Brad & Will Made a Tech Pod.

60: Forgiveness, Not Permission

Episode Summary

Big AMD energy this week. PC World's Gordon Mah Ung has gotten his hands on the new Ryzen 5000 series chips, and now he's here to tell us all about AMD's latest attempt to take the CPU crown (well, that and getting the fire department called). Then we dig into the announcement of the new RDNA 2-based Radeon 6000 cards, perhaps more affectionately known as BIG NAVI. Support the Pod! Contribute to the Tech Pod Patreon and get access to our booming Discord, your name in the credits, and other great benefits! You can support the show at: https://patreon.com/techpod

Episode Notes

Big AMD energy this week. PC World's Gordon Mah Ung has gotten his hands on the new Ryzen 5000 series chips, and now he's here to tell us all about AMD's latest attempt to take the CPU crown (well, that and getting the fire department called). Then we dig into the announcement of the new RDNA 2-based Radeon 6000 cards, perhaps more affectionately known as BIG NAVI.

Support the Pod! Contribute to the Tech Pod Patreon and get access to our booming Discord, your name in the credits, and other great benefits! You can support the show at: https://patreon.com/techpod

Episode Transcription

Brad: [00:00:00] Should we talk about giant touch screens, doing data visualizations on live cable news. 

Will: [00:00:04] Yeah, Yeah

Gordon: [00:00:05] It is better than when they started to pretend they were in the screen before. And it was just kinda like, and it would be wrong

Brad: [00:00:11] yeah, there's been weird shift away from the weatherman style of I'm on a green screen and pretending that I'm waving around inside a computer and it's more just like here's a giant television that we're shooting with a camera.

Will: [00:00:21]Well, so it's funny because. Gina. And I were talking about this last night at like three o'clock in the morning. And she said that that job, like being the screen guy is maybe her dream job. Cause it's like math. It's all of the things that she likes to do. Like weird math on the spot. You've got to do it really fast.

It's like being in another, like a high school math competition or something. And I was thinking, as she was saying this, I was like, I want to be the person that's directing these shows and that's feeding them lines. Like I want to be, I want to be Albert Brooks in broadcast news. Back in the background, dumping the lines into people's head and stuff like that.

And just riding that wave of hype all night long. 

Brad: [00:00:56] It's almost turning into like a competition at this point. Like the John King, Steve Koernke arms race, like he was like straight up posting Twitter clips of Koernke like doing crazy math in his head and stuff like that. It was just like, there, they are just in some sort of weird data visualization, pulling aggregation, arms race at this point. 

Will: [00:01:14] And it's not like it's, it's not like it's difficult math. It's only difficult math when it's 3:00 AM and you've only slept like six hours in the last 70 to 72 hours or something like that. But, um, 

Brad: [00:01:23] I’ve straight up seen people tracking the lack of clothes changing in guys, like that and like the person running the count in Philadelphia or whatever, it's like, yeah.

That guy's been wearing the same shirt and blazer for the last three days, I'm not sure if he's gone home yet. 

Gordon: [00:01:37] You know, that the other change in all the broadcast news. Have you noticed? I, I have, no, I don't think they did it this time. But do you remember when they used to have chairs on broadcast news?

And suddenly all the sets went to like, let's stand on the set. 

Will: [00:01:50] I feel like, I feel like a lot of them have. So like the, the screen guy has to be standing. It's getting, he's got to get back and forth from that big surface, over to the you'll notice Microsoft didn't sponsor this one, by the way, like last go around, they sponsored.

And I think it associated surface surfaces with like a lot of people's worst night of their lives. 

Brad: [00:02:08] You really don't want to  fall. That is the worst possible brand association that you could ever form. 

Gordon: [00:02:12] Hey, you know what I would do if I were Microsoft though, for surface, here's my suggestion. You signed bill Bellacheck as spokesperson now.

Will: Oh no, no.

Gordon: Yeah, because it's just like, you know what bill Bellacheck, even people in Boston will fire you in a season. But so you need some money now and we want to make you our, you know, I, you know, the inappropriate word these days, but you know, you just want to like, yeah, you hate it on the surface. You're now our official spokesperson for the NFL.

You see Belicheck, they're just kind of sulking. And like, I like surface. I, I, uh, I really, I use it all the time. 

Will: [00:02:48] If you're going to do that, you might as well go all the way and make them the official computer of the Houston Astros too. It's like, Hey. Look, we've moved at one level deeper. We don't need to bang trash cans to cheat effectively anymore.

We're just going to get out there and we're going to write down the secret signs on the signal, on the surface and hold them up in the outfield so that the guy behind plate can see what's going on. 

Brad: [00:03:06] I don't even want to consider how much hate mail we're going to get over this. Like I, I I'm muted. I muted the sports channel on the discord for a reason.

Gordon: [00:03:15] You got to say, it's like, Hey, we've got the TV here. Why don't we just pass that information on? Hey, that's a good idea. That's pretty impressive for some jocks to figure out and you got to say, 

Brad: [00:03:24] okay now we 

Will: [00:03:27] let's start the show. 

Brad: [00:03:28] We've got to quit while we’re behind. 

Will: [00:03:54] welcome to Brad and Will make the techpod I'm wIll, 

Brad: I'm Brad. 

Will: And today we have a very special gut guest, Gus Gus, 

Brad: [00:04:03] a gusty guest, 

Will: [00:04:04] uh, my dear friend and former colleague from the old maximum PC days, Gordon Wong now of PC world. And you may know him from the full nerd podcast. Welcome Gordon hello.

Gordon: [00:04:13] Hello. Hey, how's it going 

Will: [00:04:15] now? You know, it's uh, it's let's say we're recording on Friday morning. 

Brad: [00:04:19] It's a week. It's been a week.  

Gordon: [00:04:21] I feel like I got to have a beer, actually. I was  thinking

Brad: [00:04:24] 8:00 AM beer. 

Will: [00:04:25] Yeah. Um, but you're welcome to the show Gordon. We, we, we, uh, we brought you in this week cause like there's a bit of a bunch of stuff happening on PC hardware in the last couple of weeks with the AMD stuff specifically.

And I know. Um, some of you you can't talk about, but, uh, like we wanted to kind of check in on the state of the CPU, post the Ryzen launch a few weeks ago. Um, now that some benchmarks are starting to come out and stuff like that. And also to talk about the new GPU's that AMD announced, I think last week, but it feels like about a hundred years ago

Brad: [00:04:58] time, time might as well not exist at this point.

Gordon: [00:05:02] Yeah. No, it's it's the, the. That everybody basically, because due to, uh, our current world events, everybody's just launching parts whenever they want to. There's no more schedules or no more timing around. So it's been, it's been a crush for everybody who does it, you know, looks at hardware for a living.

But I think the, probably the best way to describe, you know, Ryzen five thousand Zen three is it's. It's hard to believe that AMD could crush Intel's parts even more because, you know, frankly, Ryzen 3000 was already very, very impressive, right? 16 core, you know, true consumer part. And now they've come back new cores.

They've said we're going to hit this performance. And for the most part, they have hit it or exceeded it as far as I've seen. And, and one thing that probably people are not seeing that or. Not actual hardware, reviewers that are not exposed to all the reviews documents is they were, they played it very, very conservatively.

Brad: [00:06:04] Really? I thought, yeah. You mean in the, in the numbers they put out? 

Gordon: [00:06:07] Yes. Yeah. So like, so like, Hey, so typically you will get vendors like here is a sanity check number. Um, uh, it basically run this test, this benchmark on your machine. If your machine is not hitting this number, call us because then you have screwed something up or something is wrong.

Right. It's I do this, you know, with Intel, I do it with the AMD and you should just do it because, so for one of these it's, Cinnabon's just like, okay, you should hit X number in, you know, cinema bench, you know, single thread. Okay. Hey, I'm I'm like, I'm there. Right? So you're saying, I think six, whatever, six 40, I'm getting six 30, nine.

They're like, Oh, you should be higher than that. We’re like what. Like, yeah. You know, we played a very conservative, you should be at least, you know, so they basically said they talked to the motherboard vendor, they got me new bios. And actually it's like, this actually brings the performance of this board in line with the other boards out there.

And all of that. I had to throw away everything I'd already done because I'd started it. Cause it was within 1% of what they had said It would be just to get one and a half percent higher. Right. So it's just because they want it to be. Higher

Will: [00:07:16] So, so this, um, I have two questions and one is kind of like a meta what it's like to test during quarantine.

Um, but, but I I'm curious, like, This is the, this is what I asked when I told you we're gonna talk. How about this today in our discord? Um, the, one of the big questions they had was about memory compatibility. They had a lot of questions about reliability with Ryzen because they know that the first gen parts had some problems.

Um, it seemed to me that most of that stuff cleared up with the second gen, especially if you were careful to stay on the Ram vendor list, but because AMD is supporting motherboards for so long, Like, can you end up in a situation where you have a motherboard that's theoretically supported, but is actually impacting your performance because it's not, you know, it's three years old and you've got you don't you're not getting the bios updates you want or something like that. That, that seems like a real kind of a real concern.

Gordon: [00:08:05] well, I think most of it is, is not really the, it's not really, you're going to see a drop off in performance. That's typically more motivated by, you know, what you went out and bought a cheap motherboard.

The VRMs, can't put out the voltage to really hit super high clocks, and that's probably, what's going to impact your performance, more of memory timings and those things. I think more of it is, you know, people like, Hey, I bought the original AM4 socket, you know, X, three 70 or whatever, and I can run the newest things.

Like, well, not necessarily. There is just some. Some boards where you just can't drop in horizon 3000, there was a big, huge kerfuffle over, you know, some old four, 400 series boards that they were going to like, you know what, we're going to leave you behind. And those people said, no, don't leave it behind.

And AMD actually to their credit came back and said, yeah. Okay, well, we'll, we're going to support it. Although it is a serious pain in the butt. To try to support older motherboards with new CPU's and is one of the reasons why Intel doesn't do it is because now you've got this, like you got consumers. We, we don't know what works with what, and then you've got people returning stuff, RMAing stuff, and then you've got to qual all this stuff.

It really gets to be a mess. You know, AMD, this was actually a weakness AMD for a long time. Cause people tout like, wow, this is awesome. That. They can do all this in AM4, you know, it really, truly is amazing to get the amount of CPU generations and performance improvements in the same socket. But the reason they do this is because in the old days, Like, you know, like, Hey AMD is like, we want to do a new socket for a new Athlon effects part or whatever it is.

Right. And we're not going to waste time for this because no, one's buying your stuff. Right. And that's, that's on, it's true. 

Gordon: [00:09:45] Like a board vendors, like we're not going to, it's not, they don't have the gravity to move things like Intel, just say, Hey, we're coming out with a new socket to match a new CPU.

Like, yeah. Hey, you're selling 80%. You know, probably at one point it was 90% of CPU sold into the entire market. Yeah, we're going to do it. We have to, right. And honestly, we like selling new motherboards, not, not having people buy, you know, try to use the old ones. So I think it's just sort of conditioned AMD like that once was really a major weakness has now turned into a major strength because, you know, for nerds like us, we love being able to just swap the CPU out.

Yeah, realistically, consumers don't ever do it. And you know, 99% of people don't do it. But for nerds, it's like so valuable. So, 

Brad: [00:10:27] sorry, I'm just gonna jump into here. This was one of my, kind of later questions, but I'm just gonna bring it up now since you're talking about it. So looking at AMD roadmap, I believe they've said that this is probably the last chip on AM4, is that your understanding?

Gordon: [00:10:39] Yeah. No. And it makes sense. Because, you know, they have said publicly, we need a major change, right? We're not just going to break it for like, they went from 400 to 500 PCIE four, you got PCI four with five seven. So with am, whatever the new socket will be. I'm thinking AM5  makes sense. You're going to have, you know, DDR four, potentially PCIE five.

So, you know, I've been saying, you know, AM5 with PCIe five and DDR five. So, you know, you have a major memory change. You it's. Okay. Most people admit it's okay. You know, this hasn't always been the case because you had people, Motherboard vendors tried to make it a DDR with DDR two motherboard. And those things werejust terrible. people don't like. 

Will: [00:11:22] Do you remember that one that had both slots you could put SDR and DDR Ram on?

Gordon: [00:11:26]Yeah yeah, yeah. Yeah. It's a mess. It's a huge, huge compatibility issues. But you know, people, memory is very, it's, it's very slow to change, so 

Will: [00:11:39] So somebody wants to buy a Ryzen machine. Buy memory off the, if you're coming into Ryzen and right now the safety, I mean, new board's coming, you're probably still safe to buy just by memory off the list and by one of the current gen boards and you're going to be okay.

Right. 

Gordon: [00:11:52] Yeah. I, you know, of course, if you want to be, you know, uh, you know, a safety Sally. Yeah. You're definitely going to go off the QVL list, qualified vendor list of memory. Those are, those are the modules that they found, but if you're building. And again, it depends on your audience. If your, some hardcore YouTube were doing extreme, overclocking with liquid nitrogen, that's a different world than I'm going to build a box off of parts on new egg or Amazon.

You go out and buy a X five 70 board, you'd go out and buy a 5,900 X and you basically buy any name, brand memory, which pretty much it's kind of work because, you know, they're the jet expects. They basically, if you're going to. If you're going to hit the jet expects, it's it's always going to work with jet expects.

It's really the going beyond the jet expects. That's the problem 

Brad: [00:12:36] is, are the XMP profiles typically above those specs? Is that actually technically, like, I've always thought of the XMP as the baseline, but is that actually out of spec for what you're getting supposedly? 

Gordon: [00:12:46] Yeah. XMP is actually beyond genec generally.

You know, some of them can be, you know, potentially within, but you know, they'll have generally like, you know, two profiles, right. And typically the higher end. Profile will be beyond the official jet expect for it. But you know, I do, I do want to say this cause people, I, I continue to hear this, especially as you get to more.

Let's face it. And this is a world AMD fans don't understand is AMD is just stunk it up prior to Ryzen. They just stunk it up for 10 years. Right. And people, when it gets to that point, like, you know what, I'm never going to buy a Fiat in my life. That Fiat was the worst car I've ever had where I was suddenly  Faits making the best damn car in the world.

It's hard for people to go like, well, I had a Fiat, well, that was like 10 years ago, man. That one thing that was a piece of junk, you know? 

Brad: [00:13:34]So this, this tracks, I mean, this is going to be a very self-serving part of this podcast, but this tracks 

Gordon: [00:13:38] AMD is not a Fiat, sorry, don't want to say that. Cause Fiat's 

Brad: [00:13:40] yeah, just attracts of one, the one with my experience over the last 20 ish years, like from, from the Durand 600 to some kind of Avalon 64, I was a diehard AMD guy that Avalon 64 was a lot like, what was it FX after that?

Gordon: [00:13:55] Yeah. I mean, it's just one thing. You don't have to remember them. You don't reserve brain space to remember the brands at this point. 

Brad: [00:14:01] Literally, when I fell off of AMD and bought a core  i7 860, and like, I've been Intel since then, and now I'm sitting here looking at these Ryzens again and like, okay. Maybe it's time to jump back over again, but here's the question I want to ask you.

Or one of many questions. Uh, I'm on a 7,700 K right now and I thought it was kind of okay. And then watchdogs Legion came on and that was like a, that was a pretty rude awakening. In terms of what my CPU is able to do at this point. Um, so I'm looking at Zen3s, thinking about getting back in, but like one of the, like you said, one of the big selling points of AMD at this point is like, I buy a motherboard right now.

I've got an upgrade path for X number of years. Like that's not going to be the case right now. Right. Because, it sounds like these boards that are out that are accommodated a Zen3 chip right now, That's it right? Like you're not going to work 

Will: [00:14:48] Oh god Gordon for the love god. Don't encourage him to wait please

Brad: [00:14:51] no, no, that's not my, but that is the case, right? 

Gordon: [00:14:53] Yeah. The case that the AM4 argument, this is gonna, I mean, 99%, this, will be the last CPU, new CPU AM4s. You never know, you know, DDR five could hit serious, you know, launching yield issues. And then maybe they say, you know what we're going to do in it.

You know, you don't know, but yeah. If you wait, it always gets better, but 

Brad: really 

Gordon: [00:15:19]you can wait. Right? You can wait, 

Brad: [00:15:21] I'm talking to, I'm not saying I'm going to wait. I just want to know what I'm getting into. Do you see any mid generation like Zen3 refresh on another socket or is this kind of one-to-one you think.

Gordon: [00:15:31] I, I think so, you know, they've already, they said, Zen 4s on the way that'll be, you know, hopefully five nanometers. The, the general roadmap is 2021 to 2022 kind of thing. 

Brad: So that’s a ways often then

Gordon: it's a ways off. And I think, you know, again, you're, you're getting stupid. Like, Oh God, I hate the like, yeah, if you always wait, it'll get better, but you never want to be caught, you know, never want to be the last person on the old part.

Right. That happened with Intel. If you had a 9900 K and you bought a 900 K, that was it. You never got, when you couldn't run the new 10, 900 K I would say though, these parts, I mean, 700 7,700 K for most games is going to be fine. And you know, generally a, a GPU upgrade is better most of the time.

But if that game is really. We're running out of gas. I'll have to fire it up. I just finally got a code for it, but 

Will: [00:16:24]it's pretty CPU Like it, especially if you turn everything up, 

Brad: [00:16:28]watchdogs  maybe watchdogs as an outlier, like, we'll see how cyberpunk does. And like what else? Assassins and, and call of duty. Like it may just be a watchdog thing, not a slower CPU thing.

I'm not sure. 

Gordon: [00:16:39] Go ahead. 

Will: [00:16:40] Oh, I was gonna say one of the big questions for me is with new consoles coming out right now that have a much larger number of cores than we're used to like this, the, the, the I'm really curious to see what that's gonna do to performance on this next wave of like AAA games. And my guess is it probably isn't gonna impact this first year crop, uh, titles as much as it does next year and the year after, especially, but.

Like with, with the way Ray tracing works and the way it changes the way the rendering pipeline works. So that you're not as you're, you're not really feel rate dependent anymore. You're entirely dependent on how much math you can do per frame. And how many light sources there are much more than like how many pixels are on the screen.

We're in a real, like benchmark is going to get real weird over the next few years I think. 

Gordon: [00:17:24] Yeah. And, uh, you know, I think I'm with you. I generally, cause you know, this has been something. This is not new, right? People say, well, you know, we've got Ryzen you’ve got eight core is costing basically, you know, nothing, you know, under $200, we got new consoles is going to change everything.

Well, I'm extremely skeptical of that because I heard the same old line with the, you know, eight core consoles. Remember you had eight core consoles. My God, this is going to change PC gaming. It didn't, those weren't exactly hot cores in those consoles. But I, I fundamentally think one of the problems. Even for triple a is, you know what?

I love making AAA games. I also like making a lot of money. I'm going to sell this game. Am I really going to leave those with a quad core CPU? Like in the  and you know, as much as the PC has moved forward a lot recently. You still got a lot, you know that, and I've been saying this for a little bit now, but you know, the, the PC strength is legacy and the PCs weakness is its legacy.

Right. Because. And the same thing with windows. They like, Hey, what, you're going to drop off this support for this thing that only five people in the world use. We're going to come, you know, we're going to March on Redmond, same thing with hardware. I don't really believe they can just simply go. Yeah, we're going to leave everybody.

Behind

Brad:  I'm surely right. 

Gordon: Turn off features. Right. They'll probably turn off features. But

Brad: [00:18:49]  if you look at the steam hardware surveys, like time after time, they bear out exactly what you're saying. So like 

Will: [00:18:53] 6,600 K is still the median and CPU on steam hardware survey last time I looked.

Brad: [00:18:57] 1060, you know, 1650, like cards like that. Dominate. The GPU lists by far by like an order of magnitude.

So yeah. You know, we're just nerds and we like to think that we have this thing under the sun. Um, so how many of the, the Ryzen 5,000 series have you gotten your hands on personally? 

Gordon: [00:19:14] So, so, uh, I originally this, another sort of inside baseball thing, they are extremely sophisticated with how they put spins on everything.

So in previous launch you might get like, Hey, we're going to send you, you know, All four, right? This is four parts here, but you basically sample the, a ones that you want to basically. So you want to shape the coverage. You ship a one part out. So this time the vast majority of people got the 12 core and 16 core Ryzen the Ryzen of nines, which are frankly, the ones I'm excited about.

Cause I love big fat, hairy hardware. Yeah. I didn't get the six core. I didn't get the eight core until yesterday. So you basically, they launched yesterday. I got them yesterday because if you're AMD you go like, you know what. We, if you want to, like, you want to just, you want to just put that knife into Intel's back, you show off 16 core, 12 core performance.

It just crushes them. Like, you know, single-threaded, multi-threaded, you know, where it's 70% to 20%, right? You look at those lower end parts where you're looking at a Ryzen 5, Ryzen 7 where it's really an eight versus eight and a six versus six. I can tell you Ryzen still. Likely I haven't tested them yet, but they're going to be figuring you still going to accept probably at least 20% faster, probably across the board.

Maybe a little bit better. That's not as impressive as, you know, 20% to 70%, 20% to say 25% or 30%. That's good, but it doesn't give you the same messaging is like, I mean, they will crush them no matter what, but, you know, but it just sort of goes, it shows and I don't know. Will, you probably haven't looked at too much hardware lately sorta like caught up in that horrible reviewers being manipulated.

Will: [00:21:02] I spent some time with like 30 eighties and I spend a lot of time with those 300 Hertz monitors that Nvidia just rolled out, which was awesome

Gordon: [00:21:07 But you'll notice the timing on when reviewers get hardware is part of the PR marketing. Right. And it's just, it's so sophisticated now because they know. If they only give these to you, you're going to look at them you know.

Will: [00:21:20]Like I suspect that they even look at like yield sometimes and think about like how, how the chips are binning and like, which, which of these things do we want to move into the market hard and which ones do we want to kind of slow play because we're going to have trouble getting them out. And it seems like.

Like one of the things that's happened since I left maximum and PC and stopped doing a lot of hardware stuff is we've gotten much better as a society at like binning a hardware and designing stuff that we can actually manufacture in mass because you don't often had pandemic aside, see like a new hardware launch rollout and all of a sudden they're selling a bazillion more of the thing than that, then they can actually manufacture.

Um, but yeah, I, I. The kind of whole, the PR cycle for this stuff is the thing that struck me going back into this. When we started this tech, the tech pod is how much that's changed and how much more savvy they are, how much more engaged with the enthusiastic they are. And, and honestly like, How much more transparent they are about like, I was shocked when we went to those Nvidia briefs for the 33, for the 3000 series, I was shocked at how transparent nVidia  was with like the things that they thought were good.

And the things that they thought were bad. Cause like the thing is that the, with all the people that are doing this now and the, the sophistication of the folks from YouTube to digital trends, to whatever, um, or sorry, a digital Foundry to whatever, like. They're there. They can't get away with being sneaky anymore.

You used to have to detectives some stuff out at maximum PC and like there's enough people working on it now that somebody will figure it out across the board. So, 

Gordon: [00:22:52] yeah. Yeah, no, it's insane. It's everything is everything from the reviewers. I feel like an old person and I am an old person versus like, you know, like, uh, uh, Steve Burke at gamers nexus.

He was like, he's doing like pressure traces where he, he got pressure sensitive. Uh, paper and he's looking at CPU and GPU mounts. And then, then, then he's looking at basically sorta like, um, then he can 3d model on that cause he's calibrated the paper. So then he can 3d model the pressure traces for how you've mounted that CPU cooler and then take it off.

Brad: [00:23:22] God gamers nexus feels like it came out of nowhere. And just like the amount of sheer ingenuity that I see going into the tests they're doing is like kind of insane from like, you know, testing, testing, water loops and looking for. Behavior that nobody else is catching and stuff like that. Like they really are hitting it.

Gordon: [00:23:37] I, yeah, no, I love Steve. He's awesome. 

Will: [00:23:40] The first one of his videos I saw, I was like, man, this guy is crazy. And then I watched the video and I was like, Oh man. Like, I wish we had him 15 years ago. It max PC, he would have been a monster, 

Gordon: [00:23:51] Yea I think you know, and you know, Steve, you know, he is a fan, you know, a lot of the younger people actually.

And I, you know, I will say this to old man Will and old man Gordon, but a lot of these people, they, you know, a lot of people, max, PC boot, they were just huge fans of it because. If you think about it, there were very few print publications that actually flew the flag for what nerds cared about. 

Will: Oh yeah.

Gordon:In PCs right. So I think in some way it's really kind of good because at least what we put on paper back then 15 years ago is sort of inspired a lot of kids. You know, now they're. Now they're doing their awesome stuff. Like, Oh, I'm an old person. 

Will: [00:24:25] Not, not to make excuses for our nonsense back then. So, so I, you know, I learned how to build a PC from Tom's hardware in that boot build a PC guidance like 1997.

That was, those were the two sources I used when I put together my K6-2 And, um, the, the, the thing about moving over and working at max PC was that it became clear that a lot of what we could do was entirely gated by like the fact that it was going to fit into like 1500 words. Right. And, and it didn't like there was stuff that like, When we did smoke testing and, you know, evacuated the building a couple of times and like tried to do latency testing in 2008 and couldn't afford a high-speed camera because it was going to be a 300 word blurb in the bottom of some article.

Right. Um, like I I'm envious of, Hey, we here's how much money we're going to make on this YouTube video. So here's how much we can spend and that stuff being relatively. Uh, kind of flat, would, it would have made things a little bit easier. Anyway, 

Gordon: [00:25:22]As well as the tools to the tools today are so sophisticated, right?

I mean like, like you said, like I, there was things like that smoke tests were Brad may, not have heard this story. We took a system and we basically, you know put plexi on the side of it. And then we took pictures as we blew a simulated smoke through the case. 

Will: [00:25:39] Gordon had his fog machine and brought it into the office

Gordon: [00:25:42] I brought the fog machine in, and then we were in, you could see, of course, you know, the, the, the smoke would just stall out near the PCI cards.

PCI, of course. And of course I th I B I remember like, Hey, uh, 

Will: [00:25:55] As you were firing up the smoke machine. I was like, Hey, there's smoke detectors in here. You'll think this stuff will trip the smoke detectors do you. 

Brad: It'll be fine. 

Will:  You were like nah It’s fine. Let's just ask forgive this not permission. And then the next thing we know, we had the whole building evacuating.

The firetrucks were on their way.

Gordon: [00:26:07] No, no. You know the dumber thing though. No. See, Brad doesn't know the actual dumb story. So the actual dumb story is. I don't know how these smoke detectors work. Do you think they're going to key off the particles and it must work off smoke, this isn’t smoke it’s just fog. Well, you know what you know, here is the dumb thing.

Now Will may not remember this. Hey, let's ask the, um, let's ask the facilities person how the smoke detectors work. She's like, well, I don't know, but don't do it then because 

Will: [00:26:35] Her spidey sense tingled hard. 

Gordon: [00:26:36] just don't do it. What are you stupid? Oh, okay. Yeah. So then there's like, Oh, let's do it anyway. And that's when we set off the smoke detectors in an entire building.

So 

Brad: [00:26:50] the whole forgiveness, not permission. Philosophy is great right up until it's the County that you're half having to ask for forgiveness. And then it might be a bit of a problem.

Gordon: [00:26:56] Yeah a fire truck did roll up for that. So 

Will: [00:26:59] well, they were used to it by then. Cause you guys had done the, uh, the fire safe test with the, with the kerosine soaked, uh, dura logs, Duraflame logs a few months before that.

So when you melted the asphalt in the parking lot or whatever. 

Brad: That's amazing. 

Gordon: [00:27:12] A gallon of gas and five. Five Duraflame logs, one small what? Two pound extinguisher will not put that out.. 

Brad: [00:27:19] Well that sounds like it sounded like a party to me. So, um, so just real quick rundown for people who don't remember all this stuff. There are four Ryzen, five thousands that have just come out.

There's six, eight, 12, and 16 cores. The prices are two 99, four 49, five 49, seven 99 correspondingly. Um, I kind of wanted to ask in your opinion for like your average use case gaming, light productivity, whatever, like, what do you think is the sweet spot there in this, in this new lineup? 

Gordon: [00:27:47] Well, it's, it's clearly going to be that two 99, the Ryzen five comes with a cooler.

It's just, you know, for most people you’re doing Photoshop, even frankly, you know, premiere, you're not doing heavy it’s like E and if you're sorta like at 75% gaming box, come on. That's, that's all you need really sort of, I think we're 

Will: [00:28:06] Is that the six core one. 

Gordon: [00:28:06] Yeah, that's six cores. It's more than enough for most people.

Brad: [00:28:10] interesting. Cause like I, and this might be fasile like, this was probably a simplified understanding of how multi-core software works, but like I'm looking at the consoles and going like, well, those have 8 cores. So I should probably get at least eight on the PC, but I guess that's not quite how it works from the sounds of it.

Gordon: [00:28:24] I w I, I will welcome the day when I am wrong and say, you don't need an 8 CPU to play because I do. I wish for that forever. 

Brad: [00:28:32] Because that'll mean there'll be doing way more interesting and like high powered stuff, right? 

Gordon: [00:28:37] Yeah. So there, there means actually game developers have decided to take full advantage of all the PC firepower we've had for a long time.

Will: Like 8 years yeah

Gordon: Like, and the reason I'm always so skeptical is Intel used to go out and literally pay developers to support their six core CPU's Gulf town Westmere. Right. They paid them right. to support it. And then they're like, look, it works. This is like 2012 or something. I like. Yeah, 10 years ago. 

Will: [00:29:03] I mean, it's really, multithreaded stuff in game.

Renders are mostly single-threaded is the thing. So like you can break out the sound, you can break out the physics, but like you run out of things to break out about around four or five cores. And then after that, who knows? Uh, I I'm curious, like, have you. As somebody who does a lot of benchmarking, have you found new workloads that are actually consumer workloads for, for like 10, 12 core CPU's aside from like the obvious the old school.

Hey man, you're you're 4k video. Like. If I'm doing 4k video edits, that's pretty much the only thing that, that spikes my, uh, 9,900 K all the cores on. And, and for, for enough time that it's worth having, you know, eight cores with 16 threads. Right. And are you seeing stuff beyond video and, and like CAD stuff and, you know, cinebench and those types of benchmarks?

Gordon: [00:29:55] Not, I mean, not really. Um, yeah. The best case you can argue. As you know, as you're saying is video editing and codes, even there are sort of tops out, you don't need a thread ripper, you know, 32 core for that, but 

Will: [00:30:08] machine learning stuff if you’re a scientist I guess right.

Gordon: [00:30:10] No, and this is the, this is the one play that Intel's betting on.

You know, AI kind of interesting, you know, they're doing all, they're interesting, you know, different, different things branded under DL boost. But if you look on the Intel on a desktop side is just plain. Done. Just done. 

Brad: [00:30:27] I apologize. I'm just going to use this podcast as an excuse to ask you every question I can think of.

Cause, cause I 

Gordon: [00:30:32] on laptops though, yeah go ahead.

Brad: [00:30:34] I was going to ask what you think about like that, that rocket Lake announcement for next spring that happened a week or two ago. So it seems pretty underwhelming. Like what's, what's your general like state of the union for Intel right now? Like where are they going?

Gordon: [00:30:45] I think Intel is, they have too many enemies to fight.

They have too many. Mouths a feed. And I think there's just trying to desktops as much as all us nerds and discord and Reddit get totally like, Oh my God. Over 16 core Ryzen, the vast majority of humans use, um, laptops. 

Will: [00:31:07]So laptops with four  cores and 16 gigs of memory. Right. 

Gordon: [00:31:09] Yeah. And Intel's Intel.

Basically. They want a different data center, which has all the money. Their problem, there is AMD or Nvidia, and then they to defend laptops, which is AMD has got stellar products on as well. But on desktop, rocket Lake will be, you know, the rumors are basically, they're going to take ice lake, you know, microarchitecture back ported on to 14 nanometer because they can't make enough 10 nanometre parts, but you take 10 nanometer, a design made for 10 nanometer and you're going to stuff it onto 14 nanometer yeah. Maybe you can get two better clocks because. Thai, uh, ice lake could not target Lake 11th gen can, but you know, it's still 14 nanometer.

You're not going to, they're not going to get the core density that AMD has. So I think probably what's going to happen is you're going to see basically the micro architecture you're using now, which is essentially Skylake plus plus plus plus plus, plus you're going to see a really decent, you know, uh, IPC increase.

So they'll get closer to what we're seeing out of Zen3. I don't know if it will be better. 

Will: [00:32:08] Sorry. IPC. 

Gordon: [00:32:09] Uh, instructions per clock. Right. So, so, you know, so I think they're going to get better on that efficiency and, you know, actually Will the whole IPC thing. Doesn't really, I really mean what it used to mean when you know, really smart people would yell at us in the nineties and say, well, that's not how you measure IPC.

Cause people now IPC AMD's definition is we take 20 applications. We lock all the CPU's at four gigahertz and we look at. Well, in the old days, they used to say, well, you gotta use, you know, this, you gotta use a spec thing and you gotta, like, it really was like high end computer scientists, kind of like kind of stuff.

But now it's really gotten very loose with IPC, but really it comes down to efficiency per clock. Right? So, so rocket lake will get them closer. May not beat them. Maybe it will be them, but that's only in lightly threaded stuff and AMDs just going to bash their heads in still with. 12 and, and 16 cores. So rocket Lake is not going to save them, but it may, it may STEM some of the bleeding.

Brad: [00:33:04] So is of again, self-serving here cause I want to know what to get, but, um, I'm doing all, you know, we're working from home right now. Like I'm doing a ton of streaming. Like I am doing a lot of playing games on this PC and like streaming them out at the same time is like the, say the 12 core overkill for my use case.

Gordon: [00:33:21] I don't think so. I think 12 is like, honestly, I think for the average, and when I say six core, I mean, that's for your average person, they're getting office, they're playing games. Like if I were to build my son a machine right now, or my daughter, I would be a six core machine. I'm not going to pay for 12 for me.

You know, I would definitely pay for the 12. I think the 12 was a stellar. It's really a great place to be 16 is awesome, but I don't do 3d modeling at home. I don't do 50 million things at once. I think 12 is really good place to be. 

Brad: [00:33:45] Yes. 16 sounds like it is overkill. Is that generally the case also kind of a related question.

I noticed that the, the base clock on the 16 is a good bit lower. And does that. Impact anything much. 

Gordon: [00:33:58] So the base clock is really sort of, you know, you're not generally going to hit it. I mean, I suppose the, you have to sort of think of, um, so the problem with these, the 16 core AM4 part is if you look at some of the performance tests we did you see when you put it against the older 16 core part, it's.

Huge difference in that sort of left that lean a lightly threaded single core stuff. As you get to the middle, it's still better, but you get to the right side is sort of, it's still faster, but it's not that huge margin. You're seeing 16 core versus 16 core. And the main reason even according to AMD is we're just, we're just power limited.

You've got 16 cores. If you're really using all 16 cores at a hundred percent for 25 an hour. We, you know, it's a hundred, five watt part. We're not this isn't a thread ripper, they're basically answers like if you really need 16 cores for two hours or four hours at a time, go out by a threed ripper we've got 32 core parts.

We've got 64 core parts. That's really, and they just have a lot, it's a bigger socket, you know, there's, there's more memory channels. That's really the tool to use. Um, so I, I, that's what I would argue 

Brad: [00:35:04] and there, are no zen 3 base thread rippers announced yet. Nah I don’t need a thread ripper.

I'm just curious. 

Gordon: [00:35:10] Not yet but I mean think about that though, if these cores are essentially power constraint in a 16 core version, you take these suckers and you put them into a next generation thread ripper is going to light things on fire, right? I mean, like the benchmarks. It's just going to be incredible. Right? So you just sort of take the shackles off of the power and thermal set, you get out of threed ripper platform and it's going to be, 

Brad: [00:35:34] Ooh, that's exciting.

Will:I like the sound of that. Yeah. Um, let's can we switch over to GPU? I think we've gotten, I think we've hit everything important on the. 

Brad: [00:35:43] I can keep asking questions for a while. 

Brad: [00:35:46] I haven't been, this is just in general. Like I, this is the most exciting the PC hardware has been to me. inI don't know how long, like 15 years, probably just 

Brad: [00:35:56] the CPU's GPUs is it's just crazy 

Um, what else was I going to ask? Do you have any experience with, uh, with graphite thermal pads? I haven't actually talked to anybody who has used them firsthand. And like, I, I imagine with the amount of CPU swapping back and forth that you have to do, I was wondering if. Those have come in handy for you.

Gordon: [00:36:12] I, I, I do not use them. I mean, I do know it's of course the people are people, especially in these, enthusiasts are always looking for the next cool exotic thing. You know, liquid metal I think is probably what a lot of people would use, but those graphite pads look. Sexy. So 

Will: [00:36:29] I look at my thermal pad a lot, Gordon.

I want it to look really nice. I find it to be it's important to have an attractive thermal bed 

Brad: [00:36:35] I want a nice sheen on it. 

Gordon: [00:36:36] That is what's so cool about it. I don't, I don't it's like, it really felt like, I mean, that would be like exotic. This is stuff that NASA had in the two thousands. Like, Hey, I just ordered this off of eBay.

Yeah, 

Will: [00:36:46] look, you can get all sorts of crazy stuff from alley express. I got replacement mouse switches the other day that are like high performance mouse switches. And all I have to do is wait a month and they show up and yeah

Gordon: [00:36:55] No, it really is a different world. Where does some of the less obvious workloads that would show off having a ton of cores?

Hmm, no, I'm just looking at some of the questions, but that's cool. 

Will: [00:37:09] Yeah. I mean, I think that was the, that was the big, that was the big one. I had a question like 

Gordon: [00:37:12] I wish, right. I Wish

Will: [00:37:13] I mean, I think streaming is a good example, actually, because I know like all of my streamer friends have switched from Intel machines with a second.

A lot of them switch from Intel machines with a second machine to do the encode and run OBS that they pipe pipe out, across via HDMI. And now a lot of them are just using a like 16, 12, 16 core Ryzen. Single single stream PC, because it's way simpler to do everything on the same PC and let OBS just grab it from the frame buffer.

Gordon: [00:37:41] Yeah. I mean, just think of what it's just amazing because you can just the core count that you can get. But I mean, even that's, I think the one thing that I really like about Zen three is it felt like. I mean to the external world that felt like we were just making excuses because clearly when Ryzenfirst came out, Intel was kicking AMD's.butt in gaming still,

Will: Oh yeah

Gordon:  20% is nothing to like that they were still better in gaming, but then you sorta like, well, AMD's argument was like, well, you're. You're playing these games, but you're also streaming. So you need more court density 

Will: yeah. 

Gordon: It almost felt sort of like you were making an excuse for AMD, like sorta like, and the thing is people really do play in stream.

That's like a new form of gaming that, you know, I'm, I don't do I'm I'm an old person, although you are young hipster Will, so you did it, but 

Will: [00:38:32] I got a beard for streaming is the thing I had to, I grew that thing out and I had to start doing something 

Brad: [00:38:36] Uh. Did I miss fun material while I was in the bathroom? 

Gordon: [00:38:38] No, we're just talking about, uh, the core density and how it used to be with when Ryzen first came out, that you had to sort of, it felt like you were making excuses cause you know, core, still kick Ryzens butt in gaming.

But then you're like, well, but if you're streaming like, Whoa, I know I can get eight cores and Intel I'd have to buy a thousand dollars CPU for that. It's just like, that was like a game changer, but it still always felt like you were making an excuse for AMD you know. 

Will: [00:39:03] And well, and if you're, if you're a professional streamer and you want your thing, you want your stream to look real good, the best way to do that until until the NVANC stuff came out with the 3000 or 2000 series cards was to run the X 64 X.264 software encoder, which will basically eat as much CPU as you throw at it at the higher quality settings.

And yeah. 

Brad: [00:39:27] Oh, sorry. I mean, I'm at the point and you know, this is this work from home for stuff. This may not last forever, but like I'm at the point where I'm taking like videos, we need to post to the site and like running them through FFMPEG to encode them to something else. And. You know, like I'm, I'm dealing with ProRes files and like production quality stuff fairly

Gordon: [00:39:42]All on the 7700 K.

Brad: [00:39:44] Yeah. It's kind of a bummer, right. That's why I'm really feeling the itch at this point is like, 

Gordon: [00:39:49] you should not wait. 

Brad: [00:39:51] I'm not going to, but it's like, yeah, this is going to take like an hour and a half to and code. And so this insanely high bit rate because yeah, 

Gordon: [00:39:57]Think of it, I mean the whole, like if you waited for Zen 4 whatever, you're looking at really a new platform, New motherboard, you are going to have to do a new motherboard.

You're going to have to buy DDR five, which I can tell you. Will not be cheaper than DDR four

Will: [00:40:12]Also at launch it will really slow too.

Brad: [00:40:12] That's what, that's what I was gonna say. And not only that, but like maybe there might be stability concerns. There might be kinks to iron out, like getting in on the ground floor of a brand new platform and generational memory and all that stuff.

Maybe sounds a little dicey, maybe like I'm so used to having to buy a new board and memory with every CPU anyway, that like it's. From Intel, but it's, it's kind of, it's just the status quo for me. So I should really get one of these probably soon. 

Will: [00:40:35] And Brad, the other thing to think about is when that new stuff lands and you decide you need that upgrade and you can use that the old one can roll over into the home server as you got a massive number of core.

Brad:Oh my God. I can have a 12 core NAS

Will: [00:40:47] Yeah. 

Brad:Holy crap. Okay. Sold.

Will: [00:40:53] Hey Gordon I have one last CPU question for you. Um, what's like when this situation happened with AMD last time, Intel, a few years later, you know, the Athlon was rising and, uh, Intel came back with a massive performance boost with core, by both fixing their janky Pentium 4 business, and like starting on some multi thought of this multithreaded path that we're, that we're riding on now.

W what, you know, but a lot of that was driven by process improvements and there's not a lot of process improvements left to come. So like what's, what's, Intel's, what's Intel's response to getting stomped. Like, do they care about this market still you think, or is this, is this a place where they're just going to see this to the enthusiast PC market, to AMD and lean into server metal and, and.

You know, with Apple switching to their own arm chips on mobile, which we still don't know how that's going to go, but it looks like it's going to be interesting, at least like and Nvidia owning the high-end data center compute stuff. Now it seems like, like I don't, I don't Intel has to be worried about some of these markets.

I don't know if they're going to be worried about all of them, but I'm curious what you think their response is going to be. 

Gordon: [00:42:03] I, you know, I, in this is. It's true. Cause you know, you know, AMD has in their presentation, what three years ago, four years. So they were basically saying in the past, Intel would run away with process and smoke us and they just see that slowing of Moore's law, not letting Intel do that anymore.

And we're seeing that every day manifest itself, Intel basically. There's heavy, heavy talks of them hiring out TSMC as a fab, right. Which is insane

Will: [00:42:33] That's bananas given the money they spend on fab. 

Brad: [00:42:35] I was gonna say, they must have invested like untold billions in their own fabrication, right? 

Gordon: [00:42:38] Yeah. They were talking about billions and billions of dollars in these fabs.

You got to pay for an, and also. You know, Intel, you know, they are, you know, a lot of people call them arrogant, but it's because they've been so good for so long. Right. How about somebody? Who's basically been the number one for 40 years of computing. Yeah, you're gonna there. I, I have to believe that they will, at some point pull it out, but sometimes unlike before with, you know, with Athlon 64 where AMD cleaned their clock for about four years and then basically Intel put them in a box and we didn't see them for 10 years.

I don't know sometimes I'm, I'm really kind of worried. I've been saying this for a while, because I don't know, because I, I don't know if Intel can pull it out this time. 

Brad: [00:43:21]It has  to be like turning an aircraft carrier. Right? Like it takes, it takes years for what they're doing now to actually manifest in the market.

Right. Like there must be huge lead times on this stuff. 

Gordon: [00:43:30] But they are smart. I mean, the one thing it's like people really just people like, they just want to hate on, the, the way the world of, of hardware is now. They just want to hate on, you know, and I will say this I'm a warriors fan. I’m not a Patriots fan, but I know how Patriots fans are.

People will hate you no matter what if they want to hate on you. 

Brad: [00:43:48] I feel like that's just any, any enthusiast community at this point, right. 

Gordon: [00:43:51] any enthusiast  community. But the reason why. Intel. They are smart. They have generally the smartest people in the world. So I really think I have to believe they will. At some point, pull it out. But like, it's certainly not like it used to be.

I don’t know what they're going to do. 

Will: [00:44:06] Yeah. I was going to say with the core stuff, you can kind of see the writing on the wall and see that they were going to A. lean on process and B just go massively. Multithreaded as fast as they could. But with this, I don't see. I mean, it could be that they go more SOC and start adding dedicated compute units and stuff like that to accelerate the specific workloads.

There's a bazillion different things. They could do that. I'm without the crutch of process to lean on, I'm really interested to see what happens. 

Gordon: [00:44:32] Well, you know, they've talked about this for a few years, but they haven't successfully pull it off in any higher end part, but you know, they've got their Emib right.

So they are basically going to be stacking parts. And the whole idea is like, they've realized that monolithic. The dyes are not going to work for them anymore. So, you know, let's just stack them.

Will:Can you just explain Emib stuff a little bit.

Gordon:  So, um, I don't, I don't remember the exact name of the acronym, but basically it lets them say, use 10 nanometers cores would say the power circuits made out of a 14 nanometer part, and then you basically, they just take the parts that they want and then they stack them.

Will: [00:45:08] Like Lego almost 

Brad: [00:45:10] sounds not entirely dissimilar to the chiplet stuff AMD is doing like just modularize the design of the, of the dye. 

Gordon: [00:45:17] Yeah. So it'll be modularized and, you know, emib is, you know, a high-speed, you know, mesh fabric type, whatever. Uh, and they've actually done this with the original KB, lake G, that was one of the test platforms where they took basically, they took a Kaby Lake quad core and they connected it with a rainy E core and they, they basically use their Emib to connect the two.

Right. So they could prove that we could do with the high-speed interface all on the same socket or the same chip package. And then of course the. Uh, the Radeon had the, it had HBM memory. So it was interesting Polarez Vega construction, but they do think, they think that'll help them beyond rocket Lake there's.

Um, you know, rumors of Alder Lake part that I think is going to be like a big little construction. So I, I'm not sure. Where that plays out, but you know, I, people, Intel is smart Intel is a really, really smart company. You don't get the 40 years of dominance because you just don't know, it's not out of luck 

Brad: [00:46:15] Yeah as much as they seem like they were maybe lost in the woods a little bit. Right now it seems like it would be foolish to write them off this soon. 

Gordon: [00:46:20] Yeah. 

Will: [00:46:21] So, um, we don't have a whole lot of time left, but I wanna make sure we talk about new GPU is a lot, a little bit because there's interesting stuff happening on the Radeon side from AMD as well. And like I said, they announced this stuff last week.

We don't have. Benchmarks or anything yet? Uh, the NDAs for that haven't expired yet. Uh, and I think you said you were firewalled from PC worlds, uh, GPU coverage, so we can talk freely. Um, but, but this, this is RDNA two, uh, the, the new Radian 6,000 series. And I'm like, it seems like the big change is that they've added Ray tracing.

Is that, is that, 

Brad: [00:46:56] well, I think the general, the bigger picture is the big changes that they are competitive in GPU again, right? Like much, like much like on the CPU side. It's like AMD actually matters in graphics cards for the first time in like how long? 

Gordon: [00:47:08] Yeah. I mean, they are claiming to compete with 3080, and so people think 3090, so.

Damn. Yeah. 

Brad: [00:47:16] Yeah ray tracing is obviously the big new feature for them here. 

Will: [00:47:19] And well these are the GPU. These are the same. This is the same, uh, graphics architecture. That's driving both or at least some subset of both of the new PCs consoles. 

Gordon: [00:47:28] Yes. Yep. The new PCs. 

Will: [00:47:30] Yes.the new living room PCs, 

Brad: [00:47:33] more PC, especially, especially the XBox sure looks kind of like a PC case.

Will: [00:47:37] I wish I PC case was, was square and Towery. That sounds like it would be good. Um, Th th there, uh, let's see, there, there, unlike Nvidia who rolled their own API before direct X and the Vulcan API weren't quite ready for RTX stuff last year or year before. Um, it seems like in the AMD parts are both native for the Microsoft direct X Ray tracing our API 

Brad: and DXR.

Will:The Vulcan API. Yeah. 

Brad: [00:48:04] So this was kind of a big revelation to me. Cause like the headline coming out of this was like, Oh, some of the current ray tracing, enabled games on the market are not going to do ray tracing on these AMD cards right out of the gate. And like the, when you walk that back a little bit to understand why it sounds like it's because they're doing Nvidia proprietor, stuff in software, which I

Will: [00:48:21], That’s my understanding yeah.

Brad: [00:48:22] I just assumed, I  assumed they were to spec in Vulcan, if not DXR, but I guess Nvidia was doing their own.

Customs voodoo in there. 

Will: [00:48:29] So yeah, like people haven't really, I went back and looked when you're talking about us before the start of the show, like remedy, doesn't say what API they're using for Ray tracing control because there wasn't there wasn't, there was only one card when they released that game, they could do Ray tracing.

So it did kind of didn't matter. It's like if you wanted to play GL quake, when the 3d FX voodoo was the only card out, it didn't matter what unless you had an SGI workstation.. 

Brad: [00:48:52]Right Well, I was totally going to say this is so sort of analogous to the 3d effects and glides situation. Right. Of, of, of games, games being kind of designed toward one specific piece of hardware.

Not, not at, not one-to-one, but it's kind of reminiscent of that. 

Will: [00:49:04] Well, so the difference now though, is that often the direct to API follows, whoever rolls the hardware out first. 

Brad:Okay. 

Will: Um, I mean, there'll be, there'll be conversations with all involved parties. So AMD, Nvidia and Microsoft, and everybody who, who works on in those worlds contributed.

But, but my guess is at the, at these high level, these, these high level APIs, generally the. The situation is that the you're just doing math. So like it's a situation where you can either rewrite the rewrite, the, you know, reformat the code for the, for the ray tracing part of the program, into the new API, or dump it into a wrapper or something like that.

And usually it's, it's pretty performant. Um, 

Brad: [00:49:46] yeah, I mean, um, yeah, I 

Will: [00:49:47] I don't I know if that applies here just for the record, but that's how it's worked in the past. 

Brad: [00:49:51] I'm not implying it's a problem in any way. I'm sure that those games will get back ported to the new hardware, I  just like knowing how things work. Yeah. 

Gordon: [00:49:57] I think your concerns are warranted because I, you know, definitely it, it has been floated that a lot of, you know, again, these were all DXR games and there was nothing proprietary.

Obviously the DLSS part was going to be proprietary. And everybody had assumed that when you got a, uh, say an Intel or an AMD card, that's supporting DXR, you would just flip on the switch and it would work too. But if these games are not going to work out of the box or easily with other direct X Ray tracing cards, it makes you wonder what exactly happened.

So it sh people should be concerned about that because, you know, I would hope, you know, people were more transparent about it. I mean, but it could be. This, this could be NVIDIA's GameWorks, right? This could be simply, Hey, you know what? We know you don't care about this, but we will do the work for you.

It's like, Oh really? You will do the work for us. They hand them the library, they use the library and use it. Right. 

Will: [00:50:58] Let me tell you if Nvidia called me and was like, Hey, we'll add Ray tracing to your game. I am literally a D I'm a small developer. I'm not making control or anything like that. But if somebody calls me and is like, Hey, we'll give you an engineer for a few months to implement this feature.

Fucking my hand is raised. 

Brad: You’re not going to say no

Will:  No

Brad: [00:51:13] Absolutely. Um, I've got, I've got the little quote snippets in front of me from AMD here on this issue, which is that. They say that the Radeon 6000 series will support quote, all ray tracing titles using industry-based standards. That's obviously DXR and Vulcan

Will: [00:51:28] well that's, I mean, that's always been their position.

They're like, they always go with the industry standard at AMD which is a good thing

Brad: [00:51:33] but, but also they had added the, that they won't support quote, proprietary, ray tracing, APIs, and extensions. So there's obviously something Nvidia specific there that they're referring to. 

Will: [00:51:42] My, my guess, is that going forward? This is much less of an issue than it is for back for older games.

Brad: [00:51:47] I mean, realistically, we're talking about like practically single digit number 

Will: [00:51:50] literally two games, cause like the call of duty, modern warfare Ray tracing, extensions, didn't matter. Um, we're talking about control and maybe Metro Exodus, right. 

Brad: And, and maybe shadow of tomb raider 

Will: Quake two RTX and the Minecraft RTX thing that Nvidia paid for.

Brad: [00:52:08] Oh, I I've got to get Minecraft RTX. I had Quake two RTX was the first thing I played on my 3080. And like, I was just drooling all over myself. Like I could not believe what I was seeing, but 

Gordon: [00:52:18] here's my question to both of you. So, cause what's going to happen is you're going to see a lot of yelling. You're going to see a lot of gnashing of teeth over.

Oh my God. You know, Nvidia is doing its proprietary thing again. This is not fair, blah, yada, that'd be the normal. You know, you, you complaint, you will see out there. Do you think that is all that. Do you think that's a little unfair because if Nvidia like Will said, if Nvidia says, Hey, we're, we're going to, we're going to do it for you.

We're going to do the work for you that you would otherwise not do shouldn't they benefit from that 

Brad: [00:52:48] Totally, they were first to market with Harbor Ray Tracing capability. Like they absolutely should exploit that. I'm not like criticizing in any way. It's just, it's more just a assessment of like, of what games are going to work on this new card and which ones aren't.

And what's it going to be like going forward? And I assume going forward a hundred percent of games coming out will work on both. 

Will: [00:53:02] My assumption is going forward, almost everything except for XBox exclusives are going to be written in Vulcan, which will be cross-platform. Cause I mean that if you, if you're building stuff for PS4, if you're building stuff for, for XBox, if you're building stuff for PC Vulcan means you write that code and then you don't worry about it is my understanding. There are graphics programs in the audience that, that I know Sony has their own thing that is extention, 

Brad: [00:53:25]They roll their own  graphics library as far as I know.

Gordon: [00:53:26] Does that bit. Is that realistic though? Because I mean, I've always thought Vulcan was, you know, One out of 50 versus, you know, 99, 

Will: [00:53:34] I keep expecting Vulcan to take over because of the cross platform nature of it. And I, it has it kind of like doom uses Vulcan to great effect. Right. But doom is the one one Vulcan game that I have played on the rag over the last five years.

So maybe I'm wrong. I don't, I don't know. I mean, DXR is the, is the no-brainer for PC and Xbox, right? 

Gordon: [00:53:53] Yeah. That's but I just feel like it's, it's just so much, there's almost so much gravity with. Direct X and the tools and Microsoft and PCs. And, you know, XBox that I just can't imagine who would do Vulcan except for, you know, people who want to support those, you know, you know, third-party or I know what's the.

Politically nice way to say Linux and BSD people these days, is there a correct 

Will: cranks is what we usually 

Gordon: [00:54:17] no 

Will: We like Linux and BSD people

Gordon: [00:54:18] Belt and Suspenders

Will: [00:54:21]Brad and I both have BSD machines running in our two. I have multiple BSD machines running in the same house. 

Gordon: [00:54:23] You have a beard? 

Will: [00:54:25]I have a sick,  a gray beard yeah. 

Brad: [00:54:26] Yeah. I am constantly, constantly Google searching for answers to the very, very minute.

Differences between BSD and Linux versions of everything. And it's not, let's just not get into it

Gordon: [00:54:36] I do like how BSD and Linux, people like yell at each other though. 

Will: [00:54:39]Look I’m moving my BSD Machine to ProxMox So I can just put a bunch of Linux VMs on it and call it a day. 

Gordon: [00:54:44] Well, the funny thing is to us, like dirty windows, peasants, it's like, you're both the same.

We don't, we don't understand

Brad: [00:54:50 You're all lunatics 

Gordon: It’s like the episode of Star Trek where one had the face right. The white face on the right and the left one was reversed. 

Brad: [00:54:57] It doesn't matter. It doesn't matter. What heretical denomination do you ascribe to? You're all 

Gordon: [00:55:02] sensitive or you don't understand? 

Will: [00:55:05]Look, I can  tab complete. It doesn't matter.

Uh, Gordon, 

Brad: [00:55:08] sorry. 

Will: [00:55:09] I was just going to say the thing to remember about Linux and BSD is that you spend like four years. Kind of going cross-eyed every time you have to fix something and then eventually it just clicks and you understand it's like another language 

Gordon: [00:55:22] really

Brad: [00:55:23]it's like,  honestly, that moment, that moment, every time it happens is worth all of the toil and the heartbreak, because like, it's, it feels like magic when it's like, Oh my God, I feel like it's like, it's like in D and D how you have to like, learn your spells, you know?

And they like. Exist in your brain. And then you cast the magic, yeah, it's just like, wow. I learned a magic spell at some point without realizing it. Now I can just do this incantation, but 

Gordon: [00:55:43] I have to admit, I haven't tried it since the 1990s. 

Brad: [00:55:46] It's way better than installing Slack. Where from, from floppy disks, it's come a long way for sure.

But, uh, last thing I was gonna say, I mean, we're getting off topic a little bit here, but the last thing about graphics APIs, I was going to say is. It seems like the X factor there from a little bit. I know about it is that the new unreal engine stuff that Epic is kind of moving away from hardware, graphics, APIs entirely and starting to try to do more stuff in back in software.

Again, just some of the stuff 

Will: [00:56:10] with their massive poly counts and all that stuff

Brad: [00:56:13] yeah all their virtualized, geometry and stuff like that. It sounds like that. And like, I, I, there's a Tim Sweeney talk from 10 years ago where he's talking about getting away from the, the hardware, the GPU graphics APIs, and moving back to software.

And it seems like that's actually the direction they're going now. So. Um, I'm getting in over my head here. I'm just getting a bunch of blank stares here. So we'll revisit that topic later. 

Will: [00:56:34] We'll just chop this bed out. It's fine. Um, 

Brad: [00:56:37] Oh, if you take it out, then I can't look prescient on when you, everyone is saying this in a year or two

Will: [00:56:43] Uh, let's. Let's talk about, okay. So there's a couple of things that these cards do that, that everybody else isn't really doing. I mean, first off AMD is the only one with a PCIE 4 implementation on the motherboard side. Right? Cause there's no PCI four on Intel. Is that correct? 

Brad: [00:56:57] I think Rocket Lake has it , is that right.

Gordon: [00:56:59] Yeah, Rocket lake is rumored to have it. 

Brad: [00:57:00]It's supposed to come next spring, but 

Gordon: [00:57:02] it will happen. 

Will: [00:57:03] I'm like, does that matter? It doesn't seem like that matters right now. Is that fair? 

Gordon: [00:57:08] I don't, I don't think it has mattered. It's just like the old, like every time. Oh, I need x 16 PCIE 3.  Cause it's so much better than by 16 PCIE two. And then I don't think anything is really soaking up all of that.

Bandwidth that you're getting at a PCIE for graphics

Brad: [00:57:26] The ampere cards, support PCIE four, right. 

Gordon: [00:57:30] Yes those are PCIE Four. And, uh, I will say one of the things, wrap it a little bit back to Zen three, as I said, like, look, if you go from your, your ridiculous 2080TI card to a 3080, you will actually see an increase in performance on Zen.

They didn't say anything that was necessarily. PCIE 4 at all, but I, my guess is for most consumer applications, it's not making a difference. It hasn't, it just hasn't. I wish it would, but you know, again, are you going to be a developer, does going to make a game? You can sell to five people, right?

Brad: [00:58:03] Yeah. I'm sitting here looking at 3080 benchmarks between PCIE three and four. And it's not a big difference. 

Will: [00:58:08] It seems like the PCIe stuff is maybe going to make more difference when the direct storage APIs start coming online next year, the year after

Brad: for sure. 

Will: And there's actually a use case for all. Like, it wouldn't surprise me if we see video cards in the next few years that have slots for, for SSDs.

Gordon: [00:58:23] Yeah Wouldn’t that be weird? I remember AMD actually showed that off four or five years ago. 

Will: [00:58:27] Oh really? I didn't know that. 

Gordon: [00:58:29] Yeah, they had a, it was a high-end compute card that had  M.2 it had loads of storage. 

Will: Oh that's right

Brad: [00:58:35] Oh, that would be cool as hell. Um, this has got, I do have another question kind of in this vein about, uh, the new Radeons, um, Any thought on this smart access memory, which is what they're calling this like kind of little bandwidth boost.

If only if you have a Zen 3 and a Radeon 6000, it's almost like I see people. For intuitives like an RPG set bonus. You know, if you have, if you have both AMD parts, you get this little boost, 

Will: [00:58:58] the McDLT. 

Gordon: [00:59:00] Yeah. I don't, I don't know any details of it. I was trying to figure out like, well, what is this?

Because this is, you know, it's basically the way it works is your CPU has to communicate with the GPU. Do you were a very small window? They say typically it's 256 megs. Um, this would allow the CPU to access basically all of the memory on the graphics card only with Sam. Right? So, um, yeah, it sounds like a lot of people sort of thinking that it's, you know, bar or at base address register some part of PCIE, but it's not really clear to me what it is.

They also did say that, um, it's something they hadn't told any developers about. So I think it could. Be pretty cool, because if you think about it, you, this could be performance boost for them. It also could be something that is already well being used in Xbox PlayStation. And I also think that what's interesting to me is where this could go, which is so if your CPU is accessing, whatever is in the graphics cards, memory.

Is there a certain point where, you know what, why don't we just let the graphics card memory controller run most of the call, most of the plays, right? And then maybe the CPU just does whatever housekeeping with the system DRAM. You know, but that, but then you start to think like, well, huh, you look at Xbox, isn't the Xbox and PlayStation architected that way. There's a single member controller for the x86 cores, as well as our graphics, cores. Why in the world do you need that, that, uh, Memory control on CPU, right? Why don't we set the CPU, do that?. 

Will: [01:00:33] Interesting Um, can we talk about DLSS a little bit, I feel like this is the other big differentiator between the AMD cards and the Nvidia cards.

And like when DLSS rolled out a couple of years ago with the 2000 series, it kind of sounded like snake oil. Um, and, and the,TLDR, if you don't know what DLSS is, basically they do a ton of frame analysis and machine learning stuff to build a machine learning model to super sample. Which can then be used either to upscale low resolution stuff, to play like games at now, eight K apparently, or, um, to increase image quality as if you were doing super sampling, uh, AA anti-aliasing without the four X performance hit that you get by doing actual, super sampling.

Um, and, and the upshot is. At 1080 P 1440 P 4k and 8K at 1080 P you can get increased visual quality at 4k and 1440 P you can run at like 10 80 P frame rates with much higher resolutions. And like looking at the image quality comparisons and actually using it in person, it doesn't feel like, like snake oil anymore.

Brad:  No

Will: Um, 

Brad: [01:01:41] I, I I'm with you. I wrote this feature off until I tried it myself and like, Holy crap. 

Will: [01:01:46] Well, well, and, and, and it feels like the thing that, the reason it doesn't feel like snake oil is you might have a minor, you might have an increase in like weird glitches occasionally where the machine learning kind of fails in interesting ways.

But more than that, it smooths out the frame rate in such a way that like, if you're sensitive to micro stutter and to frame rate hitches, you it's magic. Like you're trading a little bit of image fidelity for a lot of frame rate stability. And, and it seems like an increasingly big deal, especially if you're playing it 4k.

Um, like realistically nobody's playing at eight K for the next several years on these cards. Yeah. But at, at, at 4k, which I think is something that we should address and is, is like a place that people are gonna be playing games over the next two, three years. It really makes a difference. 

Brad: [01:02:34] Even at 1440 P like it just in a couple of weeks I had, since I got my 3080, I've been messing with it, like in control I did some side by side on, and I was hard pressed to tell the difference between the DLSS. 

Will: [01:02:45] Well, you have to pause the game and get into photo mode and stuff like that, and do zoom ins and things, things like that in my experiance to see the difference.

Brad: [01:02:51] Yeah to see the difference between 1440p and whatever it was rendering out with DLSS on, which was like 1700 by something.

So like dramatically lower than actual native resolution. I basically couldn't tell the difference. Except that the frame rate was smoother.

Will: [01:03:02] except for the frame rate is, is like butter. The so you can tell in geforce experience if you have an Nvidia card, you can flip on the performance meter now, and they show you the 99% frame time, which is like, it shows you what your average frame rate is with the fastest 99% of frames. 

So you, so you, um, you see what your, where your sags are better. Um, and it, and it shows up a lot there. I AMD doesn't. Like this, this hardware has the compute cores to do a DLSS equivalent, but they don't have a solution ready. And I like there's there's it seems like to implement this. There needs to be both backend.

Uh, machine learning stuff set up on the AMD side, where they can run these games on these massive compute clusters and generate really high frames and basically, you know, run through every part of the game over and over and over again, until they get enough data to, to let the machine learning algorithms fill in the gaps based on the context of what's around those pixels.

Cause cause basically what they're doing is taking one pixel and turning it into four right. Um, and uh, you know, I think, I think. I would not have said that this was a weakness before, and I feel like it maybe is after spending some time with the 3080 cards and the DLSS two stuff. 

Brad: [01:04:13] Uh, it's probably worth noting.

Um, I don't know if you saw that blog post Microsoft put up about the rDNA two features in the series X. 

Will: [01:04:20] This is the one where they said, basically, this is this in response to the, to the rDNA, two being the direct X compatible that xbox compatible 

Brad: [01:04:30 This was their kind of chest beating. Like we're the only console in the market with full rDNA to spec implemented, which kind of meaningless, because obviously Sony is just doing their own thing to, to achieve equivalents.

But the reason I bring it up is there is a. Phrase in here where they refer to improve visual quality via techniques, such as machine learning, powered, super resolution, which has to be the same thing. Right. That's basically, but, but that's lower case. So that's not like their fancy name for a feature that they were coming up with to compete with DLSS but it sounds like they were working on it

Gordon: [01:04:58] And I think AMD is actually talked about a super resolution feature. Right. So that's leveraging that. Yeah. 

Will: [01:05:05] They said it's coming, it'll run on this hardware. It's just not, it's just not there yet. 

Brad: [01:05:08] Just not at launch. 

Will: [01:05:09] Okay. I mean, I assume it requires partnership because, you know, they're going to have to build hooks into the game to be able to render on the, on the compute clusters at those high resolutions and to play the game in an automated way.

Like there has to be some backend work done. And this is an instance where NVIDIA's spending a ton of money on dev REL makes a. Big difference for the number of games that support this. But I mean, 

Gordon: [01:05:29] you almost, 

Will: [01:05:30] even with DLSS one and two, there's still only a handful of like, it's like watchdogs and control and death stranding, and maybe Metro are the ones that kind of mattered in the beginning.

And I think Valhalla maybe supports it as well, but it's not like every game comes out and supports DLSS 

Gordon: [01:05:47] I, I wonder if though how this plays out, because I I've heard this from other people too in my own brand is like, yeah, DLSS 2.0 is no joke. It's awesome. Right? Yeah. And it is a serious, um, peg plus for buying an Nvidia card over an AMD card.

I do wonder how it plays out in the long-term though, because it's very dependent on game developers. So yeah, if you have X-Box and it's like, if you're doing super resolution for Xbox. Um, you might as well just throw it in the PC version too. Doesn't this? It seems to me that it will play out for AMD in the end. 

Brad: Oh, sure.Yeah, 

Gordon: it'll be very similar to,free sync free sync and G sync battles. Right. Gsync was clearly superior, but free sync was cheaper and everybody did it. So. 

Brad: And open 

Gordon: well sorta open, but like in the end, you know, free sync sorta won, but until Nvidia flipped them on their back and called it, Oh, it's G sync compatible.

So like, uh, no, no. It's free sync compatible, Okay I want free sync compatible I mean I want G sync. 

Brad: [01:06:50] It's like a marketing master stroke that even free sync monitors are just branded as G sync Compatible now. 

Gordon: [01:06:56]Just imagine like waking up like, Oh, let me read the paper. Like, Hey, where's all our free sync monitors. They're all marked as gsync now. 

Brad: [01:07:02] These Asus monitors.I just got are only free sync, but they just had a giant fat G Sync sticker on them. That's like a coup that they pull off. So

Gordon: [01:07:11] you have to respect that. 

Will: [01:07:13] Well The other thing I wonder is for stuff like live games, like Fortnite supports DLSS now, and it means that you can turn on the new Ray tracing features in Fortnite and get a playable frame rate on a 3080, but.

Like fortnight's a live game. Epic is constantly changing assets in that game. And I wonder what overhead it adds for the developer in terms of like how early do they have to have that stuff ready before they can push it out? And I know being like a massive multi-platform game, like Fortnite is like, they have to worry about everything from the switches and the PS4 and Xbox is qual process, but, but still like you're, you're in a situation where.

Like they can't just dump a patch out the changes, a bunch of assets and change everything over to Halloween skins because that's going to mess up the DLSS presumably, and they're gonna have to retrain all those algorithms. So like, I I'm. It's going to, like I said this at the beginning of the show, it's a real weird few years for benchmarking, because on one hand you have like some magical mystical.

We're going to machine learning this shit up, but it actually works. And then on the other hand, who knows what Ray tracing does to benchmarks and like at all, all that's there kind of off so

Gordon: [01:08:16] well, yeah. I mean, it's the nice thing about the CPU testing is it's pretty clear. I spend less time doing this encode , or I do spend less time doing a render.

Graphics people have it really hard, because I do feel like visual quality is going to play into this again. Yeah. Because yeah, you've got Ray tracing. You've got VRS as well. So like what part is what vendor, you know, sort of giving you worst graphics on, you know, so I. 

Brad: [01:08:44] Yeah, it's a lot more subjective for them than just your, your, your world of just hard numbers and, and time spans and stuff.

Right. 

Will: [01:08:50] I remember the hours I spent in the old days at maximum PC talking to the art designers about using nearest neighbor up rezzing. When I was giving them these really low Rez quincunx versus super sampling aa shots. And I was like, look, you got a neck, nearest neighbor, this or else, it's good. You're going to, you're going to by linearly filter out all of the difference in the pictures and they're going to look exactly the same and then we'd get the fucking magazine back. And because it's, uh, you know, it's, it's a print thing. The dots are only so small. It was blurred out. Anyway, it was a complete 

Gordon: [01:09:19] yeah it didn't matter

Will: [01:09:20] the YouTube guys have it so easy.That's all I'm saying. 

Gordon: [01:09:22] Do you think though, it's going to get back to the visual quality Wars of those days, where they would, um, they would, so for, for those are not old timers, ATI, and nvidia would snipe at each other saying your AA sucks , look at this. 

Will: [01:09:35] or like color rendering. They would complain about color rendering back in the day, your colors are washed out.

don't get the washed out colors. Um, I, I mean, I think that the thing that's changed is that the worlds are so visually complex now and the, because of the pipelines are fully programmable in ways that weren't, they weren't back then it's really up to the art directors and the designers of the games, what they look like.

And, and like, we're not, I think the challenge is. Does AMD's DLSS implementation or Microsoft's DLSS implementation deliver the same kind of frame rate smoothing at the gate. The resolutions people actually play games at less than how, like you're going to be able to snipe jank out of these machine learning upscaling algorithms, wherever you like.

If you, if you pause the game, get loaded up into photo mode and start looking around, you're gonna be able to find weird stuff where the machine learning does weird mill machine learning stuff. Right. And there's not going to be any. Rhyme or reason to it. Cause the machine learning is weird often and you know, you're.

I don't think there's going to be a clear Victor, no matter what in that, in that space, I guess. I think it's, I think it's, you're going to look at performance and you're gonna say, okay, this one, the 99% frame rate, uh, rate is here. This one is here. I think this would probably better this go round 

Brad: [01:10:56] Gordon had a really good point though, about the console specifically, the X-Box being, uh, AMD based since it's direct X compatible, that a lot of stuff will trickle down from consoles back to PC for the AMD hardware.

Cause that's just how it works. You know, like console is kind of dictate development for everything. 

Will: [01:11:10] Well, and it seems like this, like, it seems like there's a benefit to having what is essentially the native direct X GPU, right? Like in terms of, if you want to play the big games that are coming out on consoles, it seems like this maybe is the right choice.

DLSS aside. 

Gordon: [01:11:27] yeah, 

Will: [01:11:29]we have  Brad. I know you'd have just a couple of minutes left. Um, 

Brad: [01:11:33] should we just quickly run through the models? 

Will: [01:11:35] I think we should do that. 

Brad: [01:11:36] There's what is a 6,800 6,800XT and 6,900XT? Yeah, those do those map pretty cleanly to 30, 70, 80, and 90. 

Will: [01:11:46] It seems we don't know performance yet. I mean, we know the performance they said I'm AMD is generally pretty reliable on that.

Um, but I'd like to wait until the actual benchmarks land and we don't have these cards. Um, the, the memory bus is the same on all three cards, which is really a trip. They, I don't, I didn't, I wasn't able to find memory clocks. Everybody was talking about mega Texels for mega textures per second at, at, and bandwidth. Uh, I mean, I could have done the clock math, but it was like three o'clock in the morning. I was putting together the show notes last night, after watching all the election stuff. And I wasn't capable of that. 

Brad: [01:12:17] Um, Steve Kornackie’s not the only one who can't do math. 

Will: [01:12:20] Look, it's math and math on math doing live math is hard, but the prices are, but the 6800 is 580 bucks. It's got 38, 3,840 shader units. The XT is 4,608 shader units for 650 bucks. And the 6,900 is a grand for 5,100 shader units. The compute cores are like, they jumped from 60 to 72 to 80. And like, that's the place that I'm kind of interested to see what the impact is.

I think like, I think you're going to do either the, either of the 68 hundreds are going to be fine at your 1080P's your 1440p’s. And probably when they get their DLSS equivalent online at 4k as well for single-player stuff, especially like if you're playing single-player stuff at 4k 30, you're, you're going to be fine on any of these cards.

The 6,900 feels like. Card that like shoots the gap between the 30, 80, and 30, 90 on paper. And I don't 

Brad: [01:13:14] and in in price for that matter

Will: [01:13:15] that. And in price 

Brad: [01:13:17] a thousand dollars, 

Will: [01:13:18] I don't know I'm going to be, I'm super interesting with the benchmarks look like I bet it's a real low, I bet it's a part that they like.

If you're gonna spend that much money, you're probably gonna buy a 30, 90. And get go all the way. Sure. But I mean the 30 90 has a lot of weird overhead. Like you got to it's, it's pretty much the maximum size for a PCI express card. It's um, do you need a giant power supply for that thing? It's a big chunker like maybe, and it takes up two and a half, three slots.

So like maybe this, maybe since this, this does, this is a normal sized card. Maybe that's the argument is if you need the most power that'll fit in your mini ITX case. Here's here's your, here's your best bet. 

Brad: Sure

Gordon: [01:13:57]. Yeah, less power. I do think that the arc 6,900XTis the most disruptive of those because yeah, they, you know, I think nvidia really did sort of price that 30, 90.

Probably not expecting AMD to be as close as, as they claim they are. So I think the, the 6,900 XT, if it's, you know, 98%, the performance of a S you know, a 30, 90, then a lot of people are going to say, well, why do I want a 30 90? Right. So 

Will: [01:14:28] I think even if it's 90, like if it's, if it's 33% of the price, For 80, 90% of the performance, and you're not doing heavy metal like this, like realistically the 30, 90 is a card for people who want to play games at ridiculous resolutions, professional streamers, and like hardware nerds.

Gordon: [01:14:45] machine learning, scientists

Will: [01:14:46] machine learning, scientists,yeah stuff like that

Gordon: [01:14:48] It's, it's sort of pitched for that. It's, it's almost, it was priced at a price where they thought like, you know, no, one's going to buy this. I think that's where AMD did most of their. Messing with Nvidia. The other parts are probably correctly priced to compete. And you know, there, I think the price is sort of tell you about where they're going to perform.

You know, there'll be right there with their equivalent cards and. You know, they were going to expect that, you know, Nvidia might cut prices a little bit, or maybe just it'll be more attractive. 

Brad: [01:15:17] That's that's really the thing to me here on both on both fronts, CPU and graphics. It's like, it's just great to see AMD come out, swinging again, after so long, we need to be even more competition in that space for pricing, for feature development, for availability, for that matter.

Like it's nice that it's nice that these AMD cards are. Pretty good, apparently in a world where nobody can get their hands on the 30, 80 right now. 

Gordon: [01:15:37] Yeah. They are very confident that they are, they are back, they are going to be swinging with and yeah. Yeah. 

Brad: This is better, better for everybody. 

Gordon:I mean, as we saw with Intel, uh, it would be nice if we got to that with Nvidia, but I don't think we're there yet because of the, you know, Nvidia is not in the place that Intel has been.

So I think Nvidia will,  it was sort of, they're always four steps ahead of everybody. It feels like sometimes

Will: [01:16:00] their R and D pipeline is deep and well-funded so 

Brad: it's crazy. 

Will: Yeah. Um, I guess that'll wrap it up for us, uh, as always, uh, this is the part of the show where we think our patrons, the wonderful people who make the show possible every week.

Uh, and, and thanks to all. Let's see we're at, uh, hold on. We are at 

Brad: [01:16:21] drum roll 

Will: [01:16:22] 1,259 patrons now. So thank you to all 1,259 of the wonderful people who make the show possible, 

Brad: [01:16:27]I just heard the little windows chime sound in my head when you said that. 

Will: [01:16:31] Um, but it's a special thanks to our executive producer tier patrons. Jacob Chavel, Andrew Cotton and David Allen.

As always an extra special, thanks to Julian for taking care of the transcripts for the tech pod for us. If you would like to read the tech pod machine translated with the help of our good friend, Julian, you can, uh, they're on the Page @ techpoddotcontent.town. And please tell your deaf or hard of hearing friends.

Uh, so, and let us know if you're using the transcripts so that we know whether it's worth continuing to do them. Thanks, Julian. If you would like to know how to join the tech pod discord, the fabulous, fabulous. And well-informed tech. Hey Brad, there's a new channel. Brad:Oh, 

Will: wait. No, that was the other discard.

Brad: [01:17:16] No, you have to make that shift. What was, what was the channel in other discord

Will: [01:17:18] Well so the other discord has a  Booze, booze and drinking channel, which isn't like my jam, but people are really into it. Um, a parenting channel because we don't have a parenting channel and there's a lot of parents

Brad: That’s not the worst idea.

Will: yeah. 

Brad: [01:17:30] Yeah. We have a food channel where occasional whiskey talk pops up, so that's fine.

Parenting would go well, right under work and career stuff. 

Will: [01:17:37] That’s true. 

Brad: I would say 

Will: that's a grownup conversation. 

Brad: That's right. 

Will: For grown up places. I've been enjoying the movies and TV channel that you added a couple of weeks ago though. That's been, that's been nice for me to come and talk about my star Wars business. 

Brad: [01:17:48]  just hang out at a time when people need a place to hang out 

Gordon: [01:17:51] your star Wars business.

Will: [01:17:52] Well, how do you feel at the Mondo Gordon? 

Gordon: [01:17:54] I like it. I feel it's everything. If you could sort of, if you could take every movie made since the end of return of the Jedi and just basically blast them into, so our Sarlacc pit and we just had the Mandalorian. I’d be totally fine with that. It satisfies the whole star Wars craving.

I mean, the closest thing it's probably gotten to me is a rogue one, the end of rogue one, anyway, I mean, that was really very star. We didn't know everything has to be about Han Luke and Leia, you know, and, and I think it really does it well. 

Will:Agreed. 

Gordon: So 

Will: [01:18:29] if you want to find out about the tech pod discord, how to, how to back the patron, you can go to patrion.com/tech pod that's patrion.com/techpod Gordon.

Uh, this is the part of the show where we plug things. 

Gordon: [01:18:40] Where can people find you? You can find me over@pcworld.com and also you can catch our live show. Oh, the full nerd on YouTube Twitch. And we do that live. So we take questions. It's awesome because, um, you can actually have conversations with us in real time on YouTube and they're reasonable conversations.

I'm going to say that reasonable conversations on YouTube live with people. Yes. 

Brad: [01:19:04] What is this? What is this promised land you have discovered? Bring me there. 

Gordon: [01:19:08] Yeah. Yeah. 

Will: [01:19:09] If people want to find you on Twitter or anything like that. Are you still doing social media 

Gordon: [01:19:11] Oh yeah I think,I have no idea because I don't know how to use Twitter and never got the manual, but I think it's Gordon at Gordon on Twitter

Will: [01:19:20].Very good 

Gordon: [01:19:22]Oh I had to check. 

Brad: [01:19:23] Thank you for letting us pick your brain

Gordon: [01:19:26]Not that’s it because there’s  an actual person who has Gordon mam on Twitter. Who's not me. 

Will: [01:19:31] Oh, I'm sorry. I look, I didn't have that problem on Twitter. I managed to nail that one, but, um, uh, yeah. Thank you so much for coming by Gordon , we'll have to have this again, sometime this really fun.

Brad: [01:19:42] really appreciate it.

Will: [01:19:43] Like, I feel like I should get you to yell it. So I was hoping I was gonna ask you about the Mandalorian. You're gonna start yelling, but, um, I guess 

Gordon: [01:19:49] No it's so clever. It's so clever, you know, and it's just, 

Will: [01:19:52] have you looked at how they do the volume, how the volume, how, the way they shoot that thing works. Have you watched any of the behind the scenes on that?

Gordon: [01:19:58] I haven't, but I was talking to my friend who was looking for like media advice on how to pitch the thing she's doing. And she's like, we're actually doing it. They're not doing it, the Mandalorian, cause they're still doing post effects. So 

Will: [01:20:10] well it's, it's the, the behind the scenes on Disney plus for the first season of the Mandalorian shows how that, how the volume works and how they use unreal to like change backgrounds and stuff live and how they get all those reflections on it's a it's unbelief.

It's magic. Uh, super cool, but, uh, thank you so much for coming by every check out Gordon at PC world and the full nerd, and we will see you all next week. Bye everybody. .