Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

Discussing Next-Gen CPUs, GPUs, and consoles w/ High Yield [Guest Telegrams] (Patreon)

Published:
2026-02-27 21:33:54
Edited:
2026-02-27 21:34:10
Imported:
2026-02
Tags:

Content

Once again Max will be joining Broken Silicon! This time we expect to discuss what's going on with Nvidia's graphics, Panther Lake, and of course all of the next-gen AMD products that will be rolling out over the next few years....and that obviously includes Zen 7!

So in reality - ask us anything! This episode will be HEAVILY steered by what YOU want us to talk about! You have ~48 hours to submit, but it would be appreciated if you submit by this evening.

Check out High Yield: https://www.youtube.com/@HighYield/featured

Last time High Yield was on: https://youtu.be/dDVeH3gBOtI?si=7345wvMzwrPFBkOQ

Recent Zen 7 Leak: https://youtu.be/huRWonnGiCo

Recent RTX 5090 Ti ES Leak: https://youtu.be/p5pZuUmpIyY

Nvidia Graphics in Intel Leak: https://youtu.be/hvuN_4C5fx0

High Yield's PTL Analysis: https://youtu.be/g3xpRXD0I30

MLID PTL Analysis: https://youtu.be/Lc05AAMG0Xo

Comments

Lucas

Resident Evil Requiem appears to use PSSR2. What are the chances Sony lets AMD copy their homework again and we see FSR4 on Rdna 3 or at least 3.5 in a reasonable timeframe? I think it’s kind of silly it’s taken this long already.

BespokeEP

I don't think it's that simple, AMD and Sony developed the base of PSSR/FSR together. Then Sony would have tailored made parts of PSSR to suit a console. Since PSSR launched first (real world testing is one of the most important things). PlayStation/AMD were able to see problems and fix them for FSR 4. Then Sony started to add those fixes to PSSR all while making sure things would also be compatible with PSSR 3.0 on PS6 which I imagine is why things took so long. From the sounds of it, devs that already have PSSR in their games don't need to do a thing, they will automatically be improved when the official update rolls out in a few weeks. This puts PS6 at a huge advantage over Magnus, where at launch PS6 will have 100's if not 1000+ games that will instantly run better with no additional work from the devs. Where Xbox will have to ask each dev to go back into all their old games and add FSR5? support, which I imagine most would be reluctant to do. And if games have fixed res or a lower DRS bound... Capped frame rate.. it won't matter how much more powerful Magnus is, because all those games were designed with limits and PS5PRO extended those limits. PlayStation definitely thought about this when deciding to release the pro or not

BespokeEP

Sorry I drifted from your actual question with my own excitement for the future. PSSR has been so heavily customer into a console environment, Sony's low level API... It would be a lot of work to port it to PC. I don't think people realize why AMD won't ever bring it over... Cause then they have a commitment to fixing things, supporting it... It would be a lot of work, especially for a company who wants you to upgrade... Nvidia has been doing it for years, don't really get why AMD is getting all the hate

Lucas

Half the FSR4 implementations so far have been through the driver side upgrade. To give devs a real reason to implement it properly they need more compatible hardware out there. It has been proven beyond all doubt that it is possible, so I don’t understand why they drag their heels on this. On top of this, FSR4 hasn’t materially changed since its initial release other than calling it FSR redstone, presumably because they are morons and every good idea needs to be matched by a bad one to restore balance. I can’t write a line of code and can’t even estimate the work required to port PSSR to PC or improve FSR4. But I do see what NVIDIA is doing with DLSS and although the number of models and the different use cases one or the other might be most suited for are getting harder to navigate, the fact is undeniable that they are making great strides with their image reconstruction. AMDs software team has made some Ray Reconstruction equivalent in the meantime that is implemented in one game, seems to be having some detrimental effect on image quality in that game and is in general a beta feature at most.

Lucas

Do you see any reason, apart from product segmentation of course, why consumer platforms are still stuck on 28 pcie lanes? Soon we will be able to get 32 core CPUs on a mainstream mainboard, but are still restricted to one x16 slot and a couple ssds at most.

Gob

Seeing PSSR2 in action on my PS5 Pro on my TV at home in RE Requiem blew me away, more than PS4 Pro ever did. Have you seen PSSR2 in person yet on your TV? If so, what are your thoughts? You’ve said many times that 4090 performance is more than enough for most people. Do you think PS6 can be that for consoles? Path Tracing, 4K120 capability, fast storage and good AI upscaling. Beyond that, it seems like the main thing PC is bringing to the table is higher framerate and frame generation. Are there some “next generation” features that will be worth caring about?

TheParadoxy

You (Max/High Yield) recently did a Zen 6 mock up for semianalysis where it's stated that Zen 6 server uses bridge dies. What are your thoughts on how this advanced packaging will transform compute? Could you imagine this technology being employed in future consumer GPUs for faster memory? I know you've talked about the lower latency of strix halo with the sea of wires, do you expect bridge dies to have a similar benefit? The semianalysis uses the label CoWoS-L in the BoM estimate. Do you view this as a general term for this type of packaging that encompasses things like ASE's FoCoS-B? How interchangeable do you think these advanced packaging technologies are? Sorry for all the questions please answer as many or as few as you like.

Valentyn

Greetings Tom and Max! I keep thinking about the Intel and Nvidia partnership. So far this year, we've seen Intel release Panther Lake with impressive iGPU performance, as well as Nvidia killing high-end gaming GPU supply to focus on the AI demand. (Here in Switzerland, the 5070 Ti currently costs 50% more than the 9070 XT as a result.) And I keep thinking what this means for Titan Lake. Now, 2029 is admittedly still far away and it's impossible to tell what the market will look like then - whether there will still be a huge demand from AI data centers, or perhaps some kind of robotics and self-driving cars compute, or even trump coin mining. But if I imagine Titan Lake launching in today's market, I can see a lot of ways that this could backfire for Intel. Will there be sufficient GPU tile supply? Stable graphics drivers? Efficient power management between CPU and GPU? I guess what I'd like to discuss here is: will adding Nvidia GPU tiles benefit Intel? Or would Intel be better off continuing to focus on their own (Xe) iGPUs?

Valentyn

I guess we have to discuss the AI-generated elephant in the room that is the shakeup of XBOX leadership with Asha Sharma taking charge. Having an AI executive with no video game industry experience in charge makes it difficult to believe in Sharma's plan to "bring about a return to Xbox"... but since we are anyway just speculating at this point, let's try to be creative and come up with hypothetical POSITIVE outcomes: can you think of a way that this could turn out well for Xbox and for the Xbox gamers?

Prior is the theme 999

What are your thoughts on nvidia pulling back recent drivers because of them causing hardware issues. This almost seems like they weren’t tested before release. From not releasing drivers to games like Starfield, to poor launch drivers for Blackwell, to now possible damage to a GPU because the fans won’t run. It’s a small trend, but a trend nonetheless.

MelodicArmor

Welcome back to the podcast! Something I have been thinking about recently is the "design doctrine" that AMD, Intel, Nvidia, Sony, Qualcomm, Microsoft and others are executing for all their chips. It is interesting to see how some of them are staying with monolithic designs, primarily for cost reasons. However what I have noticed is that the inflection point, that is approaching fast, is that the packaging costs for chiplets, tiles and 3D stacking is steadily moving to a happy medium. Based off this trend, do you think that a defacto silicon module methodology is going to become the mainstream way of chip design? Personally I think a future where you make a handful of designs and can swap them out like the data cards for Starships in Star Trek is not a far fetched idea. Thoughts?

Valentyn

The core counts on the desktop Zen7 are truly impressive. Finally some progress! But I wonder: won't it be held back by dual channel memory (on AM5)? Compare this to Threadripper and Epyc that starts at 4 channels and goes as far as 12 channels. Are generous amounts of cache supposed to ameliorate that?

KarbinCry

Where do you feel the common knowledge and discussion (in our niche area) fails to properly show the beautiful complexity of computer hardware? You know for me, it's very much the narrowness of discourse. Sometimes it feels it's just Nvidia/AMD/Intel, just TSMC, just ASML. When there are massive players like Broadcom, Mediatek, or GloFo, ASE, UMC, or Lam, TEL, Applied Materials...

KarbinCry

In view of Zen6 using ASE packaging, is the more technical tech press at fault for "CoWoS" being used as synonym for "advanced packaging"? Do you think this issue will be fixed, or will we keep using TSMC branding forever, just as we are incorrectly using "IPC"?

TheParadoxy

Panther lake has looked really exciting and many have said that Intel is finally bringing competition back to the market. But, on the server side, people seem less enthusiastic. Semianalysis called Clearwater Forest a failure saying it has "relatively low bandwidth at only 35GB/s per 4-core cluster in accessing the base die’s L3 and mesh network." What do you think is going on with Intel's advanced packaging and why are the narratives for Intel's consumer hardware so different than Intel server? Does Intel need to have competitive enterprise hardware in order to be competitive in the consumer space? What is the most important thing for Intel to focus on for future consumer products?

Rayan Malik

To High Yield: Did you ever sort out the die size of Maia 200? I remember you got a different result to Dr. Ian, something about the size of SK Hynix HBM stacks not being the same as other vendors.

AnotherChris

Greetings, Max what is your take on a technology or designs that could help alleviate the Ram crisis? It seems like they need an infinite capacity of ram, What in your opinion what could help satisfy these data centers, or is simply more Ram.

EXTENDIFY

what are ur expectation for next Gen memory controllers (zen6, nova lake)

Cleansweep

Between Panther Lake and the upcoming AMD APUs, do you think we've hit the point where OEM office desktops will start transitioning to socketed or soldered on APU systems? Outside of jobs requiring a lot of compute, an APU with a lot of threads and good enough GPU will get a lot of office/light workstation work done.

Jihad ROUANI

I love high yield videos dissecting die shots, when are we getting the next video? 👀

Nicolas Terroir

Hello, why x86 has been so much more effective those past generation ? What change on those die ?

Nicolas Terroir

On the new snapdragon 8 elite gen 5cpu, I have report of those chip consuming around 15W. Is Qualcomm starting to running into a wall in trying to increase performance ?

Nicolas Terroir

Why RDNA4 is not used in APU ? It seems AMD skipping those new gpu in integrated graphic ?

HaveAGoodDay

Has a list of Games leaked about which titles will get PSSR2 upgrade?