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 Arrow Lake, Panther Lake, and Future Zen Architectures w/ High Yield (Guest Telegrams) (Patreon)

Published:
2025-05-09 03:40:10
Imported:
2026-06
Tags:

Content

The next episode of Broken Silicon will see us be joined by High Yield once again! Obviously you can ask him ANYTHING about GPUs, CPUs, or semiconductors you want - but of course we will also be focusing on Arrow Lake, Zen 5, and RDNA 4 at a minimum.

Additionally, I (Tom) plan to share some stuff coming in a future Zen leak with him ahead of time so we can at least talk about Ryzen's trajectory at a macrolevel to a degree before said leak(s) come out... ;)

You have ~12 hours to submit!

Last Episode HY was on: https://youtu.be/EdVLIw6-TKg?si=by1dMj6E2IXI3SMK

HY Arrow Lake Deep-Dive: https://youtu.be/wusyYscQi0o?si=ZwDq7ifiLvoEKQGq

HY RDNA 4 Deep-Dive: https://youtu.be/u8cfrJTdo0E?si=OIH_sal4qi7-bhON

HY V-Cache Video: https://youtu.be/OlRLuajAgIc?si=p8awdBfTfMb1V3YX

Comments

Smashaxely

Hello I was wondering just where does die size go from here. I mean TSMC is developing 2nm right now(or finalizing it) but how small can they really go? Are we reaching a plateau on die size or is there some fancy tech to allow even smaller forms? 1nm would be next but after that would you switch to picometers?

AnotherChris

Hello, I have a question about cpu cache, if they keep increasing it how much more can they do until the gains get smaller? Will they start doing more interesting things regarding how items are stored and moved through the cache? What do you think the biggest bottleneck in the CPU currently is?

KarbinCry

Hi. For people not watching your stuff, High Yield, can you quickly go over what you think the V-cache packaging process is like step by step?

AI-Soldier789

Good day, AMD should, if they can not cancel the 9060XT just print new boxes and name the 8 gig variant the 9060. Reviews will be horrible for the 16 gig version even if it is an outstanding product for the same reason as the 5060TI. Even worse i would argue as reviewers will point out the negativity around the RTX 5060TI launch.

kwerboom

Hi. I've watched some of High Yield's YouTube videos and found them above my head but fascinating to listen to for all the details. Now for my questions. Was Zen 5's uninspiring benchmark performance (a.k.a. Zen 5%) really partly down to the memory controller? I watched the Level1Tech's video showing Windows had issues that Linux didn't and later on Microsoft cleared those up with patches, but that still didn't provide a "magic" massive boost to performance. Also, tests in other reviews later on between the 9700X and the 9800X3D seemed to imply that the 3D V-Cache is doing the heavy lifting of providing a buffer to make up for bad system memory access. Would AMD have been better off designing a new I/O chiplet with a better memory controller for Zen 5? Could you see a chunk of Zen 6's performance uplift vs Zen 5 in reviews just being a new I/O chiplet?

Nicholas Buckner

Hello. What design would you provide for the Steam Deck 2? Tom has talked about 1W-per-core APU in future AMD architectures that could provide some insane battery life. Would you like to see that chip scaled 2x/4x inside a pc with a CD-player sized form factor?

KarbinCry

Do you think the shattering of global trade links could result in regionalisation of computer technology development? For example in CPU. Japan/Taiwan/Korea have Fujitsu, Andes, Samsung. Europe has some ARM (Sophia Antipolis) and the EPI. Could this be how x86 falls?

Aluveitie

Today's best CPUs (in terms of single core performance and power efficiency) come from Apple. For one, M series use a 16k page size and Apple optimizes the cache for it. Meanwhile AMD64 is still stuck 4k for backwards compatibility and they still need to build their CPU for that. As the cost of keeping backwards compatibility is ever increasing, how can AMD and Intel move AMD64 forward to keep up? At which point will the cost of keeping backwards compatibility will be too high?

KarbinCry

Sorry I haven't yet found the time to watch High Yield's analysis of Arrow Lake die. Based on what you saw, did Intel's decision to cut out SMT pay off in significant area savings?

KarbinCry

How would you describe current core architectures in one word (I will give examples for me): Lion Cove - bloated Skymont - misled Zen5 - misunderstood M4 P-core - wide M4 E-core - useful Tellum II (IBM system Z) - GOATED

Falto

Is Intel going to be competitive at all for the foreseeable future? With them canceling Jim Keller's project and with Arrow Lake being a regression in many ways, I have a feeling that unless AMD pulls a Skylake stagnation (what Intel did from 2016-2021), Intel will always be behind AMD.

KarbinCry

We can say that Raptor Lake is running out of breath, Meteor Lake made no splash, Lunar Lake is seen every once in a blue moon and Arrow Lake missed the target. What do you think Panther Lake will do in this series?

KarbinCry

Does either of you have experience running Arrow or Lunar Lake machine? Is the scheduling better, and if so, is it good enough? Can we expect Panther Lake to work well?

KarbinCry

With Zen7 or 8 using TSMC 14A, do you expect these to have 3D stacking as requirement, facilitated by BPD removing the issue of routing power up through the cache die?

Dark Side of the Force

Hallo. I was in a discussion earlier in Discord about AMD abandoning low end product, instead just shift their high end product into lower and lower tier as the time goes on. Just to coincide with Intel dropping price on Arrow Lake. Is there a path for Intel to compete on price again when their current gen product are all expensive to make?

crumbling_earthblocks

When will I be able to play Witcher 3 in 4k/90/ultra/RT on a mainstream APU like Strix Point?

Dark Side of the Force

Have you seen any details on Nova Lake Desktop that justify it being on a new socket being only ~+100 pins over current LGA1851? Would it be a big mistake for Intel to do so?

Cleansweep

Given the state of the market, how do you think AMD will position Gorgon Point AM5 APUs? With the GPU price and availability issues, will there be a temptation to milk the market by pricing high?

Dark Helmet

What elements of RDNA 4 made is such an architectural improvement over RDNA 3? What do you expect of Nvidia's upcoming architectures in terms of gaming improvements vs DC AI improvements? Is there a next "big thing" technology to improve graphics realism or just more of the same ray tracing/path tracing?

Dark Helmet

How is High-NA EUV an improvement over current EUV tech and what improvements in CPU, GPU, DRAM, Cache would it enable?

Dark Helmet

What do you think of Intel's foundry prospects while not being a fully separate company? What can they do to gain the trust of their prospective customers?

transpose1195

How do you see chiplet usage continuing to evolve? AMD seems to mostly continue their pattern of chiplet reuse across products with a few one-offs here and there, while Intel doesn't really seem to be following this model. What are the pros and cons of these different strategies?

Nicholas Buckner

Given 3D V-cache helps with battery life, which handheld do you think would see this used first? Vita 2 / PS6 handheld? Switch 3DV? Steam Deck 2 / 4? [can they even count to three]

Chris Rijk

AMD and TSMC recently did a little press release saying that AMD would be the first to use TSMC’s upcoming 2nm node. I think the good news from this and Tom’s leaks is that AMD is going to be aggressive about adopting 2nm. But rumours about 2nm wafer prices and the fact that Apple wasn’t first suggests to me that it was too expensive for them and that Zen 6 could be quite expensive at launch. Would you agree and what are your thoughts on the trend for the latest silicon becoming more and more expensive?

Chris Rijk

Are you aware of any major new microarchitecture features for advanced CPUs or GPUs that could help improve performance? Intel’s Rentable Units come to mind but I don’t know if that will ever see the light of day. Are we just going to get “more of the same”?

Chris Rijk

Arrow Lake has not performed well but do we really understand why this is the case? What parts of the design do you think most need fixing for Panther Lake and beyond?

Beter

Any bets, informed or not, on what an RDNA5/UDNA config will look like?

Prior is the theme 999

There hasn’t been enough time to give an opinion, but what are your early thoughts on Lip Bu Tan? Is he positioning the ships rudder to a better course or does it appear that nothing has changed at all?

Red

Hi, Do you think Blackwell has some sort of hardware flaw? I saw recent reviews for Doom dark ages, and the 5080 loses to the 4080 super constantly, and does not seem to be the only game. I find this weird for an architecture that is pretty much Ada refresh. And for Tom: Any news on Strix Halo laptops, besdies mini PCs? I want one for AI and gaming on the side, and can't find any in sight.

Kiln God

Tom and Dan - My only question is when will Halo next be out with Zen 6 cores and 120 gpu cores?

TechnoLadz

120 GPU cores is probably unlikely lol 😂

Prior is the theme 999

About a year ago high yield posted a video on backside power delivery and said it has the potential to be the next big thing in semiconductors. Curious where your opinion on that is whether you see something better on the horizon or with the recent APU push and the benefits backside power will have to them and handles/phones that this is it definitively it.

TechnoLadz

With Intel officially showing documentation for Intel 10A, which looks due Late 2027 or 2028, what products do you expect to be on it? How well do you expect it to compete with the likes of TSMC's A14 node?

TechnoLadz

Do you think AMD will release a 9600X3D, or do you think they will skip 6 cores for X3D again too, or make it a MicroC exclusive like they did the 5600X3D?.

Nicholas Buckner

Why did it take AMD (and Nvidia on Windows) so long to start to develop on ARM? I know the instruction set is constantly being thrown out and refactored every few years, and the license holder isn't the nicest, but... What do you think of the technical possibilities of AMD developing for ARM? What would you do to clean up the x86 instruction set(s)?

TechnoLadz

If Xbox is going into the handheld game for 2027+, do you think they would use X3D cache, since that would help them get additional performance at lower voltage?

Prior is the theme 999

If Nvidia, AMD, and Intel respectively hired you for your input to win the next gpu generation how would you do it? What individual advice would you give them so they can beat the other?

Prior is the theme 999

Is Nvidia blundering or is RADEON doing something really impressive? A smaller die drawing less power for identical performance? If they made 41 die do you think they would’ve won the crown?

Kiln God

I figure an Instinct MI325 has 306 compute units... therefor I'm not asking that much! LOL

Kolapso

What are your predictions about intels next DGPU, be it Battlemage or Celestial? Thanks!

Snarfnado

With regards to AMD and zen 6, do you have any idea if they will finally attempt to put dual caches one each ccd for next gen X3D. I think this would be a huge win for AMD to leverage this as another jump forward in what they offer over intel.