Question Zen 6 Speculation Thread

[inquiss]

Define "budget/mainstream stuff". We've gotten off into the weeds a bit. What is the leakers consensus? That consumer desktop will use N2 for CCDs and N3 for small monolithic APUs?

Post in thread 'Zen 6 Speculation Thread' http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/zen-6-speculation-thread.2619444/post-41456108

 

[Josh128]

Nope. Intel had mountains of cash and here we are. It is doing the disservice to the design teams to say it's all thanks to the process.

Other than that, I agree. The 5d chess people are playing to convince others their guestimate is better is a bit boring.

That happened to Intel because they chose to rest on their laurels rather than continue to drive advancements despite lack of competition. They wasted a shat ton of money on new ventures and subsidy bribes to put their foot on AMDs neck instead of letting their technology and performance do that for them.

Apple has apparently learned from Intels hubris. They have shown no desire to rest on their laurels despite having the fastest 1T perf in the world.

 

[igor_kavinski]

Nope. Intel had mountains of cash and here we are.

True but they wasted that money on a lot of side projects that didn't go anywhere in the end. Plus, they neglected their foundries and kept accepting excuses for the process delays from whoever was the head of foundry at the time. They should've kept firing the person in that position until they got someone who actually knew what needed to be done to make the process work.

 

[511]

lol who will dethrone them as the 1t king?? Nobody else is currently even in the same ballpark.

ARM x930? also since it's come to ST i think AMD and Intel have a chance this time to get back the ST crown from ARM especially Intel with APX since it helps quite a lot from their initial findings.

Here is the proper ST benchmark not Meme-Benches after the addition of SME. If Intel and AMD can get a SPEC score of 15-16+ i think they will get back on top.

SPEC CPU 2017 | David Huang's Blog

 

[Thunder 57]

True but they wasted that money on a lot of side projects that didn't go anywhere in the end. Plus, they neglected their foundries and kept accepting excuses for the process delays from whoever was the head of foundry at the time. They should've kept firing the person in that position until they got someone who actually knew what needed to be done to make the process work.

You sure do like to fire people. "Fire anyone who disagreed with Jim Keller!" "Keep firing people until they find someone who can make the process work".

 

[511]

You sure do like to fire people. "Fire anyone who disagreed with Jim Keller!" "Keep firing people until they find someone who can make the process work".

If he were the CEO there will be 2 employees left Jim Keller and Him 😂🤣.

 

[igor_kavinski]

If he were the CEO there will be 2 employees left Jim Keller and Him 😂🤣.

I'm a very serious but realistic taskmaster

I don't know about anyone else here but I keep seeing people in positions of power that they could never deserve. These people keep leeching off of corporations and society because they have mastered the art of bull****. I hope all such people at Intel will get weeded out before the end of this year.

 

[511]

I'm a very serious but realistic taskmaster

I don't know about anyone else here but I keep seeing people in positions of power that they could never deserve. These people keep leeching off of corporations and society because they have mastered the art of bull****. I hope all such people at Intel will get weeded out before the end of this year.

Don't worry Lip Bu Tan is the perfect guy but none compares to Andy Grove.

 

[OneEng2]

do you think AMD can beat Intel on costs? This is not possible considering the only reason Intel is selling so well cause the customer is buying cheap Raptor Lake.

Yes. I believe that AMD's cost structure is far superior to Intel's. I also believe that Intel maintaining market share at the expense of profit. The reason that they are selling Raptor Lake cheap is because that is all the market will bear. ARL sales are kept low on purpose because they lose more money on every ARL they sell (because it costs more).

Intel is simply minimizing the bleeding.

Key qualifier: datacenter CPU market.

Datacenter CPU market is ~$5 billion per quarter
Client CPU market is ~ $10 billion per quarter

A lot of "conventional wisdom" in this area is just retardation. You will see this from many otherwise knowledgeable posters. Saying that datacenter profits are somehow superior. But it is just falling for retardation.

For example, AMD datacenter profit was $932 million but it's possible that nearly a half of it was from GPU. So perhaps $500m from server CPU.

Client CPU profit was $496m, nearly equal to datacenter CPU.

This is just quick illustration that "giving up" client "in interest of" server is nonsense.

To zero in on just the CPU market:

Client: 29bn: https://www.linkedin.com/pulse/united-states-x86-microprocessor-market-size-top-k7i8e/

Server: 14.7bn: https://www.custommarketinsights.co...arket is expected to,USD 48.9 Billion by 2034.

Note: The profit and growth per unit lies in DC. Also Note, I have only suggested that it would be a viable strategy for AMD to optimize the design Zen 6 for DC vs Desktop because of the growth and sales. I have NOT suggested that they can not maintain superior performance in both. I only suggested that IF they had to choose, they would lean into DC.

If leakers are telling you "no they're using N2 on everything except budget/mainstream stuff" then what's the point in attempting to deduce the accuracy of this leak? Either they're right or they're wrong. Nobody arrived at that conclusion (that N2 would be used in server/workstation/premium consumer) through deduction.

Either the leaks are accurate or they aren't, accept it or don't.

Crazy how people speculate in a speculation thread .

 

[511]

Yes. I believe that AMD's cost structure is far superior to Intel's. I also believe that Intel maintaining market share at the expense of profit. The reason that they are selling Raptor Lake cheap is because that is all the market will bear. ARL sales are kept low on purpose because they lose more money on every ARL they sell (because it costs more).

Intel is simply minimizing the bleeding.

the only thing affecting cost structure for Intel is Intel 7's cost rn and also their investment in Fabs but once they migrate to 18A in volume these will be sorted out along with Intel 4/3.

 

[igor_kavinski]

https://twitter.com/x/status/1692989174909997350

Have these temporary registers for APX and AMX been discussed here?

 

[MS_AT]

Have these temporary registers for APX and AMX been discussed here?

There is nothing to discuss. If you want to know more read up on calling conventions TLDR to best of my knowledge they are simply called temporary because they are not required to be preserved across function calls.

 

[Doug S]

Not really got do with cash, Apples hardware Sillicon engineering VP just wants to be number 1 all things chip related. On the other hand look at Apple AI performance, they suck at that cause the VP for AI is a weak person.

It’s all to with a goal and people that demand that change. 2026 is year where Apple might get dethroned as 1t king cause the competition is really high. We got like 5-6 CPU architectures and all on N2.

It isn't like Apple has held #1 ST ever since M1 was released. They were briefly #1 with M1 then they got beat by newer x86 stuff. Pretty sure the same happened with M2. It wasn't until M3 that Apple took the lead and remained in the lead.

If we recall just over a year ago the general sentiment on this board was that AMD and Qualcomm were the ones to watch in summer 2024 and Intel and Apple were going to make weak releases with hardly any improvements over their previous generation. Apple ended up releasing M4 a few months earlier than expected and got bigger gains than anticipated by most, AMD didn't come close to living up to the crazy Anandtech 40% hype, Qualcomm failed to set the world on fire so the only call the "general consensus" of this board had right was Intel.

Now it seems like most expect Apple to hold the ST crown for the foreseeable future, when the M1 & M2 experience showed that a lot of was timing i.e. whoever had released most recently. Yes Apple has M5 coming up this fall but there will be a lot of others releasing products between M5 and M6, there's a chance we see some on N2 before Apple ships its first N2 since TSMC's production cycles haven't been aligned to Apple's iPhone cycle as well with N3 and N2 as it was with N7 and N5.

 

[Tuna-Fish]

There is nothing to discuss. If you want to know more read up on calling conventions TLDR to best of my knowledge they are simply called temporary because they are not required to be preserved across function calls.

Yep, terminology varies, in other places they might be called "caller-saved".

 

[igor_kavinski]

Yep, terminology varies, in other places they might be called "caller-saved".

Could these temporary registers be used by AMD for supporting APX and AMX extensions in Zen 6?

Would they need to create additional temporary registers after reserving these?

 

[Tuna-Fish]

Could these temporary registers be used by AMD for supporting APX and AMX extensions in Zen 6?

Would they need to create additional temporary registers after reserving these?

The terminology is still confusing you. These are just the architectural registers for APX and AMX.

They being called "temporary" means that after a function call, the function that was called can freely use them to store values without first preserving what already exists in them, and likewise cannot assume that their values will be preserved across another call.

An ABI must define these things so that different compilers can emit functions that can be linked together.

 

[igor_kavinski]

These are just the architectural registers for APX and AMX.

BUT BUT BUT...will Zen 6 have APX and AMX?

 

[Tuna-Fish]

BUT BUT BUT...will Zen 6 have APX and AMX?

I dunno. This document can't tell you, because it only defines how APX and AMX registers should be handled, if present. AMD64 just the proper name for the ISA that's often called x86-64, it doesn't imply AMD is involved.

 

[Kepler_L2]

BUT BUT BUT...will Zen 6 have APX and AMX?

No

 

[OneEng2]

the only thing affecting cost structure for Intel is Intel 7's cost rn and also their investment in Fabs but once they migrate to 18A in volume these will be sorted out along with Intel 4/3.

Hardly.

Intel has more mouths to feed, fab plants to feed, Fab R&D to feed, 18A development to feed (and future nodes), High NA to feed, etc, etc, etc.

Currently, on top of all that, Intel is ALSO feeding TSMC.

My contention is that even if Intel were to go back to being 100% vertically integrated, their cost structure per wafer would be WAY higher than AMD using TSMC (even if they used all N2 for Zen 6).

 

[MS_AT]

No

No for both, or no for AMX only? I really could use APX for my AVX512 code...

 

[Kepler_L2]

No for both, or no for AMX only? I really could use APX for my AVX512 code...

Both

 

[MS_AT]

Both

A pity, maybe Zen7 then

 

[Thunder 57]

I'm a very serious but realistic taskmaster

I don't know about anyone else here but I keep seeing people in positions of power that they could never deserve. These people keep leeching off of corporations and society because they have mastered the art of bull****. I hope all such people at Intel will get weeded out before the end of this year.

I absolutely agree. I've seen to many buttheads in positions they don't deserve because they can speak the right BS, or because they are buddies with the right people. One of my favorites was an old boss who, whenever there was an internet issue at a remote site, would say their was a "national outage".

 

[Thunder 57]

Both

I didn't think AMD was even going to bother with AMX. Surely they plan on using APX though?