• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Question Zen 6 Speculation Thread

Page 389 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
LightningZ71 said:
Arrow lake doesn't have bad single core throughput. Not amazing, but not bad.
Click to expand...
There was that leak from Intel projections putting Nova Lake at just 1.1x single-thread, implying it may not be big jump over Arrow Lake 1T. Which sounds bad if it's legit and not massively sand-bagged but perhaps the talk about Cove architecture lineage and its team being the sucky party of Intel these days are true, and they do need the E-Core/Austin reboot ASAP that badly...
 
adroc_thurston said:
Psure Granite Ridge sits 5G (or upper 4s) straight in Y-Cruncher under adequate cooling.
Click to expand...
Normal or BBP mode? Normal bottlenecks faster on MBW, while BBP in Mystical's teardown article hits current limit and dives under 4GHz.

Generally I have seen dense matmul of right sizes go closer to lower 4GHz than higher.

So for me 5GHz+ for dense matmul and similar would leave me impressed, seeing it has more cores.

For code compilation if it could hit close to 6GHz all core would also be lovely.
 
MS_AT said:
Normal or BBP mode? Normal bottlenecks faster on MBW, while BBP in Mystical's teardown article hits current limit and dives under 4GHz.
Click to expand...
Normal.
MS_AT said:
Generally I have seen dense matmul of right sizes go closer to lower 4GHz than higher.
Click to expand...
Tbf dense matmul forces clockdowns on pretty much everything and everyone. Including DC GPUs.
MS_AT said:
So for me 5GHz+ for dense matmul and similar would leave me impressed, seeing it has more cores.

For code compilation if it could hit close to 6GHz all core would also be lovely.
Click to expand...
Yeah. Two nodes are two nodes.
Plus it looks like a faster core inherently (at least from what's been published so far).
 
adroc_thurston said:
Yeah.
Apple, QCOM and (duh) ARM itself.
Click to expand...

Beating Apple and Qualcomm might make the fanboys cheer but it doesn't do anything for AMD's business since they don't compete with them. OK technically they kinda/sorta compete with Qualcomm, but Qualcomm has to show they can manage more than a low single digit percentage of Windows market share before AMD has any reason to give a crap about them.

AMD's real competition is Nvidia and Intel. That's where the money is, and the better they compare against Nvidia or increase their lead over Intel the happier their stockholders will be.
 
Doug S said:
Beating Apple and Qualcomm might make the fanboys cheer but it doesn't do anything for AMD's business since they don't compete with them. OK technically they kinda/sorta compete with Qualcomm, but Qualcomm has to show they can manage more than a low single digit percentage of Windows market share before AMD has any reason to give a crap about them.
Click to expand...
They very much all compete for the same PC TAM.
Doug S said:
AMD's real competition is Nvidia and Intel. That's where the money is, and the better they compare against Nvidia or increase their lead over Intel the happier their stockholders will be.
Click to expand...
Intel's a non-factor. NV doesn't play in CPU markets, really.
 
adroc_thurston said:
They very much all compete for the same PC TAM.
Click to expand...

As I've said many times before, people make the Mac vs PC decision first, before they look at anything else. Few people are cross shopping Mac vs Windows so the impact of AMD catching up to or even beating Apple's performance would be extremely limited.

Plus the segment where Apple is most likely to take a bigger share of the PC market is one where Apple has never played before with the $599 Neo. Zen 6 won't be competing there, certainly not now with memory/storage prices inflated - the low end PC market is going to mostly disappear for a while.
 
Doug S said:
As I've said many times before, people make the Mac vs PC decision first, before they look at anything else. Few people are cross shopping Mac vs Windows so the impact of AMD catching up to or even beating Apple's performance would be extremely limited.
Click to expand...
No, Apple is very much eating into PC share now irregarding of mac vs PC or whatever.
Why do you think the $600 Macbook exists?
Doug S said:
Plus the segment where Apple is most likely to take a bigger share of the PC market is one where Apple has never played before with the $599 Neo
Click to expand...
They're eating into it across every segment.
PCs need the perf. Thus, AMD has to deliver.
 
adroc_thurston said:
No, Apple is very much eating into PC share now irregarding of mac vs PC or whatever.
Why do you think the $600 Macbook exists?

They're eating into it across every segment.
PCs need the perf. Thus, AMD has to deliver.
Click to expand...

Apple's market share has increased quite modestly since they introduced Apple Silicon Macs, if you average it out over the whole time. You see sensationalist articles talking about big gains Apple is making that are based on quarterly results that hit the quarter after Apple introduces new mass market Macs like MBA. I'm sure we'll see the same based on Q2 and Q3 numbers for the Neo, we'll see people claiming that Apple is on a trajectory to displace Microsoft lol

But if you look at the numbers a few quarters after they've made all the splashy introductions the numbers are down, which is why I say you need to look at numbers averaged over time, ideally over the entire time since M1 Macs were first released.

So I don't buy that AMD considers Apple a competitor. I can promise you Apple does not consider AMD or Intel their competition.
 
Dragonetti031 said:
First off all I'm not spamming chatGPT garbage here , but information found in various articles from:

tomshardware.com - AMD EPYC Venice boasts 256 cores and bandwidth galore — next-gen server CPUs arrive in 2026
guru3d.com - AMD Zen 6 Architecture Surfaces With 8-Wide Dispatch and 512-Bit Vector Support.
wccftech.com - AMD to Pivot from SERDES to a “Sea-of-Wires” D2D Interconnect in Next-Gen Zen 6 CPUs, Bringing Major Power-Efficiency and Latency Gains
overclock3d.net - AMD Zen 6 to get “sea of wires” interconnect upgrade?
techpowerup.com - AMD D2D Interconnect in "Zen 6" Gets Sea-of-Wires Upgrade.
hothardware.com - AMD Reveals New Zen 6 Details In First Official Document.
hothardware.com - AMD Zen 6 CPUs Look Poised For A Major Interconnect Upgrade
Click to expand...
You need to understand that all the websites listed above are run by people with little to no clue, who mostly just parrot back what either MLID said in one of his videos or some leaker linked on xitter.
Actual tech journalism is pretty much dead, occasional valueable articles on chips&cheese being the exception.
 
reaperrr3 said:
You need to understand that all the websites listed above are run by people with little to no clue, who mostly just parrot back what either MLID said in one of his videos or some leaker linked on xitter.
Actual tech journalism is pretty much dead, occasional valueable articles on chips&cheese being the exception.
Click to expand...
Me sometimes being in RGT videos brings me physical pain.
 
OneEng2 said:
Prior to the current insanity, most people were guessing small IPC increases for Zen 6 over Zen 5 (10-15%). I'm still guessing this is correct since the number of transistors dedicated to raising IPC per core isn't going up that much (the additional transistors are MOSTLY going into more cores per CCD).
Click to expand...

We have some idea about die sizes, but not about transistor count and densities. Transistor density is one area where AMD excels.
 
Jan Olšan said:
AMD usually has a different rated (printed in specs) boost clock and actual maximum boost clock the CPU can use (for single thread tasks/on preferred cores of course). In chips like 5950X, 7950X and 9950X, there's 150 MHz "extra clock" the chip may achieve. But it tends to depend on load, temperature and other factors, which is probably why they officially give a lower rating.

When I used a 5950X, I was getting above 5000 MHz at least according to the maximum clocks logged in HWiNFO.

Getting the extra 150 MHz has (IIRC) gotten harder (need to get the temperature almost impossibly low, like 50C, or something?) on Zen 4 and Zen 5, but technically they can also go to 5.85 GHz in stock configuration. It's always a question how to factor the viability of that extra clock range into 1T comparisons and IPC guessing, exactly.
Click to expand...
They gave it a lower ranking because for zen 2 they put the actual maximum boost and because some (many) customers didn't reach it they got slated and so just advertised less than the max from then on.
 
inquiss said:
They gave it a lower ranking because for zen 2 they put the actual maximum boost and because some (many) customers didn't reach it they got slated and so just advertised less than the max from then on.
Click to expand...
It actually started with Ryzen 7 2700X (officially 4.3, really 4.35 GHz boost) and before that, Ryzen 1000 SKUs had something similar, except it was openly marketed and called "XFR".
(And I think Ryzen 9 3950X actually also had a 50 MHz extra range, with a 4750 MHz actual maximum).
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top