• 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.

AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

Page 255 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.
AMD has to make on 14 nm Process: Enough Ryzen CPUs, Enough Polaris GPUs, Enough Vega GPUs, Enough Raven Ridge APUs.

And GloFo has ONLY 60K wafer/month Fab, and will it increase in upcoming months to 72K wafers/month. This is not enough.
 
That doesn't mean that Ryzen will be made there. Could be Raven Ridge.. could just be overflow capacity in the hopes that AMD will be able to sale as many as they can make...

Either way it frees up production, so if they offload Polaris and possibly Vega to Samsung, more wafers for Ryzen at Global Foundries.

Don't forget AMD is rumored to have two 8 core dies taped out (ISSCC), one might be for Naples so they have those to pump out on 14LPP also.
 
Either way it frees up production, so if they offload Polaris and possibly Vega to Samsung, more wafers for Ryzen at Global Foundries.

If Samsung's process is better than GloFo's (even marginally so), then I don't think it would make much sense to produce low-margin Polaris products at Samsung and high-margin Ryzen parts at GloFo. The other way around would be more logical.
 
AMD has to make on 14 nm Process: Enough Ryzen CPUs, Enough Polaris GPUs, Enough Vega GPUs, Enough Raven Ridge APUs.

And GloFo has ONLY 60K wafer/month Fab, and will it increase in upcoming months to 72K wafers/month. This is not enough.

Maybe do the math , they don't need 50 million units per quarter
Not that GloFo has 60k wafers 14nm capacity to begin with, 60k is the entire fab but not all is 14nm
AMD can likely do with 10k wafer starts per month for initial Summit Ridge supply, then shift some output for Naples but that's very little at first and adjust ahead of Ravern Rydge.Ofc that's if Summit Ridge sells very well and that output would be as high only to keep up with the initial demand. Add GPU and they need 15-20k wspm at best this year, assuming APUs don't ship as early as July.
Consoles not factored in.
 
Last edited:
I dont like the OP being butthurt and not adding more results as memes being spammed in this thread. Can we have a thread for benchmarks started from someone willing to post results, even if they dont fit his/hes/their agenda?

https://socialtechwork.com/threads/amd-zen-key-dates-and-information.2495226/

Someone started one a while ago because of the blatant misinformation that was being spread in this thread at that point. I don't think he's updating it anymore, last update was January 11.
 
Fixed for you. You don't actually need ST performance, as there is no current ST game 🙂

All dx11 games have a main thread which will be the bottleneck. So a low clocked multi-core will choke even if the game can run 100 threads. That is the fundamental issue with dx11. And patching on dx12 on a dx11 and older based engine can only help so much.

Plus I have a 144 hz monitor and want to at least run at 120 FPS after an upgrade. So CPU performance becomes very important.
 
All dx11 games have a main thread which will be the bottleneck. So a low clocked multi-core will choke even if the game can run 100 threads. That is the fundamental issue with dx11. And patching on dx12 on a dx11 and older based engine can only help so much.

Plus I have a 144 hz monitor and want to at least run at 120 FPS after an upgrade. So CPU performance becomes very important.

Uhh, what about my Overwatch game running at like 60-70% across all 8 logical cores in DX11?
 
All dx11 games have a main thread which will be the bottleneck. So a low clocked multi-core will choke even if the game can run 100 threads. That is the fundamental issue with dx11. And patching on dx12 on a dx11 and older based engine can only help so much.

Plus I have a 144 hz monitor and want to at least run at 120 FPS after an upgrade. So CPU performance becomes very important.

You are neglecting to consider multi-threaded game logic, sound threads, system servicing threads, graphics driver threads, background tasks, and a great deal more that has gone into making better use of more cores.

Test after test shows that modern games (DX11+) making increasingly better use of more threads. An i7 can be 30%+ better for gaming... a six core can add some more on top. Eight cores is simply not helpful at this point... but they will be before not too long.
 
It doesn't change the fact that in DX11 the main GPU thread is not multithreaded.

Even Bulldozer when OC to 5ghz eliminates almost all SC IPC deficits. Ryzen at 3.4 will do fine. Ryzen at 4ghz will be futureproof.
 
It doesn't change the fact that in DX11 the main GPU thread is not multithreaded.

Even Bulldozer when OC to 5ghz eliminates almost all SC IPC deficits. Ryzen at 3.4 will do fine. Ryzen at 4ghz will be futureproof.

The main thread just creates a graphics queue in the driver for many games, with game logic occurring in other threads. That graphics queue can be serviced by multiple threads in the driver, though it needs to be reordered for packing to the GPU. This is why we are actually seeing DX11 games use more and more threads effectively.

Then you have networking threads, sound threads, etc... on even the most serialized games.
 
Uhh, what about my Overwatch game running at like 60-70% across all 8 logical cores in DX11?

You can be CPU limited at 12.5% usage. (eg 1 logical core fully used). That is the issue with the main thread. Mabye not true for overwatch but just pointing out that windows task manager CPU usage is pretty much useless for the issue at hand.
 
How do you guys believe these CPUs will be reviewed: with XFR or without it?

I bet a lot of big names will explicit disable this feature saying its an overclock and compare with Intel's only boost vs boost.
I really doubt they'd get away with it. lol

Although I wouldn't be too surprised if someone O/C'd the Intel chips to "make it fair".
 
C4tXkorUMAA7e70.jpg

Funniest for me is that I am a buddhist 😉.

And no, they did not stolen the logo, because that would mean that monkeys that can draw a circle on the paper could steal the logo.
This is the 21st century though and people have been sued for less.
 
All dx11 games have a main thread which will be the bottleneck. So a low clocked multi-core will choke even if the game can run 100 threads. That is the fundamental issue with dx11. And patching on dx12 on a dx11 and older based engine can only help so much.

Plus I have a 144 hz monitor and want to at least run at 120 FPS after an upgrade. So CPU performance becomes very important.

If true, this still doesn't invalidate my point. You need to watch for review and look to gaming performance. That would be your answer. Looking at a syntetic benchmarks ro check ST performance won't give you a reliable one, as games are not ST.
 
You can be CPU limited at 12.5% usage. (eg 1 logical core fully used). That is the issue with the main thread. Mabye not true for overwatch but just pointing out that windows task manager CPU usage is pretty much useless for the issue at hand.

You do realize that you can view the usage of each individual core (actually hardware thread) in task manager, right?
 
Status
Not open for further replies.
Back
Top