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

[The Stilt]

Socket - Platform name - Segment - Die - Package type

AM4 PGA - Summit Ridge - Desktop - Zeppelin die - SCM
SP3 LGA - Naples - Server - Zeppelin die - MCM
SP4 BGA - Snowy Owl - Server - Zeppelin die - MCM

Exactly the same as with Orochi die (Bulldozer / Piledriver). The same die was used in AM3+, C32 LGA and G34 LGA.

 

[Atari2600]

Not correct. Die design does matter in socket choice

Not really. The example you've given is completely non applicable.

Could a high-end AM4 motherboard support MCM package? Maybe - depends on how many memory channels are required.

Would a consumer/prosumer require 4 channel memory? Maybe. But having a 16-core, 32 thread platform with 2 channel memory may be a useful differentiation from the Opteron market.

 

[Nothingness]

Not when bound to process.

I worked the figuee out myself using the instr totals.

Didn't go off percentage from the tool.

So you claim this stands?

When I profiled the test on Bobcat and Jaguar, every cache access at every level was a miss with 100% DRAM hits. Probably due to the cache size/associativity

Branch Rate 3.93%
Mispredict Rate 1.01%
IC Cache Miss 100%
DC Accesses 100%
DC Misses 100%
L2 Misses 100%
Cycles FPU Empty 100%
Retired x87 FP Ops 0%
Retired SSE Ops 100%
DRAM Accesses 100%

These cache misses make absolutely no sense at all. One can't have 100% instruction cache misses. Any loop would make sure you're not at 100% I$ miss. Are you claiming the program has no loop?

 

[Sweepr]

Summit Ridge 8C/16T Cinebench R15 MT & Fritz Chess Scores Leaked (WCCFTech)

Comparison #1:

Core i7-7700K (Stock) 971* cb
Core i7-7700K (OC) 1108* cb
Core i7-6900K (Stock) 1565 cb
Core i7-6950X (Stock) 1863 cb
AMD RYZEN (8/16) 1188 cb

*http://www.bit-tech.net/hardware/cpus/2016/12/05/core-i7-7700k-performance-overclock-preview/3

Comparison #2:

Comparison #1:

Core i7-7700K (Stock)
35.52
Core i7-7700K (OC)
41.44
Xeon E5-2670 (Stock)
41.88
Core i7-6900K (Stock)
47.80
Core i7-6950X (Stock)
51.50
AMD RYZEN (8/16)
36.86

Core i7-6900K 30% faster @ FritzChess, 31% faster @ Cinebench R15.

http://wccftech.com/amd-ryzen-zen-cpu-benchmarks-leak/

 

[Arachnotronic]

Summit Ridge 8C/16T Cinebench R15 MT & FritzChess Scores Leaked

Core i7-7700K (Stock)
35.52
Core i7-7700K (5GHz)
41.44
Core i7-6900K (Stock)
47.80
Core i7-6950X (Stock)
51.50
AMD RYZEN (8/16)
36.86



Core i7-6900K still comfortably faster in both tests.

nice find, Sweepr.

 

[Charlie22911]

At what clocks? I'm finding it hard to accept Zen is scoring that low in comparison...

 

[Atari2600]

Core i7-6900K 30% faster @ FritzChess, 31% faster @ Cinebench R15.

Hmm, yes. More in line with (reasonable) expectations.

Big improvement on the current offerings, but cannot fully bridge the gap.

 

[Zstream]

Post link?

 

[PhonakV30]

Still too Vague.Name , Clock , Source ? how do you know It's Ryzen?

It was from Wccftech but they don't provide Link for "Chinese forums", also bottom "Your System......

 

[Sweepr]

Trying to find the original source, WCCFTech says it comes from chinese leakers but they don't provide a link (shameful): http://wccftech.com/amd-ryzen-zen-cpu-benchmarks-leak/

 

[StrangerGuy]

I'm still scouring for the Chinese source to provide a translation but yikes, I'm getting Bulldozer vibes again which isn't good.

 

[MajinCry]

Isn't Cinebench biased against AMD? I remember there being a big stink about it, and Agner being rather miffed. http://www.agner.org/optimize/blog/read.php?i=49#49. Sure, Intel can't require benchmarks to have a biased cpuid check, but that doesn't stop 'em pushing for it.

We need open source benchmarks.

 

[Glo.]

It is completely contradicting with what AMD provided.

So we have two possibilities:
AMD is lying, about Zen performance.
Anonymous leaker is lying about Zen performance.

Im guessing, we have to wait and see. I do not believe for a second in this leak. Biggest gripe here is there is no specifics. No core clock, anything.

 

[.vodka]

Right. No specifics. In the fritz chess screenshot, there's CPU-z running. Was it so difficult for the leaker not to crop the picture?

I'm waiting for a more complete leak with actual specs to complete the picture.

It is completely contradicting with what AMD provided.

So we have two possibilities:
AMD is lying, about Zen performance.
Anonymous leaker is lying about Zen performance.

Im guessing, we have to wait and see. I do not believe for a second in this leak. Biggest gripe here is there is no specifics. No core clock, anything.

Yes, it's a contradiction to what we've seen. You don't get the performance shown at the demos with the numbers in these leaks. Or you do get these numbers with a sample running way lower than 3.4GHz. We don't know.

Until there's more information to go with this leak, it means nothing.

 

[Zstream]

It is completely contradicting with what AMD provided.

So we have two possibilities:
AMD is lying, about Zen performance.
Anonymous leaker is lying about Zen performance.

Im guessing, we have to wait and see. I do not believe for a second in this leak. Biggest gripe here is there is no specifics. No core clock, anything.

How? In many cases equals broadwell and others haswell. I don't see that as lying or a fail. However I've grown out of the fanboyism that's still held by many.

 

[majord]

Fritz score screams of ES clocks, unless a Zen Core (w/smt) is slower than piledriver module clk/clk

 

[SpaceBeer]

It's "strange" leak, but it's not entirely impossible. I mean, it would mean in best case Ryzen is little bit faster than i7-6900K (Blender), and in worst case is little bit faster than i7-5930K (CB). We have had it before - some programs can utilize one hardware better than the other (7-zip vs WinRAR, Blender vs 3DS Max, Excel vs Calc...). Sometimes you need high clocks, sometimes more cores or cache, sometimes high RAM bandwidth... There is no one CPU to rule them all

 

[Sweepr]

If these scores are true then we know why Intel is reportedly planning to charge the same US$400-1800 for Skylake-X / Kaby Lake-X next year.

0-20% faster than Core i7-7700K in MT applications means it's a solid product, but Intel probably still dominates per core (ST) performance and would holds its own in many scenarios.

At similar clocks a Core i7-6800K basically matches Ryzen's CB R15 score and beats it in Fritz Chess with less cores (6C/12T vs 8C/16T). And Skylake-X will no doubt outperform Broadwell-E thanks to newer cores and 1MB L2 per core.

 

[inf64]

I think there is no way those scores are done at 3.4Ghz. If I had to take a guess I'd say ~3Ghz base clock.

 

[Glo.]

Fritz score screams of ES clocks, unless a Zen Core (w/smt) is slower than piledriver module clk/clk

It says there: P3 1.0 GHz "something" 480.

 

[KTE]

Or the temps are high so its lowering the MHz.

Its equally possible that ST is no where near Skylake, as I've questioned from the beginning.

Sent from HTC 10
(Opinions are own)

 

[Glo.]

Or the temps are high so its lowering the MHz.

Its equally possible that ST is no where near Skylake, as I've questioned from the beginning.

Sent from HTC 10
(Opinions are own)

Clock for clock, Zen should be on Haswell/Broadwell Level.

 

[Arachnotronic]

I think there is no way those scores are done at 3.4Ghz. If I had to take a guess I'd say ~3Ghz base clock.

Why not?

 

[PhonakV30]

Or maybe Poor MT Implementation due to New SMT in C15? Perhaps they need to update Cin15.

 

[blublub]

Are CB15 and Blender so different in execute that results could vary so much?