AMD Zen ES leaked benchmarks: performance similar to a Core i5 4670K?!?!

[themaniac]

On ‎8‎/‎12‎/‎2016 at 1:39 PM, cesrai said:

You're not serious right ? remember Intel cinebench scandal ? 

don't think I've ever heard it, also I have been corrected earlier that it is completely possible to optimize for one companies cpus

[S w a t s o n]

Zen needs to succeed for the good of the cpu market and if it "only" has the perf of a 4670k that is simply not good enough. This chip has been in develeopment for years, designed by the legendary Jim Keller. They really needed to exceed or match intel's highest end offerings with a product stack that goes down just like intels. If this much effort and design work can only match a 4670k we're pretty much fucked. In a gen or two AMD will be back in the dust. I know zen will have like a crazy ass 20 core version or something but that's to compete with xeon chips for server. They needed to have a 8/10 core beast ready for enthusiats.

[Kenjionigod]

21 hours ago, Morgan MLGman said:

Nah, I didn't mean anything about performance as I personally would like to wait for some more info that isn't 1. AOTS and 2. ES. This is also a misleading title imho, as people might think that an 8c/16t Zen CPU will be of the same performance as an 4670K, which is ridiculous, as FX-8cores outperform it multithreaded-wise already

 

What I meant is that there's so much talk in this thread about "TDP", as if it really mattered that much... People get Haswell-E/Broadwell-E i7s without even mentioning their TDP that's above FX-8 line "cause FX CPUs are hot as fuck and will make your PC a toaster", I've had an OC'd 125W FX-8350 and now I've got a 65W i7-3770S which has half 8350s TDP.

And guess what, there's no difference. What's funny, FX had lower temps under load.

This. I'm so tired of people saying that AMD runs hot. I have an overclocked 8300 and 52 degrees is the hottest it ever got under full load in the hottest core and it's air cooled. Plus, TDP. TDP doesn't mean that's the most power it'll use. That's not a single CPU or GPU that using whatever the TDP is, and it doesn't matter what company it is. AMD, Intel or Nvidia. Also considering the 8300 and others like the 8320e have a TDP of 95w on 32nm, I don't know why people are so ready to call out 95w on the new 22nm process as being impossible and unlikely so quickly.

[patrickjp93]

1 hour ago, themaniac said:

don't think I've ever heard it, also I have been corrected earlier that it is completely possible to optimize for one companies cpus

Maxon used Intel's custom compiler for the Cinebench code, but ICC only optimizes for Intel chips, and otherwise has an x86 chip run a non-optimized code path. It really wasn't Intel's fault, but it was slapped with an anticompetitive fine nonetheless. It was never proven Intel incentivized Maxon to use its compiler in any way, and the warning has been in the compiler manual from the start. Cinebench these days uses GCC.

[Prysin]

14 minutes ago, patrickjp93 said:

Maxon used Intel's custom compiler for the Cinebench code, but ICC only optimizes for Intel chips, and otherwise has an x86 chip run a non-optimized code path. It really wasn't Intel's fault, but it was slapped with an anticompetitive fine nonetheless. It was never proven Intel incentivized Maxon to use its compiler in any way, and the warning has been in the compiler manual from the start. Cinebench these days uses GCC.

actually, there was some lil turd who tested R15, and found the exact same code paths as those back in the day. Cannot remember where i saw the thread, but it was written about in this news forum about 6-9 months ago

 

[patrickjp93]

36 minutes ago, Prysin said:

actually, there was some lil turd who tested R15, and found the exact same code paths as those back in the day. Cannot remember where i saw the thread, but it was written about in this news forum about 6-9 months ago

 

The paths exist so that various CPUs can use what instruction sets are native to them. It's called an instruction dispatcher. If CPU A has AVX but CPU B only has SSE 4, then CPU B would crash when it tried to accept the binary code for AVX.

[Prysin]

Just now, patrickjp93 said:

The paths exist so that various CPUs can use what instruction sets are native to them. It's called an instruction dispatcher. If CPU A has AVX but CPU B only has SSE 4, then CPU B would crash when it tried to accept the binary code for AVX.

except all current gen AMD CPUs has AVX 1.... think even a few of them has AVX 2. So if that is the difference, then its again, because they just copy pasted intels compiler paths into their "new" compiler.

[patrickjp93]

Just now, Prysin said:

except all current gen AMD CPUs has AVX 1.... think even a few of them has AVX 2. So if that is the difference, then its again, because they just copy pasted intels compiler paths into their "new" compiler.

The path is not CPU_Name dependent, and are you then saying older PCs shouldn't be able to run this benchmark? It's simply a design choice and using a compiler flag like -march=x86_all, generating separate code paths for no MMX, MMX, SSE1/2/3/4, AVX1/2, and FMA3/4 (only AMD CPUs use FMA4).

[Prysin]

2 minutes ago, patrickjp93 said:

The path is not CPU_Name dependent, and are you then saying older PCs shouldn't be able to run this benchmark? It's simply a design choice and using a compiler flag like -march=x86_all, generating separate code paths for no MMX, MMX, SSE1/2/3/4, AVX1/2, and FMA3/4 (only AMD CPUs use FMA4).

my bad, was based on POV-Ray tests that was done in the wake of the 15 year settlement bonanza

http://www.extremetech.com/computing/193480-intel-finally-agrees-to-pay-15-to-pentium-4-owners-over-amd-athlon-benchmarking-shenanigans

[cesrai]

14 hours ago, themaniac said:

don't think I've ever heard it, also I have been corrected earlier that it is completely possible to optimize for one companies cpus

Looks I missed that post then, so don't mind me just passing by.

[ScratchCat]

AMD said that Zen has a 40% higher IPC compared to Excavator. The Athlon X4 845 is based on excavator (might not have L3 Cache , idk) reviewed here :http://www.pcgameshardware.de/Athlon-X4-845-CPU-261962/Tests/Excavator-Benchmarks-Test-1191570/

It scores 321 in cinebench r15 at 3.8ghz and 1.4x321=449 which is lower than a haswell 3.6ghz 4690k.

Not looking good for AMD

[RxL]

Regardless of all the techie info about Zen all i got from this thread was Zen = AMD took close to 5 years or took 5 years (WHEN ZEN IS FINALLY LAUNCHED) that it is still 3 years behind and barely edges past an i5 with 4 core 4 threads 22nm process while zen is an 8 core 16 thread on a 14nm process regardless of the core speed of the both CPU's i don't think it bode's well for AMD imo i dont it looks good for AMD atm but thats just my opinion

[vorticalbox]

On ‎13‎/‎08‎/‎2016 at 9:41 PM, Kenjionigod said:

This. I'm so tired of people saying that AMD runs hot. I have an overclocked 8300 and 52 degrees is the hottest it ever got under full load in the hottest core and it's air cooled. Plus, TDP. TDP doesn't mean that's the most power it'll use. That's not a single CPU or GPU that using whatever the TDP is, and it doesn't matter what company it is. AMD, Intel or Nvidia. Also considering the 8300 and others like the 8320e have a TDP of 95w on 32nm, I don't know why people are so ready to call out 95w on the new 22nm process as being impossible and unlikely so quickly.

they are only 95w because they unclicked the base from 4 to 3.2 GHz. My 8350 heats up to around 50c at stock. I'm using h100i water temps you know with amd having no temp sensor.

[79wjd]

1 hour ago, ScratchCat said:

AMD said that Zen has a 40% higher IPC compared to Excavator. The Athlon X4 845 is based on excavator (might not have L3 Cache , idk) reviewed here :http://www.pcgameshardware.de/Athlon-X4-845-CPU-261962/Tests/Excavator-Benchmarks-Test-1191570/

It scores 321 in cinebench r15 at 3.8ghz and 1.4x321=449 which is lower than a haswell 3.6ghz 4690k.

Not looking good for AMD

An 845 also isn't a true quad core, it has the same 2 modules 4 core layout as an FX4 cpu (where there are shared resources). 

[The Benjamins]

2 hours ago, ScratchCat said:

AMD said that Zen has a 40% higher IPC compared to Excavator. The Athlon X4 845 is based on excavator (might not have L3 Cache , idk) reviewed here :http://www.pcgameshardware.de/Athlon-X4-845-CPU-261962/Tests/Excavator-Benchmarks-Test-1191570/

It scores 321 in cinebench r15 at 3.8ghz and 1.4x321=449 which is lower than a haswell 3.6ghz 4690k.

Not looking good for AMD

but the consumer platform will have up to 8 cores with hyper threading when Intel only sells up to 4 cores on its consumer line.assuming the 40% is accurate a 8/16 model should easily get a score of 1200+.

[patrickjp93]

3 hours ago, The Benjamins said:

but the consumer platform will have up to 8 cores with hyper threading when Intel only sells up to 4 cores on its consumer line.assuming the 40% is accurate a 8/16 model should easily get a score of 1200+.

No, that's still the enthusiast platform. AMD said Raven Ridge is mainstream and will have both APUs and plain quad cores.

[KB_TheDireWolf]

Is this good or bad? Not sure...

[patrickjp93]

On 8/13/2016 at 4:41 PM, Kenjionigod said:

This. I'm so tired of people saying that AMD runs hot. I have an overclocked 8300 and 52 degrees is the hottest it ever got under full load in the hottest core and it's air cooled. Plus, TDP. TDP doesn't mean that's the most power it'll use. That's not a single CPU or GPU that using whatever the TDP is, and it doesn't matter what company it is. AMD, Intel or Nvidia. Also considering the 8300 and others like the 8320e have a TDP of 95w on 32nm, I don't know why people are so ready to call out 95w on the new 22nm process as being impossible and unlikely so quickly.

You do know that's the reading of the package which isn't actually on the chip, right? AMD runs hot and hungry, it just survives hotter temps.

 

As for calling out the 95W TDP on Zen, it's because getting Intel's performance is not going to come cheap thermally or electrically, no matter how good Jim Keller actually is.

[Nena Trinity]

Who cares? It is a engineering sample, how can we be sure its clock speed is not reduced by 50% of the real product? We have yet to see the real things specs, which means nothing is guaranteed this can even be the budget Zen for all we know...  

[themaniac]

yeah sure an 8c 16t budget cpu

[patrickjp93]

15 minutes ago, Nena360 said:

Who cares? It is a engineering sample, how can we be sure its clock speed is not reduced by 50% of the real product? We have yet to see the real things specs, which means nothing is guaranteed this can even be the budget Zen for all we know...  

Because the promised TDP of that product is 95W, and getting close to Intel's performance is not going to come free in terms of thermals or electrical usage. It's not like Jim Keller is a god and Intel's been sitting on its ass.

 

This isn't budget Zen. That will be Raven Ridge in mid to late 2017. This is the enthusiast lineup.

[Tam3n]

Can't really guess too much with this. We need a real cpu benchmark result like Cinebench. I think it would still be kind of sad if Zen 8 cores 16 threads is just as fast as Haswell 4 cores 4 threads...

[MrDynamicMan]

On August 11, 2016 at 0:08 AM, CtW said:

So a chip that has yet to be released is as fast as a mid tier chip from 3 years ago? Wow that's so impressive!

Seeing as performance has barely improved... Yeah, that is pretty impressive.

[Nena Trinity]

9 hours ago, patrickjp93 said:

Because the promised TDP of that product is 95W, and getting close to Intel's performance is not going to come free in terms of thermals or electrical usage. It's not like Jim Keller is a god and Intel's been sitting on its ass.

 

This isn't budget Zen. That will be Raven Ridge in mid to late 2017. This is the enthusiast lineup.

Well I hope they make a 125w Zen then... V_V

[patrickjp93]

5 hours ago, MrDynamicMan said:

Seeing as performance has barely improved... Yeah, that is pretty impressive.

It's not Intel's fault consumer programmers won't push the envelope and use newer instructions that do more. Performance capability has more than doubled since Nehalem. You can't blame the hardware provider for the idiocy of software developers.