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

[USER8000]

And those ES did not even have the auto OC feature. Before new horizon event, i would not even imagine such a feature. They hid it very well and probabily is absent in most of the ES (in those given to journals for sure), so also in this. If this feature manage to give +200:+300MHz average, other than the promised base frequency, we can have even 4GHz mean frequency in most workloads...

I had forgotten about that!

 

[lifeblood]

So when the thread title change to "Summit Ridge (Zen) Benchmarks CHECK PAGE 153" ?

If I had to wallow through all 153 pages of this thread to reach the benchmarks than, dammit, everyone else should too!

The best part of Ryzen is that this isn't only about Ryzen, it should have some very interesting effects on motherboards as well. The auto-overclock feature should stimulate some imaginative motherboard specific cooling solutions. Many people are like me and don't really mess with overclocking, but if the CPU does it automatically then it might encourage more people to invest in water cooling or other non-mainstream cooling systems.

 

[blublub]

I wonder how those French journalists? / Reviewer's did get their hands on an ES incl. MB.

So just a question here:
Could that be a really really clever fake?

 

[raghu78]

Just to keep things in perspective. This is not the final chip. We do not know final base clocks, all core turbo, max turbo and most importantly max OC speeds of Ryzen. btw the max turbo is 3.4 Ghz

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http://www.cpchardware.com/cpc-hardware-n31-debarque-kiosque/&edit-text=

CPC Hardware @CPCHardware

Important precision on the Zen benchs: although the Turbo max was supposed to be 3.5 GHz, our sample is never mounted at + 3.4 GHz.

The all core turbo is probably even lower at 3.2-3.3 Ghz. 6900k runs at 3.5 Ghz in FP workloads like Blender which are more power hungry than INT workloads. In most games the 6900k is running at 3.7 Ghz as very few games scale to 16 threads (maybe BF1) and that gives it a 8.8% clock speed advantage at minimum. The 6700k is running with a turbo clock speed advantage of 23.5%. 4200/3400 = 1.2352 . We still need to wait for final product to come to meaningful conclusion. One thing seems to be clear - this chip is definitely Broadwell class for IPC.

 

[jpiniero]

I love how some are spinning the results whilst ignoring the Core i7 6900K which is running quad channel memory and appears to have higher clockspeeds,on motherboards with production BIOSes.

Quad Channel doesn't really do much for the 6 and 8 core models. It's definitely a waste. I'm assuming their Ryzen ES would be full core turboing to 3.4 (as opposed to the 6900K which is using the not-overclocked full core turbo of 3.5)

Clearly AMD is having clock speed issues so you really shouldn't assume that it'd be able to do more than 3.5, overclocking or not. It feels too eerily like Polaris where the design is very much competitive but undermined by GloFo yet again.

I wonder how those French journalists? / Reviewer's did get their hands on an ES incl. MB.

So just a question here:
Could that be a really really clever fake?

It could be.

 

[USER8000]

I wonder how those French journalists? / Reviewer's did get their hands on an ES incl. MB.

So just a question here:
Could that be a really really clever fake?

Its a well known French tech magazine so unlikely.

Quad Channel doesn't really do much for the 6 and 8 core models. It's definitely a waste. I'm assuming their Ryzen ES would be full core turboing to 3.4 (as opposed to the 6900K which is using the not-overclocked full core turbo of 3.5)

Clearly AMD is having clock speed issues so you really shouldn't assume that it'd be able to do more than 3.5, overclocking or not. It feels too eerily like Polaris where the design is very much competitive but undermined by GloFo yet again.



It could be.

But it does do something and people can try and look at it anyway they want - this is an engineering sample against a mature production CPU with motherboards sporting stable BIOSes. Ryzen is not out for a few months anyway and even AMD said Turbo was not switched on during their demo. With Turbo I expect the 8C/16T Ryzen will be very competitive with the Core i7 6900K.

If anything AMD has been deceitful on purpose about the performance of Ryzen. They said a 40% IPC uplift - this seems far more than that.

 

[PhonakV30]

Quad Channel doesn't really do much for the 6 and 8 core models. It's definitely a waste. I'm assuming their Ryzen ES would be full core turboing to 3.4 (as opposed to the 6900K which is using the not-overclocked full core turbo of 3.5)

Clearly AMD is having clock speed issues so you really shouldn't assume that it'd be able to do more than 3.5, overclocking or not. It feels too eerily like Polaris where the design is very much competitive but undermined by GloFo yet again.

Really? even Core i7-6900K is not good at OC.Forget about 6700K or 7700K.Or wait for 4C/8T then compare it .

 

[AtenRa]

Matching Skylake Core i5 6600 (3.3GHz base , 3.9GHz Turbo) in games with just 3.4GHz turbo is exceptional. Only thing left to know is how much it can OC, if the 4C 8T and 6C 12T can reach 4.4GHz it will be a huge hit.

 

[Sweepr]

Power consumption is in line with AMDs test roughly at i7-6900k level. IPC is way below Skylake I would say. Because i5-6500 and i7-6900k run at a similar clock speed and Broadwell is ~12% above Skylake in the gaming tests, which shows that a thread/core advantage makes a decent difference in the gaming tests.

I guess Broadwell has higher IPC than Skylake, given that 8C/16T Broadwell-E is 13.3% faster than 4C/4C Skylake-S in the games tested (3.2-3.7 GHz vs 3.2-3.6 GHz). Fair comparison.

 

[inf64]

No, Broadwell-E (6900K) has higher clockspeed and bigger cashes than Skylake-S in this test while being just ~3% behind Skylake IPC wise (as per AT review). Games they tested are clock/IPC taxing except BF1 which can tax more cores.
I can't wait for retail top SKU reviews and Francois Piednoel reactions over twitter

 

[Abwx]

I wonder how those French journalists? / Reviewer's did get their hands on an ES incl. MB.

Amazingly the guy that did the Zen test at CanardPC (and who once merged his site wich CanardPC) is the very one that got an Athlon 64 sample (at 1.4GHz) 8 months before the CPU was launched, so history seems to repeat itself..

http://www.x86-secret.com/popups/articleswindow.php?id=67

 

[bjt2]

If I had to wallow through all 153 pages of this thread to reach the benchmarks than, dammit, everyone else should too!

The best part of Ryzen is that this isn't only about Ryzen, it should have some very interesting effects on motherboards as well. The auto-overclock feature should stimulate some imaginative motherboard specific cooling solutions. Many people are like me and don't really mess with overclocking, but if the CPU does it automatically then it might encourage more people to invest in water cooling or other non-mainstream cooling systems.

I imagine also tuning shops: "pimp your rig!"

 

[LTC8K6]

Just to keep things in perspective. This is not the final chip. We do not know final base clocks, all core turbo, max turbo and most importantly max OC speeds of Ryzen. btw the max turbo is 3.4 Ghz

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http://www.cpchardware.com/cpc-hardware-n31-debarque-kiosque/&edit-text=

CPC Hardware @CPCHardware

Important precision on the Zen benchs: although the Turbo max was supposed to be 3.5 GHz, our sample is never mounted at + 3.4 GHz.

The all core turbo is probably even lower at 3.2-3.3 Ghz. 6900k runs at 3.5 Ghz in FP workloads like Blender which are more power hungry than INT workloads. In most games the 6900k is running at 3.7 Ghz as very few games scale to 16 threads (maybe BF1) and that gives it a 8.8% clock speed advantage at minimum. The 6700k is running with a turbo clock speed advantage of 23.5%. 4200/3400 = 1.2352 . We still need to wait for final product to come to meaningful conclusion. One thing seems to be clear - this chip is definitely Broadwell class for IPC.

Perhaps they needed a better cooler on Ryzen to get a higher turbo?

 

[inf64]

Perhaps they needed a better cooler on Ryzen to get a higher turbo?

Or it is just an ES and not a fully enabled retail SKU

 

[bjt2]

Clearly AMD is having clock speed issues so you really shouldn't assume that it'd be able to do more than 3.5, overclocking or not. It feels too eerily like Polaris where the design is very much competitive but undermined by GloFo yet again.

Don't spread FUD, please! You forgot the New Horizon event, in which a 3.4 base Zen was shown, with more than 3.4GHz BASE clock promised at launch...

 

[bjt2]

Or it is just an ES and not a fully enabled retail SKU

Noone imagined the auto OC feature, so i think that AMD hid it very well... How? I think that all ES around in these months had the feature disabled...

 

[sirmo]

Amazingly the guy that did the Zen test at CanardPC (and who once merged his site wich CanardPC) is the very one that got an Athlon 64 sample (at 1.4GHz) 8 months before the CPU was launched, so history seems to repeat itself..

http://www.x86-secret.com/popups/articleswindow.php?id=67

Wow, good find. I remember a lot of folks predicting Hammer's failure due to low clocks back then. Let's hope history repeats itself.

 

[jpiniero]

Don't spread FUD, please! You forgot the New Horizon event, in which a 3.4 base Zen was shown, with more than 3.4GHz BASE clock promised at launch...

They said it would run 3.4+, which I took to mean they will guarantee 3.4 at full turbo but not more. Unless Zen's not coming out any time soon, the New Horizons demo have to of been on basically a final chip and we can assume they could not get it to more than 3.4 esp with the rumors they upped the voltage.

 

[krumme]

They said it would run 3.4+, which I took to mean they will guarantee 3.4 at full turbo but not more. Unless Zen's not coming out any time soon, the New Horizons demo have to of been on basically a final chip and we can assume they could not get it to more than 3.4 esp with the rumors they upped the voltage.

No they said not. Watch again. It was base 3.4 or up plus turbo. What that remains we will see.

This is more or less bwe ipc as i can tell. Branchpredictor and prefetch cache wathever all seems fine or we wouldnt be looking at those results.

 

[inf64]

They said it would run 3.4+, which I took to mean they will guarantee 3.4 at full turbo but not more. Unless Zen's not coming out any time soon, the New Horizons demo have to of been on basically a final chip and we can assume they could not get it to more than 3.4 esp with the rumors they upped the voltage.

Actually AMD stated they will launch SKU with 3.4+Ghz BASE clock. Not all core Turbo etc. Base clock. Big difference there. It means they guarantee at least 3.4Ghz in FP/power hungry workloads within 95W TDP envelope. Any other less power hungry workload will go even higher via advanced Turbo/AutoOC features. Now strap a solid water cooler on such a chip and see what it does to clocks.

 

[jpiniero]

Actually AMD stated they will launch SKU with 3.4+Ghz BASE clock. Not all core Turbo etc. Base clock. Big difference there. It means they guarantee at least 3.4Ghz in FP/power hungry workloads within 95W TDP envelope. Any other less power hungry workload will go even higher via advanced Turbo/AutoOC features. Now strap a solid water cooler on such a chip and see what it does to clocks.

See, I think AMD is calling the base clock what most people would call the turbo clock and using the Auto OC to hide that fact. It's going to be interesting at least to see how different coolers affect the clocks.

 

[inf64]

See, I think AMD is calling the base clock what most people would call the turbo clock and using the Auto OC to hide that fact. It's going to be interesting at least to see how different coolers affect the clocks.

Why do you think that? Base clock means what is says: minimum guaranteed clock for a SKU. I think you are just confused.

 

[IntelUser2000]

6400 runs at 3.3 to 3.1Ghz and 6500 at 3.6 to 3.3Ghz, come on now, that's the less than 10% difference you see.

Core i5 6400 is 2.7-3.3 and 6500 is 3.2-3.6.
http://www.cpu-monkey.com/en/compare_cpu-intel_core_i5_6400-527-vs-intel_core_i5_6500-524

Base difference: 18.5%
ST Turbo difference: 9.1%
MT Turbo difference: 13.3%
Game difference: 4.1%

This means for every 100% increase in clock speed, we are getting anywhere from 30.8% to 45.1%. Put it another way, that's anywhere from less than a 1/3rd to less than 1/2. Quite typical of games though, because games don't go full CPU-limited until you get an overkill of a graphics card.

So if we take inf64's points and scale up Zen from 3.4 to 3.7 we go from 97.4 to 101.1, best case scenario. Worst case is 100. That's a clock-per-clock difference of 6-7% from BDW-E. If we assume 4GHz for Zen based on inf64's "vs 6700K" we get 105, "best case".

Of course, we cannot forget the effect of multi-threaded coding on gaming performance. 6600K vs 6700K clock differences.

Base difference: 14.2%
ST Turbo difference: 7.7%
MT Turbo difference: 11.1%
Game difference: 18.2%

That's super-linear scaling folks. That means based on the fact that same thread count i5s were showing well below 50% scaling per clock speed, Hyperthreading is benefitting the 6700K CPU. That means naturally 8 core CPUs are going to fare better than 4 core 8 thread ones.

 

[happy medium]

First AMD RYZEN Review Leaked – Aggregate Performance 46% Faster Than FX-8370 with 93 Watt Power Consumption

http://wccftech.com/amd-ryzen-review-leaked/

http://www.cpchardware.com/cpc-hardware-n31-debarque-kiosque/

 

[bjt2]

They said it would run 3.4+, which I took to mean they will guarantee 3.4 at full turbo but not more. Unless Zen's not coming out any time soon, the New Horizons demo have to of been on basically a final chip and we can assume they could not get it to more than 3.4 esp with the rumors they upped the voltage.

The new Horizon event is an official AMD event, in which was clearly stated 3.4GHz+ BASE clock. Turbo was not disclosed.

Saying that 3.4GHz is max turbo is FUD, because it's false. Maybe you have not watched thoroughly the video.