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

[superstition]

Early ES results shouldn't be taken as the final product.

Does anyone have overclocked FX to compare with this, for curiosity? 4.6 GHz or so?

 

[Sweepr]

And the game definitely takes advantage of more than 4C/8T:

 

[Bryf50]

What's next, are you going to say this non-K 6700 is overclocked to 5 GHz+ as well?

• Average: 105.6 FPS

• Normal: 118.2 FPS

• Medium Batch: 109.2 FPS

• Heavy Batch: 92.7 FPS

http://www.ashesofthesingularity.co...-details/9ea1e763-93f4-4954-be0a-7b6dc6682699

Please drop the excuses and don't go personal next time.

Version: 0.94.18701.0

Here's a middle of the pack 5960x that scores close to the Zen leak. 5960x = 3.2 ghz Zen confirmed.
http://www.ashesofthesingularity.co...-details/796b78ba-301c-46d1-886b-190cd55a187f

 

[Zstream]

Sweepr trying to misguide people? Say it ain't so..

 

[Sweepr]

Maybe you can settle this by running the same version on your 6700K at stock then.

Ps: Your scores are not from the same version of Zen's results either!

 

[Abwx]

http://wccftech.com/amd-zen-es-benchmarks/

 

[Arachnotronic]

The vitriol I see directed towards Sweepr here reminds me eerily of the hate that some people directed towards the skeptics back pre Bulldozer.

 

[Sweepr]

Even if we use the results posted for the 'stock' 6700 by the other user:

• Average: 68.4 FPS

• Normal: 75.0FPS

• Medium Batch: 69.8FPS

• Heavy Batch: 61.8 FPS
  
  We still have a non-K 4C/8T Skylake-S beating 8C/16T Summit Ridge in a benchmark that takes advantage of the extra cores.

http://wccftech.com/amd-zen-es-benchmarks/

Below Haswell i7 as well then.

 

[Abwx]

Even if we use the results posted for the 'stock' 6700 by the other user:

• Average: 68.4 FPS

• Normal: 75.0FPS

• Medium Batch: 69.8FPS

• Heavy Batch: 61.8 FPS
  
  We still have a non-K 4C/8T Skylake-S beating 8C/16T Summit Ridge in a benchmark that takes advantage of the extra cores.

Below Haswell i7 as well then.

The chip is running at 2.8 since Aots use several cores, as for your other user he s using a GTX1080, so much for your numbers, as "relevant" as the ones you posted in the OP, but is it a surprise..??.

 

[FlanK3r]

I will test it soon (maybe Friday) at i7-6950X (I can try few steps of clocks or disbaling cores) +RX 480. What GPU drivers the guy used?

 

[superstition]

Flank3r... Can you also also test an FX at 4.5, since that is a readily obtainable overclock without high-end boards or cooling (making low-end stock FX clocks rather irrelevant)? Or, are those lighter grey parts in your sig no longer owned by you?

 

[jpiniero]

but is a theorical limit computed by Aots.

I thought it was scoring it by taking the GPU out of the equation. Obviously the 480 is going to be faster than a 460.

Anyway, it seems to be hitting expectations - SB IPC but no real threat to Intel.

 

[superstition]

What if Zen ships with solder instead of bottlenecking paste? That could help with the overclocking, perhaps. We have yet another delidding topic that just popped up here even though people have tried to argue to me that delidding is an extremely fringe activity because there isn't really an incentive to do it.

 

[FlanK3r]

Flank3r... Can you also also test an FX at 4.5, since that is a readily obtainable overclock without high-end boards or cooling (making low-end stock FX clocks rather irrelevant)? Or, are those lighter grey parts in your sig no longer owned by you?

Working Ashes of SIngularity offline? My PC with FX is near my parents home, and there is not internet conection (ooops) This week Im in home and next week Im going to my parents for one week (FX-8370 computer there).

 

[frozentundra123456]

One can argue about the ashes benchmark till the cows come home. To me it is a totally unreliable tech demo masquerading as a "game", that is all over the place depending on the phase of the moon or whatever. I mean, it is supposed to love cores, but I believe even a i3 was close to 8 core FX in the initial benchmarks.

The significant thing is the clocks. If these are final clocks, at SB or slightly better IPC, then AMD is going to be back to the usual technique of competing on price instead of having a top end product. However, I would not be surprised if those are close to the final clocks, if they are really sticking to 95 watt TDP. Just look at Polaris on the same process. Power use and frequency are certainly not spectacular, and the efficiency gain is far below what was promised by AMD.

 

[dark zero]

It's funny that nobody here is smart enough to realise that the linked benchmark shows a 6700k which is overclocked to 5GHz or higher.

See the best result of the 6700 at 3.4GHz which is basically the exact same CPU except at a lower clock rate.

• Average: 68.4 FPS

• Normal: 75.0FPS

• Medium Batch: 69.8FPS

• Heavy Batch: 61.8 FPS

http://www.ashesofthesingularity.co...-details/e440453e-8f8d-4fca-a73f-2de853e6fd2e

compared to the heavily overclocked 6700k the OP posted

- Core i7-6700K 4C/8T - 4.0 GHz (2015 Skylake)
Average: 107.3 FPS
Normal batch: 125.7 FPS
Medium batch: 113.8 FPS
Heavy batch: 89.2 FPS

http://www.ashesofthesingularity.co...-details/6096dd00-7e11-4368-b588-97413922f5c7

Let's compare the average 68.4 FPS and 107.3 FPS. 107.3/68.4*100= 156%
So that means the overclocked 6700k is 56% faster than the stock 6700.
That's a massive difference. Far bigger than a simple overclock to 4GHz would suggest.
If we multiplicate the stock clock rate of the 6700 at 3.4GHz with 156% the result would be 3.4GHZ*156/100=5.3GHz.

and 5.3GHz can actually be easily achieved with the 6700k.

It's also interesting to note that the game's FPS don't scale with more cores very well. See the result of this 16 core Xeon so this is most likely just a comparison of the single core performance which again mostly depends on the clock speed.

http://www.ashesofthesingularity.co...-details/f37e31c4-299d-44d4-b5d2-40462a551e54

Really? 5.3 GHz on i7 6700k? And my SB i5 2500K can reach 6 GHz. On air!


Now going seriously, AMD definitely got the help of Samsung. Wondering if those are the full cores or were using only 4 of them. Or maybe it was not using their "hyper" threads.

Then. The clock rate is low... But if we scale down all the processors, definitely we see AMD getting finally back to the game (in the 15% of distance of Intel). Samsung helped AMD big time and has the stamp and the help from them.

Is like showing how strong is Samsung to Intel and they are not allowed to lose brutally. And this is only the beginning. Also Apple will love this a LOT since they would finally see someone who can change without losing performance.

AMD didn't managed to beat Intel in their own game, but at least is back to the fight.

 

[frozentundra123456]

What if Zen ships with solder instead of bottlenecking paste? That could help with the overclocking, perhaps. We have yet another delidding topic that just popped up here even though people have tried to argue to me that delidding is an extremely fringe activity because there isn't really an incentive to do it.

Really obsessed with this solder/paste thing aren't you? How much overclocking do you think skylake is sacrificing by using TIM? Most skylake chips can hit 4.6 or 4.7 relatively easily I believe. Even with the magic solder, I cant imaging getting 5.0 or above, so at most they are sacrificing a few hundred mhz. But in any case, Zen will compete with the HEDT lineup (or it had better, or it will be a complete fail), which *are* soldered.

 

[rainy]

The vitriol I see directed towards Sweepr here reminds me eerily of the hate that some people directed towards the skeptics back pre Bulldozer.

Please stop posting this nonsense - OP clearly don't like AMD and nonetheless he created a lot of threads about their products (both CPU/GPU) in the last a couple of months.
Quite surprisingly I don't recall a single one with positive outcome for AMD. /s

Btw, you're defending him just because he's your forum friend or/also because he wearing the same blue and green glasses?

 

[JoeRambo]

Heh, thread is turning into "shoot the messenger" one. Sad to discuss people when AOTS benchmark tells the story of disastrous deficit of performance in what should be SHOWCASE for 8 core / 16 thread chip.

 

[dark zero]

Really obsessed with this solder/paste thing aren't you? How much overclocking do you think skylake is sacrificing by using TIM? Most skylake chips can hit 4.6 or 4.7 relatively easily I believe. Even with the magic solder, I cant imaging getting 5.0 or above, so at most they are sacrificing a few hundred mhz. But in any case, Zen will compete with the HEDT lineup (or it had better, or it will be a complete fail), which *are* soldered.

SB chips reached 4.6 GHz in the worst scenario. Their 5.0 GHz were Average chips and the golden ones reach 5.2 on air and 5.5 GHz on water. Those are REAL monsters. Also, they were soldered.

 

[The Stilt]

Early ES results shouldn't be taken as the final product.

Does anyone have overclocked FX to compare with this, for curiosity? 4.6 GHz or so?

Zen was supposed to launch in Q4 2016 so this is definitely no "early ES" we are talking about here. AMD has no reason to hand out those "early ES" anywhere since the partners tested their AM4 boards with Bristol Ridge initially.

Based on the delta between all core and single core turbo frequencies, you can bet that there is basically nothing left in these chips clock wise. No different to Polaris in that aspect.

Zeppelin desktop and server SKUs are soldered, based on what I've heard.

 

[VirtualLarry]

you can bet that there is basically nothing left in these chips clock wise. No different to Polaris in that aspect.

Sad... but this is AMD we're talking about.

Then again, you don't expect your star batter to hit one out of the park on his first swing all the time, do you?

I would expect things to improve in time for Zen++ aka Zen APU.

 

[Arachnotronic]

Sad... but this is AMD we're talking about.

Then again, you don't expect your star batter to hit one out of the park on his first swing all the time, do you?

I would expect things to improve in time for Zen++ aka Zen APU.

Problem is, Intel isn't sitting still either.

 

[superstition]

Really obsessed with this solder/paste thing aren't you?

Obsessed? No. I haven't delidded any chips nor am I the one who:

a) created a delidding machine to sell

b) posted the vast majority of topics and posts about delidding

c) wrote professional articles about Intel's thermal bottleneck via TIM (including those claiming that the Devil's Canyon "improved TIM" wasn't improved at all).

d) did systematic testing for said articles

Not obsessed but mindful.

 

[The Stilt]

I would expect things to improve in time for Zen++ aka Zen APU.

Raven is not Zen+ based. It uses the same cores as Zeppelin does.