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

[imported_jjj]

1080 is used by some who want to be able to get a high enough FPS to take advantage of high refresh rates. There is also the issue of not masking the performance of CPUs by having the GPU be the main bottleneck.

However, it is true that it's much more ideal to have higher resolution results to look at as well, for comparison. Having a ton of extra FPS at a lower resolution than you're going to play the game at is fairly meaningless unless it helps in minimums and percentiles (smoothness) in the higher resolution. In that case, though, the higher resolution testing should show that anyway.

The issue is that they mislead, many people think it's better with high clocks and it's pretty clear that many do think that.
At a reasonable config (as in well balanced system) and res, with latest gen CPUs,you are not gonna be clock bound but you might see upsides in some games that can use more cores or like large L3$. The future is likely to favor more cores too.
Ofc there are cases when one could build a budget constrained system with an unlocked Pentium and a GTX 1070 and aim for 4k lower settings or 1440p and then you are CPU bound and must look at how well you can get away with it.

 

[imported_jjj]

From some Broadwell-E reviews

Anandtech w/ GTX 980 and R9 290X ,mostly at 1080p http://www.anandtech.com/show/10337...6900k-6850k-and-6800k-tested-up-to-10-cores/8
Minimal differences both ways in Alien: Isolation , Total War: Attila ,Grand Theft Auto V , Middle-Earth: Shadow of Mordor
In GRID: Autosport small differences . With GTX 980 more cores do better ,with R9290X odd results http://images.anandtech.com/graphs/graph10337/81867.png

Hardware Canucks at 1080p with GTX 980 http://www.hardwarecanucks.com/foru...-broadwell-e-i7-6950x-i7-6900k-review-11.html
Many cores does better in Doom, Dying Light , GTA 5
Minimal differences in Middle Earth , The Witcher 3

Guru3D testing done at 1440p with GTX 1080 ,discarding tests at 1080p http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,16.html
Minimal differences in Tomb Rider , Hitman , Tom Clancy's The Division
Far Cry Primal likes many cores by a small margin

Tom's Hardware with GTX 1080 at 1440p http://www.tomshardware.com/reviews/intel-core-i7-broadwell-e-6950x-6900k-6850k-6800k,4587-3.html
Ashes of Singularity and F1 2015 favor many cores
Bioshock Infinite minimal advantage for fewer cores higher clocks but at 120+FPS where it doesn't make a significant difference even with high refresh rates

Vortez with GTX 950 at 1080p http://www.vortez.net/articles_pages/intel_broadwell_e_core_i7_6900k_6950x_review,17.html
Doom minimal differences
Battlefield 4 small difference that favors many cores

Overclock3d with GTX 980https://www.overclock3d.net/reviews/cpu_mainboard/intel_6950x_6900k_6850k_cpu_review/17
Small differences favoring clocks in Sleeping Dogs and Tomb Rider
Minimal differences in Hitman Absolution

Hardware.info
https://nl.hardware.info/reviews/67...arks-gtx-960980-f1-2015n1920x1080-mediumultra
With GTX 960 at 1080 medium settings - minimal differences in F1 2015 while at ultra settings F1 2015 favors clocks
With GTX 980 at 1080 medium in F1 2015 , clocks win by a substantial margin but at up to150 FPS while at ultra settings clocks win by 110 vs105 FPS so again rather high FPS
GTA V with 960 at 1080p medium, 6700k and 6900k come out on top while at Ultra settings very odd results at unplayable FPS
GTA V with 980 at 1080p medium, many cores wins but at 120 FPS , while at Ultra settings there is a minimal advantage for many cores - 41 vs 39 FPS
Mad Max with 960 at 1080p minimal differences with both Medium and Ultra settings
Mad Max with 980 at 1080p same minimal differences at both settings

Sweclockers http://www.sweclockers.com/test/221...-och-i7-6800k-familjen-broadwell-e/12#content
with GTX 980 TI at 1080p
Dragon Age: Inquisition , GTA V minimal differences
The Witcher 3 small differences favoring many cores for min FPS but at high FPS, 92 vs 97
Fallout 4 favors clocks , at over 60FPS min framerates here
Tomb Rider minimal differences but with DX12 many cores take the lead

KitGuru with GTX 980 Ti at 1080p http://www.kitguru.net/components/ryan-martin/intel-core-i7-6950x-broadwell-e-10-core-cpu-review/7/
Ashes of the Singularity and Middle-earth: Shadow of Mordor minimal differences
Grand Theft Auto V small win for clocks at 84FPS so small relevance

Hexus http://hexus.net/tech/reviews/cpu/93017-intel-core-i7-6950x-14nm-broadwell/?page=6
GTX 1080 at 1440p and 4k
Dirt Rally small advantage for more cores at 1440p and minimal advantage at 4k
Rise of the Tomb Raider minimal advantage for more cores
Tom Clancy's The Division minimal advantage for clocks

Computer Base https://www.computerbase.de/2016-05/intel-core-i7-6950x-6800k-test/5/
At 1080p with GTX 980 TI as far as i can tell - don't know german
Games tested Assassin's Creed Unity , Call of Duty: Advanced Warfare , F1 2015 , GTA V , The Witcher 3 , Total War: Attila
Overall minimal advantage for more cores
In individual games it goes either way with minimal and small differences with most games at very high FPS
EDIT: They have more tests than the ones i've initially spotted, the most interesting ones are multitasking.

For testing done with GTX 1080 at 1080p, results are discarded for flawed and misleading methodology that creates an artificial bottleneck just for the sake of it.
PC Gamer http://www.pcgamer.com/the-broadwell-e-review/
Techspot http://www.techspot.com/review/1187-intel-core-i7-6950x-broadwell-e/page6.html

 

[Abwx]

So about +3% in those games at such res... IF the Launch freq - ES tested = +20%.

(6400->6500 gets +3% in games at such res).

Wrong, as is usual with you, for one 6400 to 6500 the difference is 4% in the graph and turbo frequency is 9% apart, how did you come to the conclusion that there s 20% frequency difference between these CPUs...??

Because you did look at the base frequency perhaps, certainly what count in single theaded games, but since it s to mislead people why not after all, each one with his own means...

 

[lolfail9001]

My entire point was that people like you end up believing the very opposite of where the truth is.

So, where the truth is? That if i run a game on same hardware on higher resolution it will run faster ?

If you test a GTX 1080 at 1080p to induce a bottleneck that shows higher clocks getting 165FPS instead of 150FPS for many cores, it doesn't help anyone

No, it helps me to conclude that if bottleneck is indeed CPU based (and yes, you can and will get GPU bottlenecked on 1080 in 1080p on some games), i will not get higher than 165FPS on this CPU and 150 on that CPU on higher resolution and more powerful GPU. It is not rocket science.

Do the same tests at 1440p and 4k and the reality changes quite a bit.

They both start to run at 170 FPS? I am intrigued, maybe 1440p monitor is worth the premium after all.

You are better off with many cores.

This conclusion does not follow from your post Now, onto your evidence... Wait, what is your point here? That 7700k is not the best gaming CPU and that say 6850k or, Zen, should it deliver, would be better choices of a CPU for that? Because of what, exactly?

For testing done with GTX 1080 at 1080p, results are discarded for flawed and misleading methodology that creates an artificial bottleneck just for the sake of it.

You are really the sort of people that never understood the actual flaw with [H]'s CPU testing in games, are not you?

 

[scannall]

It would really be great if people would stop derailing this thread with petty bickering.

 

[Dresdenboy]

The issue is that they mislead, many people think it's better with high clocks and it's pretty clear that many do think that.
At a reasonable config (as in well balanced system) and res, with latest gen CPUs,you are not gonna be clock bound but you might see upsides in some games that can use more cores or like large L3$. The future is likely to favor more cores too.
Ofc there are cases when one could build a budget constrained system with an unlocked Pentium and a GTX 1070 and aim for 4k lower settings or 1440p and then you are CPU bound and must look at how well you can get away with it.

It would be interesting to know whether 5%, 50%, or even 90% of gamers outside of tech forums populated by many well paid people living in high salary countries would have to look for a good balance between CPU and GPU performance to get as much as possible for their given/set budgets. Spending $200 more on the CPU to get 20% more fps at 720p (and maybe 5% at 1440p) might be the wrong way then as the same money would could easily increase performance by 1.5X to 2X on the GPU side alone for res >=1080p.

Power consumption and other use cases might come in as additional requirements.

 

[TheELF]

might be the wrong way then as the same money would could easily increase performance by 1.5X to 2X on the GPU side alone for res >=1080p.

But only if your CPU can give you 1.5x to 2x more performance,and that's why we see benches with no gpu bottlenecks.

 

[Atari2600]

Concerned about gaming perf..?.

http://www.guru3d.com/news-story/fr...ng-sample-amd-ryzen-processor-benchmarks.html

Now just imagine that the retail SKUs will have 20% better ST perf than the early sample used in this review..

By a quick and dirty calc and assuming that performance is linear with the turbo clock - Zen is **clock-for-clock** right on top of the i7 6900k. As in, within 1%.

If its the base clock that both CPU are actually running at, then the i7 6900k is quicker per clock.

Assuming AMD release Ryzen vanguard at 3.6/4.0, then it beats the i7 6900k in outright performance, by around 3-6%.
Indeed, Zen would beat all bar the i7 6700k, which is around 3-6% quicker, but note that is due to the higher clocks.

If AMD could produce a 4C8T Zen clocking at 4/4.2 like i7 6700k, then they'd likely beat it too (assuming games don't really scale beyond the 4C8T option).

 

[Abwx]

By a quick and dirty calc and assuming that performance is linear with the turbo clock - Zen is **clock-for-clock** right on top of the i7 6900k. As in, within 1%.

If its the base clock that both CPU are actually running at, then the i7 6900k is quicker per clock.

Assuming AMD release Ryzen vanguard at 3.6/4.0, then it beats the i7 6900k in outright performance, by around 3-6%.
Indeed, Zen would beat all bar the i7 6700k, which is around 3-6% quicker, but note that is due to the higher clocks.

If AMD could produce a 4C8T Zen clocking at 4/4.2 like i7 6700k, then they'd likely beat it too (assuming games don't really scale beyond the 4C8T option).

According to Canard PC Zen didnt make it more than 3.3GHz and the 6900K has a turbo that can get up to 4GHz with Turbo boost 3.0.

So the frequency difference in ST could be as high as 20% and with linear scaling Ryzen would be notably faster, now if scaling is say 60% then it would still outmatch the 6900K in those tests.

 

[imported_jjj]

It would be interesting to know whether 5%, 50%, or even 90% of gamers outside of tech forums populated by many well paid people living in high salary countries would have to look for a good balance between CPU and GPU performance to get as much as possible for their given/set budgets. Spending $200 more on the CPU to get 20% more fps at 720p (and maybe 5% at 1440p) might be the wrong way then as the same money would could easily increase performance by 1.5X to 2X on the GPU side alone for res >=1080p.

Power consumption and other use cases might come in as additional requirements.

Ofc maybe 1-2% of discrete GPU sales are GTX 1080 and the focus shouldn't be there by any means. And ofc the PC is not about just gaming. Here I was only addressing this specific misconception about clocks vs cores. where folks that want the most think 4 cores is best. It's not ,even with Broadwell's small IPC disadvantage on top of clocks. Testing with a high end GPU at 1080p is like disabling 2 cores in a quad to show that core scaling is more important. Quad is the sweet spot because devs have to make it work with that limitation but it's not the best.

Most of the folks i know, do look for a well balanced system for their specific needs and inside a specific budget,often very limited. In general, the fewer bucks you can spare, the more likely you are to care about bang for the buck and do more research.
The conversation here ,more cores vs 4 core, is in the Zen context, hopefully AMD doesn't shoot itself in the foot and they bring 8 cores into the "mainstream". I think it's a valid convesation in this context and has longer term implications too.
More cores at reasonable prices would also give one the additional option to pay less for a quad and invest in something else, if that better fits their needs.I am sure many will do just that.

For power consumption, in gaming, the difference should in theory be minimal but can't quickly recall some tests that validate the theory. Hexus had some basic numbers for gaming with a GTX1080 http://hexus.net/media/uploaded/2016/5/28dd7066-76f5-417a-9854-941509906224.png

 

[DrMrLordX]

It would really be great if people would stop derailing this thread with petty bickering.

This is what happens when there is a paucity of useful information.

 

[Abwx]

This is what happens when there is a paucity of useful information.

This has nothing to do with infos being scarce but more certainly to a will to theadcrap the subject, just look at Intel s related threads, wether there s news or not you wont see a bunch of derailers poping at each occasion..

 

[DrMrLordX]

Enthusiasm about upcoming products is certainly a factor. I think there are more people (overall) interested in the subject of Ryzen/Summit Ridge than anything Intel has right now, at least as a matter of curiosity. You just don't see as many posts period in those threads as you do over here. So I'll stand by my statement.

People want to talk about things related to Summit Ridge, and they've run out of subject matter.

On the Intel side, we now have exact dates for Skylake-X/Kabylake-X/LGA-2066, and we have a release quarter for Coffeelake-S. So there's news on that front. It's just not all that shocking.

 

[dsplover]

I hope the AMD delivers as I pay no attention to benchmarks as my particular applications are not games, fame rates.
I'm buying it regardless of the pro/con inferences.
On a i7 4790k @ stock speeds I currently have a 35% overhead.

Doubling the cores doubles my workload.
But if 1 Core hits 80% that's really pushing things.
I have multi core optimized apps but one app for the sake of coherency and realtime performance is uni core.
For 500 bucks this CPU could really change the way Intel holds out on production.

Supermicro should jump on this with their Server PCB style enthusiast Line of boards.

XITE'd.......

 

[CentroX]

Sweclockers member found this mystic number in amds blender test

 

[Sweepr]

According to the new information, AMD will present a further presentation of RYZEN at the end of February, and will also include the final technical data. This information will be officially presented in the first half of March - at the same time, the launch of RYZEN will take place on the market. However, for the start, there will be only the 8-core model, which can handle 16 threads. The 4-core model will be released later.

There is still a piece of information at the price. "Prices are not as cheap as you might think" should probably then imply that it can not be assumed that the 8-core processor will be found again in the low price range. Since AMD always likes the model against the Intel Core i7-6900K, which costs a little more than 1.100 euros, the first RYZEN processor may move somewhat below it, but also AMD has nothing to give away.

www.hardwareluxx.de/index.php/news/hardware/prozessoren/41678-geruechtekueche-ryzen-fertig-doch-nur-8-kern-modell-zum-start.html

 

[lolfail9001]

Fairly positive above is rewrite of this chiphell rumor.

 

[Abwx]

www.hardwareluxx.de/index.php/news/hardware/prozessoren/41678-geruechtekueche-ryzen-fertig-doch-nur-8-kern-modell-zum-start.html

According to the new information, AMD will present a further presentation of RYZEN at the end of February, and will also include the final technical data. This information will be officially presented in the first half of March -

Lol...

 

[Dresdenboy]

But only if your CPU can give you 1.5x to 2x more performance,and that's why we see benches with no gpu bottlenecks.

That's what I meant with balancing. I might spend $200 more for the CPU alone, or the GPU alone, or add $100 to each of them, etc.
If you mean 1.5-2x performance in non-bottlenecked tests with a strong GPU (e.g. pure CPU performance), what does this tell me about 1440p performance at 4x the pixel count, if I'm on a budget? Aren't the chances better to see such gains by spending the money in full or a bigger part of it on the GPU?

Of course, this also depends on the base I'm starting from.

 

[krumme]

Lol...

Yeaa. Nice catch

 

[EightySix Four]

ccording to the new information, AMD will present a further presentation of RYZEN at the end of February, and will also include the final technical data. This information will be officially presented in the first half of March

Lol...

My guess is that it means they will give a NDAd presentation to review sites at the end of February with the NDA lifted during launch in the first half of March.

 

[zinfamous]

Mt guess is that it means they will give a NDAd presentation to review sites at the end of February with the NDA lifted during launch in the first half of March.

this

 

[The Stilt]

Sweclockers member found this mystic number in amds blender test

Probably the design id.
31021 for AM3+ Piledriver.

 

[lolfail9001]

Lol...

You are right, actually, only half a month between tech day and NDA lift? I expect a full month, as with Polaris.

 

[krumme]

You are right, actually, only half a month between tech day and NDA lift? I expect a full month, as with Polaris.

Ahh ok. Unfortunately thats not what they wrote.
But hey when they sit there inventing stuff up and rewriting in 15 minutes - that is probably the most likely scenario but no less just clickbait - thats what happens in the writing.