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

[Arachnotronic]

If you want Quad Cores (Thats what Core i7 6700K is) at $300-350 for the next 10 years yes its not essential to have AMD around

5820K is $389, and Coffee Lake is expected to arrive on the mainstream platform with six Cannon Lake cores in the first half of 2018.

Anyway, most games/consumer applications still don't really benefit from a ton of cores -- they want really fast cores. Harp on Intel all you'd like, but they have been killing it with single-threaded performance for gamers in recent years (4790K, 6700K, and soon 7700K).

 

[Sweepr]

If you want Quad Cores (Thats what Core i7 6700K is) at $300-350 for the next 10 years yes its not essential to have AMD around

They are releasing 6C/12T Coffee Lake for mobile in 1.5 year - and I don't see any pressure/competition from AMD there.

 

[Arachnotronic]

They are releasing 6C/12T Coffee Lake for mobile in 1.5 year - and I don't see any pressure/competition from AMD there.

Good point. I suspect this particular SKU is driven by the fact that the gaming notebook market is booming and OEMs would like to be able to sell six core parts to gamers (helps them boost system average selling prices/accelerate upgrade cycle). Intel basically builds what OEMs want it to because at the end of the day, Intel doesn't sell PCs to people, OEMs do.

Competition from AMD, especially given how little share it has in the personal computer market, is probably not a big driver of these SKU decisions.

 

[AtenRa]

Funny they will release a 6 core mainstream SKU just after ZEN launch when they had the opportunity to do it for more than 3-4 years (2017-2018)

 

[Arachnotronic]

Funny they will release a 6 core mainstream SKU just after ZEN launch when they had the opportunity to do it for more than 3-4 years (2017-2018)

Zen APUs will be limited to quad core configurations, so I don't really see the connection here. Summit Ridge is AMD's HEDT lineup, derived from server-first dies. Intel has offered 6, 8, and even 10 core products in this space for quite a while, also derived from server-first dies.

Frankly, I'm quite curious to see if they have made any significant power efficiency improvements on the process or the core, or have implemented a much more sophisticated turbo mechanism, because six cores in the same power envelope as a quad core could potentially lead to an unfortunate trade-off in 1-4 core performance.

 

[Abwx]

Not a single Zen leak has actually been good. AMD may have produced another Dozer.

You mean not a single leak using an ES that is unlikely to work on normal conditions..?..

Because the only leak that seemed to use a functional chip did show it beat HW-E in a FP heavy regular application...

As for AMD producing an uncompetitive chip, why not mortgage your home and massively short the stock if you are so sure..?..

 

[jpiniero]

Frankly, I'm quite curious to see if they have made any significant power efficiency improvements on the process or the core, or have implemented a much more sophisticated turbo mechanism, because six cores in the same power envelope as a quad core could potentially lead to an unfortunate trade-off in 1-4 core performance.

That I think is coming (on CFL/CNL)

 

[AtenRa]

Frankly, I'm quite curious to see if they have made any significant power efficiency improvements on the process or the core, or have implemented a much more sophisticated turbo mechanism, because six cores in the same power envelope as a quad core could potentially lead to an unfortunate trade-off in 1-4 core performance.

Im also quite curious how Intel's 6-core APUs will perform in Laptops at the same 15-30W TDPs as their Quad Core brothers and not lead to unfortunate trade-off in 1-4 core performance

 

[crashtech]

I wouldn't read too much into GB3, that was a pretty flawed benchmark. I think the best indications we have on Zen performance are the GB4 results as well as the AoTS results.

AMD has built a better core than anything Dozer, so that could help them in a number of markets. But for those expecting AMD to seriously challenge Skylake in terms of per-core performance will, IMHO, be disappointed. Even if you look at AMD's own public statement of +40% IPC over XV, Zen is still likely to be behind Haswell more often than not, and very much behind Skylake.

I think Zen will offer Sandy Bridge/Ivy Bridge levels of performance-per-clock. We already know what clocks AMD is planning to ship Summit Ridge at, the question for enthusiasts/gamers really comes down to -- as jpiniero points out -- how well this thing will overclock. I suspect Intel will have an advantage in this regard with its HEDT chips.

My comment was meant to question the wisdom of taking the leak seriously, I probably should have said so.

 

[Sweepr]

Im also quite curious how Intel's 6-core APUs will perform in Laptops at the same 15-30W TDPs as their Quad Core brothers and not lead to unfortunate trade-off in 1-4 core performance

The 6C/12T SKU is 35-45W.

And this mobile version is the main reason we're seeing a core count increase in mainstream desktop, not Summit Ridge.

 

[Arachnotronic]

Im also quite curious how Intel's 6-core APUs will perform in Laptops at the same 15-30W TDPs as their Quad Core brothers and not lead to unfortunate trade-off in 1-4 core performance

I don't think there will be 6 core chips in a 15W TDP from Intel. Looks to me like 6 core is restricted to 45W mobile chips and mainstream desktop, while high end 15W/28W SKUs will have up to 4 cores and GT3e GFX.

 

[AtenRa]

Thank God we have those Intel Laptop SKUs that come and save the day for us Mainstream Desktop users with a 6-core in late 2018 early 2019 when we could have had a mainstream 6-core in 2016 if it wasnt for those Laptops and their iGPUs

 

[Sweepr]

Actually H2-2017 according to DigiTimes, though early 2018 sounds more likely (1 year after Kaby Lake-S).

 

[Abwx]

A little quick and dirty arithmetic puts the ST perf of that leaked result on par with an old Phenom, assuming linear frequency scaling.

Actually it s even significantly lower than K10 at same frequency, hence the hopes of some usual suspects, i guess that some runs for some money will be experimented by the same candidates..

https://browser.primatelabs.com/geekbench3/compare/8073317?baseline=8076878

 

[AtenRa]

And this mobile version is the main reason we're seeing a core count increase in mainstream desktop, not Summit Ridge.

Actually H2-2017 according to DigiTimes, though early 2018 sounds more likely (1 year after Kaby Lake-S).

So, Desktop 6-core Coffee lake will come first (H2 2017) but its the Laptop 6-Core SKU coming one year later (Q2 2018) thats the reason we have the desktop part ??? is this some form of surrealism or what ??

 

[Abwx]

So, Desktop 6-core Coffee lake will come first (H2 2017) but its the Laptop 6-Core SKU coming one year later (Q2 2018) thats the reason we have the desktop part ??? is this some form of surrealism or what ??

There s related threads for whom want to discuss Intel products, no need to flood this one with irrelevant debate, indeed it s no surprise that this thread derailing is initiated by the same people at each possible occasion...

 

[jpiniero]

Actually it s even significantly lower than K10 at same frequency, hence the hopes of some usual suspects, i guess that some runs for some money will be experimented by the same candidates..

https://browser.primatelabs.com/geekbench3/compare/8073317?baseline=8076878

Don't you think this kind of explains the additional delays though? Could be kind of annoying if you have to use the absolute latest OS/VM/etc in order to get reasonable performance.

 

[lolfail9001]

Because the only leak that seemed to use a functional chip did show it beat HW-E in a FP heavy regular application...

As for AMD producing an uncompetitive chip, why not mortgage your home and massively short the stock if you are so sure..?..

It's funny how you call AMD's advertisements a "leak", not to mention it beat >underclocked< HW-E.
As for AMD producing a competitive chip, why not long the stock if you are so sure?

Anyways, i've said it already: i doubt it's real.

 

[Abwx]

Don't you think this kind of explains the additional delays though? Could be kind of annoying if you have to use the absolute latest OS/VM/etc in order to get reasonable performance.

At this point the chip is already in its final stepping and what is currently under validation is the full plateform, hence some Geekbench submission, but since 6 months are left before the first server chips are launched it s not in AMD s interest to disclose anything about the final perfs, that explain those weird results in respect of what should be displayed, FI Zen should exhibit more than 40% perf/clock advantage over Piledriver based Zambezi chips in GeekBench, wich is not the case at all with those numbers.

 

[jpiniero]

It's probably going to be more than 6 months now before you see Zen Server. Maybe more like 6 from when Summit Ridge actually ships.

At this point the chip is already in its final stepping and what is currently under validation is the full plateform, hence some Geekbench submission, but since 6 months are left before the first server chips are launched it s not in AMD s interest to disclose anything about the final perfs, that explain those weird results in respect of what should be displayed, FI Zen should exhibit more than 40% perf/clock advantage over Piledriver based Zambezi chips in GeekBench, wich is not the case at all with those numbers.

It could be easily explained by the MCM crippling performance one way or another. Could be software or hardware.

 

[itsmydamnation]

Not a single Zen leak has actually been good. AMD may have produced another Dozer.

Somethings in the geekbench results look odd others align. in GB4 ST perf per clock is ~= a 2processor 22core E5 while having the extra latency of 32 cores 2 processor. But that doesn't align to your agenda now does it , or is that just your lack of knowledge of how things like cache coherency actually work?


Personal attacks are not allowed
Markfw900

 

[Abwx]

It's probably going to be more than 6 months now before you see Zen Server. Maybe more like 6 from when Summit Ridge actually ships.

It could be easily explained by the MCM crippling performance one way or another. Could be software or hardware.

They said march for server chips...


As for the perfs they displayed what they wanted, or rather that s residuals from what they wanted to test, likely that GB run accordingly and bugless, and eventually a few other parameters but for sure that the plateform was voluntarly crippled if we are to look at some multicore results that are close to single core results while others scale more or less accordingly, anyway without knowing the frequencies it s a moot point to try any guess other than it doesnt show what is officialy announced...

 

[Arachnotronic]

Somethings in the geekbench results look odd others align. in GB4 ST perf per clock is ~= a 2processor 22core E5 while having the extra latency of 32 cores 2 processor. But that doesn't align to your agenda now does it , or is that just your lack of knowledge of how things like cache coherency actually work?

In GB4 Zen @ 1.44GHz gets a single-core score of 1141.

https://browser.primatelabs.com/v4/cpu/488835

Xeon E5 2699v4 @ 3.6GHz single core gets a single core score of 3616. 3616*(1.44/3.6) = 1446.

The perf/clock of these two chips are not in the same ballpark. Skylake Xeon, which is expected to launch around when Zen server does, will have improved perf/clock and probably improved clocks by virtue of the improved 14nm+ process and the inherently higher frequency capability of Skylake over Broadwell.

It really looks to me that in terms of perf/clock, Zen is not going to be Haswell class across the board. But, hey, the first Summit Ridges will be out in what? February? When I get my Zen-powered setup built, I'll be sure to run a lot of tests on it as well as my Broadwell-E

 

[itsmydamnation]

I picked the E5 with the best MT score when i did my comparision, the one you have picked for whatever reason is an outlier compared to the other 44 core E5 results.
https://browser.primatelabs.com/v4/cpu/45716

but then look at the sub score differences between your one and my one and you will see the likely difference is memory speed. So what speed memory is Zen running?

 

[Arachnotronic]

I picked the E5 with the best MT score when i did my comparision
https://browser.primatelabs.com/v4/cpu/45716

but then look at the sub score differences between your one and my one and you will see the likely difference is memory speed. So what speed memory is Zen running?

Who knows. Anyway, I'm going to run my BDW-E at 1.44GHz (DDR4-2400) and see what single core score I get...