AMD "Summit Ridge" ZEN CPU at 2.80 GHz Beats 3.40 GHz Core i5-4670K

[WaroDaBeast]

So far we have only seen the result of one benchmark that favours AMD's moar cores philosophy.
That also highlights how poorly this E.S. of Zen will do in normal benchmarks that can use only 4 cores or less, clock speeds will be their downfall again, they need to ramp them up.

I don't get it. FX CPUs have high clocks; the area they lack in is IPC...

i5 4670 is 21% faster frequency, 100% less threads = 10% slower in fps

I stopped reading after that "100 % less" part.

 

[the54thvoid]

I stopped reading after that "100 % less" part.

We all make mistakes. Now corrected thanks to @Caring1 for pointing it out.

My error was to simply use 'negligent' logic, where 8 is 50 % of 16, 4 is 50% of 8, therefore (falsely) 100% less than 16. Of course, 4 is 75% less than 16.

You know, mathematical mistakes are often actually very interesting. They can almost seem logical.

 

[bug]

Found this discussion stating AotS can use up to 16 threads, or one thread per core.
https://hardforum.com/threads/ashes...to-16-threads-with-1-thread-per-core.1892735/

Well, those cores are only at about 50% load, so an 8 core would do just as well. Still, the i5 is only 4 cores...

 

[ZoneDymo]

I see some thing good but it's no excellent, looks pretty sub par so far. I do know that i have less hope due to this news.

Price is were it's going matter for a lot of people.



You could all so look at that as 30 playable and 23fps not playable and some 60 playable and 53 not.

Apart from those rare games that really only work on 60 fps and every minor drop results in it looking really choppy, nobody would see a difference between 60 and 53 fps..

23 and 30 is so low on the spectrum that obviously you do notice the difference so no, that comparison does not hold up.

 

[iO]

Promising, though a single AotS bench run is pretty meaningless...

But its a good sign that there are already enough ES floating around for leaks like this.

 

[costeakai]

If it turns out to be competition... it is good news for everyone.

not for me, c'se i'm broken , with a bad mid-term perspective

 

[Caring1]

I don't get it. FX CPUs have high clocks; the area they lack in is IPC...

True, and in the past they have compensated with moar cores, and faster clocks.
This example of Zen though seems to be slower clocked and would be more competitive if it was faster, they seem to have improved the IPC, although we will have to wait to see if they reached the claimed 40% improvement they wanted.

 

[v12dock]

WCCF is not a credible source let me take my grain of salt first...

 

[G33k2Fr34k]

New 8C/16T cpu is slightly faster than old 4C/4T. What a shocker.

In this benchmark, a core i7 4790 at 4GHz beats the 8-core i7-5960X at 3GHz. Generally speaking, Vulkan has better performance with +4 threads than DX12.

 

[HD64G]

Let's compare this result to one that was done by a i7 8 core-16thoreads owner guys. Not sure this game fully utilizes more than 6-8 threads in total. A wprime benchmark result would be great for this type of comparisons. And a single-threaded one also to compare the IPC where was the weakness for AMD CPUs and Zen is supposed to cure.

 

[Melvis]

 

[BiggieShady]

Maths.... (relative to Intel as base measurement)

i7 4790 is 29% faster frequency, 50% less threads = 11% faster in fps

i5 4670 is 21% faster frequency, 75% less threads = 10% slower in fps

so the frequency isn't relevant, the core count is but the 4 core/thread part is pretty much 10% slower and the 8 thread part is 10% faster. The 8% speed discrepancy between i5 & i7 seems to have little outcome effect. AotS is using threads better but it must tank at some point between 4 - 8 threads.

A better 'leak' would be a down clocked 8 core (16 thread) Sandybridge Enthusiast chip (given how old it is).

Edited for maths......

They are all at max turbo during the benchmark run. I think you are trying to do this:

[TABLE=head]
cpu|max clock|clock ratio|cores|threads|fps|fps gain|intel gain adjusted for clock|amd gain adjusted for threads|amd gain adjusted for threads and clock
Intel Haswell i7|4.0|1.05|4|8|65.4|1.24|1.18|-|-
AMD Zen ES|3.2|0.84|8|16|58|1.1|-|0.93|1.1
Intel Haswell i5|3.8|1.0|4|4|52.6|1.0|1.0|-|-
[/TABLE]

We would again see 10% better fps on zen compared to i5, if zen was equally clocked as i5 and SMT was off ... and that's with double number of cores

Edit: AMD SMT 8 to 16 threads speedup is projected from calculated Intel HyperThreading 4 to 8 threads speedup, and therefore certainly off by some margin

 

[G33k2Fr34k]

A higher clocked quad core beats an 8 core CPU in this benchmark.

 

[dyonoctis]

Here the score on a Intel Core i7 5960X (HT off, 4 GHz).

Now here is zen result on high:

However do note that the Zen benchmark is using mssa, while the other benchmark does not.

 

[EarthDog]

According to performance numbers of an AMD "Summit Ridge" ZEN CPU engineering-sample put out by WCCFTech, AMD's claims of IPC gains are gaining credibility, and showing signs of the gaming PC processor market warming up again. An engineering sample featuring 8 cores and 16 threads (via SMT), beat Intel's Core i5-4670K processor. This sample featured clock speeds of 2.80 GHz, with 3.20 GHz boost.

The "Summit Ridge" sample provided 10 percent higher frame-rates than a Core i5-4670K, in the "Ashes of the Singularity" 1080p benchmark. The chip is still convincingly beaten by 12 percent, by a Core i7-4790 (non-K), running at 3.60 GHz, with 4.00 GHz boost. This shows that AMD could leverage the new 14 nm FinFET process to crank up clock-speeds, and produce SKUs competitive with current Intel "Skylake-D" Core i5 and Core i7 processors.



Source: WCCFTech

I certainly hope an 8c/16t processor beats a native quad core....................................wow. And then is "convincingly beat" by a 4c/8t processor.....Haswell, not even Skylake.

Well, they are close and that is half the battle!!!

 

[ZoneDymo]

WCCF is not a credible source let me take my grain of salt first...

Well lets be honest here, by your logic TPU is not a credible source either because they just forward the same information just with WCCF as the middleman.

From WCCF's article "the benchmarks given in this article are that of the Engineering Sample of AMD Zen and NOT entirely representative of the final product."

Secondly WCCF just posted/forwarded these benchmarks that some person who had the sample cpu uploaded somewhere, like any site (and case in point here TPU) would do.

So yeah, honesty....

 

[dyonoctis]

I have to say: It's great to get benchmark and all, but cinebench would have been a better idea. Or just use a 4 core 8 thread zen, seeing that 16 thread seems to be less efficient in games. I mean, the reason as to why the HT is turned off by reviewer using a 8 core intel cpu must be because it would be bad to do otherwise ?

 

[ShurikN]

I certainly hope an 8c/16t processor beats a native quad core....................................wow. And then is "convincingly beat" by a 4c/8t processor.....Haswell, not even Skylake.

Well, they are close and that is half the battle!!!

There's also a huge difference in clocks.

 

[Ubersonic]

This is quite a return to form for AMD

Reminds me of when Intel finally launched it's "Core" CPU's after years of trying to polish the P4 turd

Hopefully now we will see some real competition

 

[EarthDog]

There's also a huge difference in clocks.

Well aware... it still wouldn't matter if they were the same clockspeed...it would still easily lose to a 4c/8t CPU.

"close" is relative compared to where AMD was compared to where it is.

 

[Sempron Guy]

WCCFTECH spreads bs info. AMD gets the blame if it ain't true. Rinse and repeat.

 

[EarthDog]

WCCFTECH spreads bs info. AMD gets the blame if it ain't true. Rinse and repeat.

Bwaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaahahahha wow. AMD is the problem bud. People get irritated at THEIR claims, not these leaked benchmarks. The leaked benchmarks are where people first to start FACEPALM off of their claims.

 

[BiggieShady]

Zen would look much better next to 5960X in this benchmark ... same number of cores and threads, intel at 3.5 GHz and zen at 3.2 GHz ... and the results would be similar

 

[G33k2Fr34k]

Well aware... it still wouldn't matter if they were the same clockspeed...it would still easily lose to a 4c/8t CPU.

"close" is relative compared to where AMD was compared to where it is.

This engineering sample has a max turbo frequency of 3.2GHZ. The i7 4970's turbo frequency is 4GHz, which 25% higher. If you look at the graph a few posts above this one, you'll see that the i7 4970 is ~9% faster than the 8-core i7 5960X at 3.5GHz.

 

[ShurikN]

Bwaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaahahahha wow. AMD is the problem bud. People get irritated at THEIR claims, not these leaked benchmarks. The leaked benchmarks are where people first to start FACEPALM off of their claims.

AMD made one single claim and it's the 40% IPC increase over current arch. It's too early to know if it's true or not.