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

[lobz]

Can anyone with BCLK overclocked Skylake chime in? BCLK OC bios should make AVX on Skylake work at same speed as Ryzen theoretically does (never switch into 256 bit mode, that is), so it could help to verify that PN/Physics test are actually AVX based and there's nothing else fishy going on with them.

+1, should be fun to see

 

[lobz]

https://www.cpubenchmark.net/cpu_test_info.html



I believe the Bullet Physics Engine uses AVX2.

"Prime Number Test
The Prime Number Test aims to test how fast the CPU can search for Prime numbers, reported as operations per second. A prime number is a number that can only be divided by itself and 1. For example, 1, 2, 3, 5, 7, 11 etc. This algorithm uses loops and CPU operations that are common in computer software, the most intensive being multiplication and modulo operations. All operations are performed using 64-bit integers."

I'm at a loss...

 

[lolfail9001]

I'm at a loss...

Well, one of things AVX2 allowed is usage of integer vectors, IIRC.

• expansion of most vector integer SSE and AVX instructions to 256 bits

The other thing that i never quite heard of AVX2 using sieves.

 

[Crumpet]

I believe the Bullet Physics Engine uses AVX2.

It has an AVX mode, though it can be disabled I think

 

[inf64]

I would disregard those 2 outliers. We will have other benchmarks that can test similar things(ie. wprime). Ryzen looks extremely good. If we could only know if that sample was running with Turbo enabled or not.

 

[blublub]

I was wondering if you could tell if XFR was enabled.

I looked at the Sandra result and compared it to a random 6900K score that seemed like it had turbo disabled (3.2 ghz):

MM Integer: 718 vs 525 for the Ryzen "38_34"
MM Long Int: 211 vs 159
MM Quad Int: 3299 vs 3589
MM Single Float: 654 vs 522
MM Double Float: 374 vs 300
MM Quad Float: 13352 vs 13271

Not sure what to make of it beyond perhaps some have AVX(2) optimizations and some don't.

On this MB from the leak XFR is not supported. So no XFR but maybe normal Turbo with 3.8ghz was working

 

[Dresdenboy]

Eh, their HPC share is not going to grow with such AVX performance (that statement is independent from these benchmarks because we do not know if they use AVX).

I think many expected this already 16 months ago when it became clear that Zen would have 128b FPUs. Aside from Snowy Owl they'll not have a pure CPU based HPC contender until at least 2018.

 

[inf64]

Yep, that is why Lisa Su said that they will cover 85% of the server market with Zen. The missing 15% is HPC. Zen+ should address that I guess.

 

[Greyguy1948]

Regarding Sisoft Sandra Multimedia:
From 3.14 GHz to 3.4 GHz
(Eng Sample <-------> Ryzen)
int...... 458.....................526
long....137.....................159
quad....2.7...................... 3.6
single..348.....................522
double.201....................300
quad.....9.5.....................13.3

Floating point improvement is impressing!

 

[lobz]

Well, one of things AVX2 allowed is usage of integer vectors, IIRC.


The other thing that i never quite heard of AVX2 using sieves.

thanks

 

[Greyguy1948]

CPU Marks. Some of the tests use no L3 cache.

Test............Memory/ core
Int math.......240 kB
Compr...........16 kB
Prime.............4 MB
Encrypt........ 1 MB
String sort.....25 MB
Fp math.......240 kB

 

[Dresdenboy]

Why is sm625 calling it a turd? At those clockspeeds thats good results by ryzen.

We might have to leave the realm of hard mathematical facts and move to human sciences to understand this.

 

[sm625]

Why is sm625 calling it a turd? At those clockspeeds thats good results by ryzen.

Look at the physics score. That is what is going to condemn it to gaming mediocrity. The memory score is also a grave concern, if it is cache bandwidth that is pulling it down.

 

[sm625]

LMAO at the people going on full damage controll...why be worried guys? Ryzen is a clear success..at 3.8ghz it has almost the same ST of an i3 4160 running at 3.6ghz..

Because that i3 is 3 years old? If you take an objective perspective, you might recall that bulldozer also looked pretty decent compared to 3 year old intels chips at the time.

 

[Vesku]

Sudden spike in "concern trolls" in CPU enthusiast forums has me excited for this launch.

 

[Glo.]

Because that i3 is 3 years old? If you take an objective perspective, you might recall that bulldozer also looked pretty decent compared to 3 year old intels chips at the time.

You understand that if your claims, are not turning out to be reality, that Ryzen is a failure, you are deserving to be banned for spreading FUD, as well as vissarix?

Sudden spike in "concern trolls" in CPU enthusiast forums has me excited for this launch.

Yep, its best indication, that Ryzen is strong, and Intel is in trouble.

 

[ultima_trev]

Look at the physics score. That is what is going to condemn it to gaming mediocrity. The memory score is also a grave concern, if it is cache bandwidth that is pulling it down.

Because that i3 is 3 years old? If you take an objective perspective, you might recall that bulldozer also looked pretty decent compared to 3 year old intels chips at the time.

1) You took some cherry picked results of a 4.7 GHz i7 5960X, 4.8 GHz i7 5820K and 4.2 GHz i7 6900K to make the physics score appear far worse than it needs to be. That physics test also seems to make heavy use of AVX, where Intel will still hold a big advantage. BUT MOST GAMES STILL DO NOT USE AVX!!

Thanks to WhyCry and the VideoCardz team for posting more unbiased scores.

2) You seem to imply that going from IPC like Conroe (Piledriver) to IPC between Ivy Bridge and Haswell is somehow a bad thing?

You have got to be the single biggest troll I've ever seen on tech forums anywhere.

 

[dfk7677]

Afaik, GPUs do the physics nowdays, not CPUs.

 

[lolfail9001]

Look at the physics score. That is what...

Name 5 games with Bullet physics engine. I agree that either it uses AVX or something is gravely wrong there, though.

The memory score is also a grave concern, if it is cache bandwidth that is pulling it down.

That i may agree with.

Because that i3 is 3 years old? If you take an objective perspective, you might recall that bulldozer also looked pretty decent compared to 3 year old intels chips at the time.

In single threaded bench my recently refreshed CPU loses to 3 year old i3 too, so what? Not to mention that single threaded bench is fairly flawed: x4 845 actually posts a result nearly matching my 6400.

Afaik, GPUs do the physics nowdays, not CPUs.

PhysX in mainstream engines (Unity/Unreal) runs on CPU. GPU physics is actually fairly rare.

 

[inf64]

I did a small comparison between A12 9800 and Ryzen sample in ST test. Interesting findings.
This test clearly shows IPC improvements between 3 different AMD generations(PD,SR,XV).

A12 9800
Single Thread Rating: 1824 @ 4.2Ghz
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A12-9800&id=2861

A10-7870K
Single Thread Rating: 1542 @ 4.1Ghz => 1580 @ 4.2Ghz
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A10-7870K

AMD FX-8350
Single Thread Rating: 1505 @ 4.2Ghz

XV is having ~15% higher score than SR at the same clock
SR is having ~5% higher score than PD at the same clock

Compare with Ryzen sample:
Single Thread Rating: 2046 @ Unknown clock

Let us see what might be the unknown clock (3.4Ghz or 3.8Ghz?)

If it was ran @ 3.4Ghz, in IPC sensitive benchmark like this Ryzen would be showing:
2046 x 4.2/3.4=2527 pts or
1) ~1.39x (39%) higher IPC vs XV. Checks well with AMD's quoted >40% IPC jump
2) 68% higher IPC vs PD

If it was ran @ 3.8Ghz, in IPC sensitive benchmark like this Ryzen would be showing:
2046 x 4.2/3.8=2261 pts or
1) ~1.24x (24%) higher IPC vs XV. This is a lot lower than AMD's quoted >40% IPC jump
2) 50% higher IPC vs PD

According to above math it is **more likely** that Ryzen chip ran that one ST particular benchmark at 3.4Ghz (than at 3.8Ghz). Whether Turbo was not activated at all or something else is in question remains to be seen.

 

[lobz]

Look at the physics score. That is what is going to condemn it to gaming mediocrity. The memory score is also a grave concern, if it is cache bandwidth that is pulling it down.

You've got to be kidding. There's no way you say that seriously.

 

[CatMerc]

Afaik, GPUs do the physics nowdays, not CPUs.

Depends on the physics really. Calculating tons of tiny particles is a lovely job for a GPU, calculating the movement of a large rigid object is still better done by the CPU.

 

[lolfail9001]

According to above math it is **more likely** that Ryzen chip ran that one ST particular benchmark at 3.4Ghz (than at 3.8Ghz). Whether Turbo was not activated at all or something else is in question remains to be seen.

Do note, however, that in this particular bench XV is only 20% behind Skylake per clock, making is **even more likely** that Passmark is Con-core biased, and as such using it to draw conclusions on Ryzen via AMD's statement would be at best weird.

 

[The Stilt]

I believe the Bullet Physics Engine uses AVX2.

At least the more recent versions do.
AVX, AVX2 and FMA.

Bullet is part of my test suite because of that.
Cache size and latency has a huge effect too.

 

[OrangeKhrush]

I did a small comparison between A12 9800 and Ryzen sample in ST test. Interesting findings.
This test clearly shows IPC improvements between 3 different AMD generations(PD,SR,XV).

A12 9800
Single Thread Rating: 1824 @ 4.2Ghz
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A12-9800&id=2861

A10-7870K
Single Thread Rating: 1542 @ 4.1Ghz => 1580 @ 4.2Ghz
http://www.cpubenchmark.net/cpu.php?cpu=AMD+A10-7870K

AMD FX-8350
Single Thread Rating: 1505 @ 4.2Ghz

XV is having ~15% higher score than SR at the same clock
SR is having ~5% higher score than PD at the same clock

Compare with Ryzen sample:
Single Thread Rating: 2046 @ Unknown clock

Let us see what might be the unknown clock (3.4Ghz or 3.8Ghz?)

If it was ran @ 3.4Ghz, in IPC sensitive benchmark like this Ryzen would be showing:
2046 x 4.2/3.4=2527 pts or
1) ~1.39x (39%) higher IPC vs XV. Checks well with AMD's quoted >40% IPC jump
2) 68% higher IPC vs PD

If it was ran @ 3.8Ghz, in IPC sensitive benchmark like this Ryzen would be showing:
2046 x 4.2/3.8=2261 pts or
1) ~1.24x (24%) higher IPC vs XV. This is a lot lower than AMD's quoted >40% IPC jump
2) 50% higher IPC vs PD

According to above math it is **more likely** that Ryzen chip ran that one ST particular benchmark at 3.4Ghz (than at 3.8Ghz). Whether Turbo was not activated at all or something else is in question remains to be seen.

Seems to be according to Passmark that it was at 3.4Ghz and no turbo