AMD 32core CPUs benchmarked!!

[Space Reptile]

1 minute ago, patrickjp93 said:

No, those literally do not affect performance.

well then lets hope that sample can clock higher than 1.44 when it hits the shelves , because well , 64 cores and bearly reaching 4790k/6700k performance (in multi core gb v4)  is not that good 

 

but we will see , samples are taken w/ a grain of salt since they dont represent whatever we get in the end 

[patrickjp93]

Just now, Space Reptile said:

well then lets hope that sample can clock higher than 1.44 when it hits the shelves , because well , 64 cores and bearly reaching 4790k/6700k performance (in multi core gb v4)  is not that good 

 

but we will see , samples are taken w/ a grain of salt since they dont represent whatever we get in the end 

Geekbench is also not good at scaling to multiple threads, and Amdahl's law is still at play. I wouldn't draw any conclusions from this for that and all the previously mentioned reasons.

[Space Reptile]

Just now, patrickjp93 said:

Geekbench is also not good at scaling to multiple threads, and Amdahl's law is still at play. I wouldn't draw any conclusions from this for that and all the previously mentioned reasons.

agreed , lets just hope it kicks ass and levels the playingfield when it comes to marketshare 

[patrickjp93]

Just now, Space Reptile said:

agreed , lets just hope it kicks ass and levels the playingfield when it comes to marketshare 

That hope is delusional I'm sorry to say.

[Notional]

24 minutes ago, patrickjp93 said:

Can I hit you with a bat?

 

DRIVERS?! IT'S A CPU BENCHMARK! CPUS DO NOT NEED NOR GET DRIVERS!

 

Also, it's at 1.44GHz on an engineering sample with A0 stepping. God people, just, just READ! That's all I'm asking of you...

It does require proper scheduler support in the OS though. Especially in Linux, we've seen what proper AMD A/CPU support can do to performance.

6 minutes ago, patrickjp93 said:

That hope is delusional I'm sorry to say.

Remember it is very easy to accomplish, even with having an IPC ~10% slower than Intel. Nothing a lower price/more cores/higher clock cannot deal with. 

[SpaceGhostC2C]

1 minute ago, Notional said:

Remember it is very easy to accomplish, even with having an IPC ~10% slower than Intel. Nothing a lower price/more cores/higher clock cannot deal with. 

Or less artificial market segmentation (locked, unlocked, Z, H, etc)

[normpearii]

5 hours ago, MarcoTexture said:

impressive for 1.4 ghz. But but butttttt why did they benchmark it at that frequency? I mean what's the point?

 

Easy, this most likely a low voltage, low TDP server chip.

 

5 hours ago, PCGuy_5960 said:

Idk.... If one core can get roughly 1000pts in Geekbench then 32 cores shouldn't be above the 20.000pt mark?

Turbo?

[PCGuy_5960]

8 minutes ago, normpearii said:

 

Easy, this most likely a low voltage, low TDP server chip.

 

Turbo?

These opterons don't have turbo as far as I know....

[dalekphalm]

4 minutes ago, PCGuy_5960 said:

These opterons don't have turbo as far as I know....

We don't know one way or the other. Considering that AMD's GPU's have "Turbo" boosting technology now, I assume that they will roll out the features to their Zen CPU's too.

[PCGuy_5960]

2 hours ago, rcarlos243 said:

I made another test.

This time with Hyperthreading disabled and Uncore Frequency at 1.4Ghz

Based on these results, it would take approximately 18 Intel Cores to achieve the same multicore score as the 64 on these CPUs...

(running at 1.4GHz....)

I do hope that they improve their CPUs so that they can compete with Intel...

[PCGuy_5960]

1 minute ago, dalekphalm said:

We don't know one way or the other. Considering that AMD's GPU's have "Turbo" boosting technology now, I assume that they will roll out the features to their Zen CPU's too.

Maybe you are right...

[The Benjamins]

3 hours ago, The Benjamins said:

For reference.

My D-1541 scored 2918 / 14504 (all stock 2.1Ghz / 2.7Ghz)

https://browser.geekbench.com/v4/cpu/182531

 

4 Win 10 VM idling 

 

33 minutes ago, PCGuy_5960 said:

Based on these results, it would take approximately 18 Intel Cores to achieve the same multicore score as the 64 on these CPUs...

(running at 1.4GHz....)

I do hope that they improve their CPUs so that they can compete with Intel...

this is my score on a D-1541, and it close to the leaked score.

[Gr0egercesg]

Lookin good, Lookin good

AMD stock price = +25%

 

[ScratchCat]

Could someone tell me If geekbench actually represents the speeds between different architecures correctly now such as ARM vs x86. IIRC it was biased towards ARM in Geekbench 3

[PCGuy_5960]

13 minutes ago, The Benjamins said:

 

this is my score on a D-1541, and it close to the leaked score.

I used 18 cores as a worst case scenario....

[PCGuy_5960]

1 minute ago, ScratchCat said:

Could someone tell me If geekbench actually represents the speeds between different architecures correctly now such as ARM vs x86. IIRC it was biased towards ARM in Geekbench 3

That's Geekbench 4

[ScratchCat]

1 minute ago, PCGuy_5960 said:

That's Geekbench 4

This was information I got before GB4 was release. I think GB4 is more balenced as x86 gets a boost in points while ARM cpus I have tested get a smaller to 0 boost.

[The Benjamins]

Is this another Zen chip? 

https://browser.geekbench.com/v4/cpu/156454

[PCGuy_5960]

1 minute ago, The Benjamins said:

Is this another Zen chip? 

https://browser.geekbench.com/v4/cpu/156454

Seems like one... Impressive for just 2.5GHz!!!

[ScratchCat]

2 minutes ago, The Benjamins said:

Is this another Zen chip? 

https://browser.geekbench.com/v4/cpu/156454

Must be as its a ES and its got a 1core 2 thread ratio indicating SMT.

Nice find!

[ScratchCat]

1 minute ago, PCGuy_5960 said:

Seems like one... Impressive for just 2.5GHz!!!

Again I tested with the i5 2410m and it score 2693 at 2.9Ghz and its 5Gen only(I know there isn't much improvement).

May be actually good given that its an ES

[PCGuy_5960]

3 minutes ago, ScratchCat said:

Again I tested with the i5 2410m and it score 2693 at 2.9Ghz and its 5Gen only(I know there isn't much improvement).

May be actually good given that its an ES

It is competitive with the G4400 on the Intel side.... How good these CPUs will be depends on the pricing of them... If AMD asks $100 for that CPU then it's not good... If it asks $60 for it, then it's a good buy...

[Doobeedoo]

Hmm, ok. Though their highest clock CPU will be main interest to see that single core performance. 

[Coaxialgamer]

7 hours ago, MarcoTexture said:

impressive for 1.4 ghz. But but butttttt why did they benchmark it at that frequency? I mean what's the point?

these are very early samples , judging by the stepping . It's most likely because of that

[Coaxialgamer]

Also , it's worth pointing out the chip is most likely chocking on memory bandwith