AMD 32core CPUs benchmarked!!

[PCGuy_5960]

Here is the geekbench screenshot...

It is kind of disappointing... But nonetheless impressive for an engineering sample at 1.4GHz...

(Geekbench page)

(Source)

 

In addition, if geekbench is correct, it looks like each CPU has nearly 120mb of L3 cache

Edited September 6, 2016 by PCGuy_5960

[don_svetlio]

Disappointing? Did you even read? It's at 1.4GHz...

[FloRolf]

1 minute ago, don_svetlio said:

Disappointing? Did you even read? It's at 1.4GHz...

^this

 

try your 5820 at 1.4ghz (or any other cpu really) and it'll not do any better.

[don_svetlio]

Also, keep in mind that Geekbench may be registering AMD's version of hyper threads as cores. So those may actually be 16 cores per CPU, not 32.

[Dackzy]

dude that CPU is running at 1.44GHz... That is pretty damn good, try and underclock your CPU to 1.44GHz on all cores and no turbo

[GirlFromYonder]

At 1.44 ghz. not the ur i7 is 3.5 or something

[PCGuy_5960]

2 minutes ago, don_svetlio said:

Disappointing? Did you even read? It's at 1.4GHz...

Yes.... But it's 64 cores.... In that sense....

[don_svetlio]

Just now, PCGuy_5960 said:

Yes.... But it's 64 cores.... In that sense....

It's more likely to be 32 cores and 64 threads at 1.4 GHz without any sort of firmware optimization.

[PCGuy_5960]

2 minutes ago, don_svetlio said:

Also, keep in mind that Geekbench may be registering AMD's version of hyper threads as cores. So those may actually be 16 cores per CPU, not 32.

If that's the case, then the results would be more realistic....

[MarcoTexture]

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

[PCGuy_5960]

Just now, don_svetlio said:

It's more likely to be 32 cores and 64 threads at 1.4 GHz without any sort of firmware optimization.

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

[PCGuy_5960]

1 minute 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?

I think power consumption... As they are trying to compete with Intel and their lower TDPs

[don_svetlio]

Just now, PCGuy_5960 said:

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

I did mention the possible lack of firmware optimization, did I not? It's also not certain whether the benchmark ran properly on both CPUs or just one. Currently - this bench is useless without context.

[PCGuy_5960]

1 minute ago, don_svetlio said:

I did mention the possible lack of firmware optimization, did I not?

I didn't notice that.... Sorry

[randomhkkid]

Also remember this is almost certainly an engineering samples so non-final clockspeeds, microcode, software optimisation etc. If anything this just tells us that AMD has some working silicon

[don_svetlio]

2 minutes ago, PCGuy_5960 said:

I think power consumption... As they are trying to compete with Intel and their lower TDPs

This is certainly a server setup and TDP does not matter at all there. Hence intel's hot 22-core Xeons.

[PCGuy_5960]

Just now, don_svetlio said:

This is certainly a server setup and TDP does not matter at all there. Hence intel's hot 22-core Xeons.

Idk... Is there any other reason to run the CPUs at just 1.4GHz??

Because that is the weirdest part here...

[don_svetlio]

Just now, PCGuy_5960 said:

Idk... Is there any other reason to run the CPUs at just 1.4GHz??

Because that is the weirdest part here...

Making sure the whole platform is stable before setting the clockspeeds and TDP targets?

[Vode]

6 minutes 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?

32 energy efficient cores @ 1,4 Ghz might be very appropriate for server workloads... Also the final clockspeeds could be around 2Ghz maybe, this is still an engineering sample..

 

 

[PCGuy_5960]

Just now, don_svetlio said:

Making sure the whole platform is stable before setting the clockspeeds and TDP targets?

Maybe that is the case....

[PCGuy_5960]

Just now, Vode said:

32 energy efficient cores @ 1,4 Ghz might be very appropriate for server workloads... Also the final clockspeeds could be around 2Ghz maybe, this is still an engineering sample..

We are all hoping that this is not indicative of the final product.... Because if it is, then AMD is kind of doomed, as noone overclocks Xeons/Opterons

[don_svetlio]

Just now, PCGuy_5960 said:

We are all hoping that this is not indicative of the final product.... Because if it is, then AMD is kind of doomed, as noone overclocks Xeons/Opterons

They'd have to be really drunk or really stupid to release a new flagship that performs worse than their old flagship.

[patrickjp93]

15 minutes ago, don_svetlio said:

Also, keep in mind that Geekbench may be registering AMD's version of hyper threads as cores. So those may actually be 16 cores per CPU, not 32.

SMT is confirmed disabled. It's the 32-core.

[PCGuy_5960]

Just now, don_svetlio said:

They'd have to be really drunk or really stupid to release a new flagship that performs worse than their old flagship.

I know.... I am just saying that they'd better not release something that runs at 1.4GHz.... It should be 2.5+GHz and then it will completely crush the competition...

[Herman Mcpootis]

31 minutes ago, don_svetlio said:

They'd have to be really drunk or really stupid to release a new flagship that performs worse than their old flagship.

i think they already knew what will happen if they tried it again, last thing they need is another bulldozer