Nvidia Pascal & Volta Compute Performance Figures Released, HBM2 Power-Hungry?

[Stefan1024]

Nice to see sone actual numbers, I was allways interested in how the power consumption splitts between GPU, VRAM and VRM.

But 60 watts for the memory, are you kidding me? There must be a more efficient solution, or we just ask for to much performormance.

Also with the same 250-300 watts board power like we have no but less die size, the power desity well be aweful.

But I'm counting for the wisedom of AMD here. Give up 10-20% performance to save up to 40% energy (I don't remember the exact numbers but you get the point).

So finally I can do a passive PC even with the new generation of GPUs.

[patrickjp93]

Makes you wonder why so many people don't like him. Hmmmm

The only people who piss off both camps are those who remain completely objective and call out the bullshit on both sides.

[don_svetlio]

The only people who piss off both camps are those who remain completely objective and call out the bullshit on both sides.

Or those who present theories as fact with the sole purpose of causing unrest

[patrickjp93]

Or those who present theories as fact with the sole purpose of causing unrest

Your issue is you actually believe your own delusions.

[don_svetlio]

Your issue is you actually believe your own delusions.

My issue is 75% of your claims either have no source or use yourself as a source and in both cases that is not plausible. And I also find it odd how you failed elementary school math a while back. Supposedly a guy as smart as you would be able to do basic multiplication properly. Alas, that is not the case.

[patrickjp93]

My issue is 75% of your claims either have no source or use yourself as a source and in both cases that is not plausible. And I also find it odd how you failed elementary school math a while back. Supposedly a guy as smart as you would be able to do basic multiplication properly. Alas, that is not the case.

Where did I come up with faulty math?

 

Second, me being my own source is valid in the cases I've done that, because math is universally true.

[don_svetlio]

Where did I come up with faulty math?

 

Second, me being my own source is valid in the cases I've done that, because math is universally true.

When calculating memory bandwidth? When we were talking about the 960 and 380 memory interfaces? In your claims the 380 suddenly lost 20% of the memory bandwidth and the 960 gained 20% and suddenly your claim that they were close seemed reasonable. That is, until the spec sheets of both proved you wrong.

[patrickjp93]

When calculating memory bandwidth? When we were talking about the 960 and 380 memory interfaces? In your claims the 380 suddenly lost 20% of the memory bandwidth and the 960 gained 20% and suddenly your claim that they were close seemed reasonable. That is, until the spec sheets of both proved you wrong.

My calculations were correct actually. While I admit I did mess up one of the inputs (GDDR clock iirc), the math was never wrong.

[don_svetlio]

My calculations were correct actually. While I admit I did mess up one of the inputs (GDDR clock iirc), the math was never wrong.

I guess that's still something.

[AnnoyedShelf]

Why is this not tagged [WCCFTECH].  They're notorious liars.

[patrickjp93]

Why is this not tagged [WCCFTECH].  They're notorious liars.

Because their article is a summary of Nvidia's presentation at ISC which is linked in their article and which I linked in the OP you biased buffoon. http://images.nvidia.com/events/sc15/SC5125-energy-efficient-architectures-exascale-systems.html

[AnnoyedShelf]

Because their article is a summary of Nvidia's presentation at ISC which is linked in their article and which I linked in the OP you biased buffoon. http://images.nvidia.com/events/sc15/SC5125-energy-efficient-architectures-exascale-systems.html

Okay, my bad. Didn't catch that. No need to resort to name calling.

[patrickjp93]

Okay, my bad. Didn't catch that. No need to resort to name calling.

I'm sorry but attacking the source is always a fallacy. Look at the merits as they are. As much as academics get portrayed as snobs, it's only because we end up jaded dealing with people who seem to have no capacity to stop and think before opening their mouths. I have no shame in wearing the crime of being blunt on my sleeve.