** The Official Nvidia GeForce 'Pascal' Thread - for general gossip and discussions **

[Rroff]

Samsung have 1.4, 1.6 and 2GHz points for HBM2 IIRC.

EDIT: Its 1, 1.6 and 2 Gbit/s per pin actually for the JEDEC standard - got a bit confused.

 

[Orangey]

Now can we finally stop with the posting of rehash blogspam out of spite?

Those of us who said no Titan this year were right.

 

[Orangey]

Samsung have 1.4, 1.6 and 2GHz points for HBM2 IIRC.

EDIT: Its 1, 1.6 and 2 GB/s per pin actually for the JEDEC standard - got a bit confused.

That's Gb, little b bro.

 

[Rroff]

That's Gb, little b bro.

Was already editing it heh.

 

[Rroff]

I don't think the issue is the Samsung end BTW - everything Samsung have published indicates they are having success with it - heard a few rumours of weird configurations on the nVidia side so sounds like nVidia are having problems with implementing the interposer and/or HBM2 rather than a problem with HBM2 itself.

 

[DmitryD]

High base clocks on Tesla sized chip (with FP64 which is likely to be cut out in consumer chip) is a very good sign!

 

[Rroff]

High base clocks on Tesla sized chip (with FP64 which is likely to be cut out in consumer chip) is a very good sign!

TBH I suspect DM is right that they have a weird situation with regard to where optimal perf/watt works out that might not be the same for smaller variants/consumer chip.

 

[bru]

Those of us who said no Titan this year were right.

Well you might be, come December 31st we will know one way or the other.

 

[JediFragger]

Well you might be, come December 31st we will know one way or the other.

He can't be any wronger than Belmont, who was convinced they'd be here this month

 

[bru]

I see that the anandtech article says that this is only 56 out of 60 SM's so a full chip would actually have 3840 cores.

Powering the Tesla P100 is a partially disabled version of NVIDIA's new GP100 GPU, with 56 of 60 SMs enabled.

http://www.anandtech.com/show/10222/nvidia-announces-tesla-p100-accelerator-pascal-power-for-hpc

 

[bru]

Well I have certainly said that I would like them to be here this month, I might have even said that I expect them to be here this month, but to the best of my knowledge I certainly have not said they would definitely be here this month. Anyway this month has only just started.


For all we know NVidia might be planning to call the new GP104 Titan, that would make Orangey wrong, we would all be laughing at NVidia for a stupid marketing idea, but Orangey would still be wrong, unless of course that doesn't arrive till after 31st December. the point being the end of the year is still a good way off, we have no idea what will happen between now and then.

 

[mahius]

A disappointing start to GTC... but it's not over yet. They could be saving the good stuf for last. Fingers crossed. That or AMD has this year in the bag, this lack of consumer Pascal news is worrying.

 

[bru]

A disappointing start to GTC... but it's not over yet. They could be saving the good stuf for last. Fingers crossed. That or AMD has this year in the bag, this lack of consumer Pascal news is worrying.

That is a good point and one we shouldn't forget, GTC is a five day event, who knows what other info will come out of it.

 

[Tommy_Here]

Very disappointing to say the least, might as well get a 980 Ti if you're waiting

 

[Dicehunter]

I can hear Kaapstads wallet screaming now "No...Please...Not again, You've taken everything from me man, No more man.....No more, It's just too much !!!!"

 

[Kaapstad]

I can hear Kaapstads wallet screaming now "No...Please...Not again, You've taken everything from me man, No more man.....No more, It's just too much !!!!"

Not likely.

No big Pascal or Polaris, no purchase.

 

[Dicehunter]

Not likely.

No big Pascal or Polaris, no purchase.

Good man, Same here, IMO this is going to be another GM204 situation, Good performance but I'd rather wait on the proper "Ti" variant

 

[TheF34RChannel]

If the max Cuda cores is 3840 that's not really impressive... Let's make a Volta thread

 

[drunkenmaster]

I see that the anandtech article says that this is only 56 out of 60 SM's so a full chip would actually have 3840 cores.




http://www.anandtech.com/show/10222/nvidia-announces-tesla-p100-accelerator-pascal-power-for-hpc

Uh oh, is this 480gtx all over again, randomly announce/launch it early yet only actually promise real supply for Q1 the next year and for a more expensive market and even then it's a salvaged and not a full part.

Looking it up the Quadro 5000 had another SM disabled compared to the 480gtx but had higher clocks, 225W TDP compared to the 480gtx 250W, keep in mind that at the time the 5870 was a 188W card.

Nvidia announce the card, salvaged and already 300W and list actual availability 9 months later, really doesn't sound great. Figured they learned about large cores early on a process from 40nm as they avoided the problem at 28nm(1-2 process nodes after the rest of the industry knew not to do it).

Would Nvidia release an over 300W consumer card? 2 more SMs enabled but lower clocks, same or more SMs disabled, lower clocks and trying to hit 250W?

This is coming across as more and more of a stunt launch to beat AMD to the punch(considering they already announced Pascal as the first 3d memory card back in 2014 CES, it wouldn't be new) and they have to respin GP100 and we'll have GP110 by some point in 2017.

 

[nukemdukem]

so happy i got my GTX980 on offer - will see me through another 12 months