http://wccftech.com/nvidia-volta-gv 100-gpu-fast-pascal-gp10

>muh compute
Why does it matter to you you fucking retard? What are you thanking them for?

Every generation the new GPU gets hyped and by the time it comes out, the hype is dead and at best people are "that's kinda neat"

>>58090678
Why is this a surprise?

>>58091015
Kepler was pretty hype on release.
70% gains on fermi while reducing power draw was awesome.

>>58090678
Holy shit this is all they have to show for a Volta reveal? Nvidia is fucking finished.

Wow, so it's 50% faster than last gen? Whoop.

hilarious that HBM is already a dead end.

>>58091545
HBM, another AYYMD engineering failure like their CPUs & GPUs

>>58091545
Hynix is just slow at HBM3 or whatever they'd call it.

File: 1471808494751.jpg (90KB, 604x453px) Image search: [Google]

90KB, 604x453px

>>58090678
>>58091581
Why you guys using ayymd when talking bad about amd? I mean, amdrones using this "word" when talking good about their company and your statements looks weird because of that..

>>58091545
>HBM is already a dead end
howso?

File: 1462899337100.gif (522KB, 255x255px) Image search: [Google]

522KB, 255x255px

>>58090678
>9.5 TFLOPs FP64 Compute


Holly fuck!

File: 1473291918535.png (123KB, 785x757px) Image search: [Google]

123KB, 785x757px

>>58090678
>wccftech

>>58092162
> triple what a Hawaii from 2013 could do, in 2018, holly fuck!

Raw FLOPS aren't everything, or else GP100 would've used more space on ALUs and less on registers and IO and easily gotten beyond the 4-5 fp64 TFLOPS it has.

Both Volta and real Pascal (GP100, not the consumer Maxwell shrinks) need to show that they actual architectural refinements that matter, but they're both vaporware at the moment.

File: georgejefferson.gif (3MB, 320x240px) Image search: [Google]

3MB, 320x240px

>>58090678

but will it play Crysis?

>>58091191
Kepler is shit now.

>>58090678
Remember when nvidia launched GP100 in march this year and they are still not shipping in products?

>>58092815
How would you know

>>58092843
are you a moron? they are not for sale anywhere, and the only question is where they exist at this point.

> Nvidia's labs only
> Nvidia's partners' labs
> reject bins in Taiwan

>>58090678
Sounds more like a huge unstoppable housefire.

>>58092878
>are you a moron? they are not for sale anywhere, and the only question is where they exist at this point.

dgx1 has been shipping for ~8 months. just because a consumer can't buy an individual p100 right now does not mean they don't exist.

>>58092932
they're still in the pre-order phase for even so-called "elite solutions providers"

> http://www.colfax-intl.com/nd/solutions/nvidia-dgx-1-deep-learning-system.aspx

there may be a handful of cobbled together engineering samples floating about, but it's not a real product yet.

File: 1481733438520.png (201KB, 722x862px) Image search: [Google]

201KB, 722x862px

>>58090678
>i fell to the Pascal meme

>>58093236
>they're still in the pre-order phase for even so-called "elite solutions providers"

lol no, you can order them directly from nvidia right now. not sure why you would even view some shady reseller as valid over nvidia themselves unless you have some sort of agenda to push.

>>58093262
>New generation is announced
>People who buy/bought current generation realize there'll be better

Like pottery

>>58093331
You are an shill and you are wrong.

File: dgx1_sales.png (356KB, 993x1330px) Image search: [Google]

356KB, 993x1330px

>>58093331
>some shady reseller
lmao, I seriously doubt you've ever done any sort of enterprise procurement before.

there are handfuls of DGX-1s floating around in prestige exhibitions, but if you can't help Nvidia make better press releases, your $130k/unit will only get you a spot on a waiting list.

>>58093587
>there are handfuls of DGX-1s floating around in prestige exhibitions, but if you can't help Nvidia make better press releases, your $130k/unit will only get you a spot on a waiting list.

explain the hundreds of businesses and universities that already have them then?

>>58093731

> https://www.technologyreview.com/s/603075/the-pint-sized-supercomputer-that-companies-are-scrambling-to-get/
> The Pint-Sized Supercomputer That Companies Are Scrambling to Get
> December 14, 2016
> ...
> Fewer than 100 companies and organizations have bought DGX-1s since they started shipping in the fall, but early adopters say Nvidia’s claims about the system seem to hold up.

Again, as of last week, the number of these things in the wild still number in the high dozens, and they're just in scattered POC/R&D testbeds. They never shipped in the Spring as you claimed earlier, and it's still a near paper launch as of today. GP100 was never intended for a mass launch until 2017, and you're an imbecile or a shill to claim otherwise.

>>58093958
>Again, as of last week, the number of these things in the wild still number in the high dozens,

nothing about your source claims that, and for nvidia to hide such a thing would require lying on their quarterly reports, which is a far greater risk than just admitting to a supposed shortage.

>They never shipped in the Spring as you claimed earlier,

you mean summer, which is exactly when NVIDIA said they would begin shipping dgx1.

>GP100 was never intended for a mass launch until 2017,

gp100 was mass launched with the nvlink p100s included in dgx1, the pcie variant is coming in q1.

> you're an imbecile or a shill to claim otherwise.

not agreeing with your conspiratorial irrational thoughts =/= shilling.

>>58094014
you are retarded

File: 1439761242786.png (71KB, 324x392px) Image search: [Google]

71KB, 324x392px

>>58093350
I keep falling to /g/ memes. /g/ said pascal was going to be 10x faster than maxwell, that DDR4 was going to increase my frames and that Linux is better for gaming.