Friday, July 22nd 2016

NVIDIA Announces the GeForce GTX TITAN X Pascal

In a show of shock and awe, NVIDIA today announced its flagship graphics card based on the "Pascal" architecture, the GeForce GTX TITAN X Pascal. Market availability of the card is scheduled for August 2, 2016, priced at US $1,199. Based on the 16 nm "GP102" silicon, this graphics card is endowed with 3,584 CUDA cores spread across 56 streaming multiprocessors, 224 TMUs, 96 ROPs, and a 384-bit GDDR5X memory interface, holding 12 GB of memory.

The core is clocked at 1417 MHz, with 1531 MHz GPU Boost, and 10 Gbps memory, churning out 480 GB/s of memory bandwidth. The card draws power from a combination of 6-pin and 8-pin PCIe power connectors, the GPU's TDP is rated at 250W. NVIDIA claims that the GTX TITAN X Pascal is up to 60 percent faster than the GTX TITAN X (Maxwell), and up to 3 times faster than the original GeForce GTX TITAN.
Add your own comment

162 Comments on NVIDIA Announces the GeForce GTX TITAN X Pascal

#51
jabbadap
Uhm, only 12GB of memory? So no clam shell mode like titan has always be? I'm a bit afraid if vega is not powerful enough next high end nvidia card will be more castrated gp102 with ~3000 cuda cores and 320bit 10GB memory called gtx 2080...

EDIT: Oh and titan X will only be available direct from nvidia like gtx 1060 FE. No AIB versions.
Posted on Reply
#53
jabbadap
btarunr, post: 3493093, member: 43587"
Both GTX 780 Ti and GTX 980 Ti were launched as response to AMD products. GTX 780 Ti was launch as response to R9 290X, and GTX 980 Ti was launched to preempt R9 Fury X. So unless AMD has something that beats GTX 1080 and gets-close-to/outperforms GTX TITAN XPascal, NVIDIA won't launch GTX 1080 Ti. If AMD's response is slower than TITAN X Pascal, GTX 1080 Ti will have fewer CUDA cores. If it's faster then NVIDIA will max out 3,840 CUDA cores present on the GP102. If it's way faster, then NVIDIA will tap into GP100. So no, GTX 1080 Ti probably won't launch till Vega10 is close, and it's not.

At this rate GP100 won't make a market debut under GeForce brand this generation. It will be shelved for the GeForce 20 series.
I doubt GP100 will ever make debut to consumer market, prosumer maybe dubbed as quadro P6000. Why do you even think it would, geforce does not need fp64 compute and gp102 and gp100 has same amount of fp32 cuda cores(thus I would suspect gp102 die size is smaller than gp100). I doubt that memory bandwidth alone will make it that faster, and in the future there will be faster gddr5x memories too.

Oh and it's Nvidia Titan X. Nvidia don't market it as Geforce gtx anymore. Not that it mean much, but maybe they try to differentiate it from consumer products to more prosumer/consumer product. Well niche it's anyway and most probably very limited unit count overall.
Posted on Reply
#55
GC_PaNzerFIN
Oh my, lack of competition is very disturbing.
Posted on Reply
#56
bug
GC_PaNzerFIN, post: 3493283, member: 76628"
Oh my, lack of competition is very disturbing.
True, but who's AMD going to compete with first?
Because this is where AMD stands right now:
And if you think about the mobile market, Qualcomm is worth almost $100Bn... I don't know who would want to be in AMD's place right now.
Posted on Reply
#57
P4-630
Does it play Crysis at 1080p FHD? :D:p
Posted on Reply
#58
Cybrnook2002
So if this is GP102, then there is still another monster waiting in the background with full fledged GP100? Something higher than Titan X with HBM2?
Posted on Reply
#59
bug
Cybrnook2002, post: 3493292, member: 44647"
So if this is GP102, then there is still another monster waiting in the background with full fledged GP100? Something higher than Titan X with HBM2?
Most likely not. The GP100 seems to be reserved to Tesla line, because its design is inherently geared towards compute. With Pascal, Nvidia finally made different designs for Tesla, rather then simply enabling/disabling various blocks. Plus, the 1070 and 1080 will have the market to themselves for about 9 months, there's no need for anything faster.
Posted on Reply
#60
Nobody99
NVIDIA must think people forget easily, the audacity they had when they released the first Titan for 1000$ and it wasn't even fully unlocked and the audacity now thinking people would consider buying such a card after such a legacy that befalls the Titan name.

I might remotely considering buying it if they wouldn't have used the noun Titan in vain.
Posted on Reply
#61
iO
So, not a full 3840 shader part because of yield issues or an upcoming Black Edition milking part?
Posted on Reply
#62
GC_PaNzerFIN
iO, post: 3493299, member: 107457"
So, not a full 3840 shader part because of yield issues or an upcoming Black Edition milking part?
Yield and power/thermals are driving factor always. I don't think NVIDIA reference cooler can handle full GP102 at reasonable clocks, and if the part clocks higher with less cores compensating the difference while improving yield considerably, then why not do it.
Posted on Reply
#63
mcraygsx
Prima.Vera, post: 3493157, member: 98685"
Yeah, but 3dFX was EXACTLY in the same position AMD is now. Their new GPU release was slower and more power hungry than nVidia's, for 2 consecutive generations, just like AMD now... On 3rd strike they got bankrupted. Personally, I see AMD going into an a similar way if they don't release something good this year...
Time is catching up with AMD or vice versa. They have been playing catch up game since last two gens and finally released a new architecture that competes with 980's.

NVidia could have utilized HBM2 on TITAN Pascal.
Posted on Reply
#64
Ubersonic
bug, post: 3493288, member: 157434"
I don't know who would want to be in AMD's place right now.
Cyrix :P
Posted on Reply
#65
ViperXTR
Nokiron, post: 3493176, member: 103000"
The title is wrong though, its just called the Titan X. They skipped the GTX and Geforce-moniker on this.
I just realized it lel, so Old GeForce GTX Titan X and new nVidia Titan X it is then, still valid i guess?
Posted on Reply
#66
alucasa
1199? I blame Brexit for that.
Posted on Reply
#67
matar
Great Announcement GTX Titan X Pascal BUT @ $1200 with 12GB they should have given it at least 16GB + it's way too high come on no GeForce GPU should be over $1000 they gave it this price so they can sell the GTX 1080 Ti @ $999.
Great GPU But a Bad price...
Posted on Reply
#68
Prima.Vera
Crap Daddy, post: 3493246, member: 89540"
Titan X was announced at Stanford Uni at an AI meeting, deep learning stuff. It's the same approach as with the other Titans, offering a more affordable tool for productivity, research, scientific use while enticing the tech enthusiasts, gamers who can afford ultra high end products. Nothing wrong with the price, it's tried and tested. You don't buy a Lamborghini if you need a car to get you to your average income job place.
Car comparison again... You cannot do that man! Your Lamborghini is not becoming obsolete after 1 year, common!
Posted on Reply
#69
alucasa
Prima.Vera, post: 3493379, member: 98685"
Car comparison again... You cannot do that man! Your Lamborghini is not becoming obsolete after 1 year, common!
Yeah, it's always cars. They can't get over it. You'd think the Internet people would have more creativity.
Posted on Reply
#70
dozenfury
Seems pretty underwhelming on those specs, especially at $1200. But I'll be curious to see benchmarks. SLI'ing 2 1080s for basically the same cost and maybe 50w more power usage would seem far faster and more appealing, even with the headaches of SLI. Or better yet just wait for 1080ti.

But I'm sure there are plenty of people that will jump on this for the bragging rights, or high-profile Twitch streamers where it's just kind of a business cost.
Posted on Reply
#71
GhostRyder
Ugh, that price is OP... Supposedly there is going to be a 12gb and an 16gb edition but that was just rumored though it would be interesting to know if they do that where both will be priced. Well seems a bit quick for this release, wonder how soon GTX 1080ti will follow.

Only selling these from Nvidia's website from now on...Guess they don't want to let OEM's release hybrids or waterblock editions of the card anymore...

Hmm, question now for me is how close will the 1080ti be to BF1 or if this will be the only one of the top chips available...
Posted on Reply
#72
Captain_Tom
Wait.... 60% faster than 1070? Isn't the 1080 already 20-30% faster than that?

That makes this card only 30% stronger than the 1080! Good lord these Titan cards keep getting more and more pathetic.

No worries that Vega will crush this anymore...
Posted on Reply
#73
bug
matar, post: 3493375, member: 96547"
Great Announcement GTX Titan X Pascal BUT @ $1200 with 12GB they should have given it at least 16GB + it's way too high come on no GeForce GPU should be over $1000 they gave it this price so they can sell the GTX 1080 Ti @ $999.
Great GPU But a Bad price...
Captain_Tom, post: 3493403, member: 155434"
Wait.... 60% faster than 1070? Isn't the 1080 already 20-30% faster than that?

That makes this card only 30% stronger than the 1080! Good lord these Titan cards keep getting more and more pathetic.

No worries that Vega will crush this anymore...
This is not a gaming card, it's pointless to compare it the GTX 1080.
Fwiw, even Nvidia isn't sure what this is, because now it's officially not part of the GeForce line and it was never part of Quadro or Tesla lines either. I've always viewed Titan cards as a "look what we can do" card that I didn't care much about.
Posted on Reply
#74
Slizzo
bug, post: 3493296, member: 157434"
Most likely not. The GP100 seems to be reserved to Tesla line, because its design is inherently geared towards compute. With Pascal, Nvidia finally made different designs for Tesla, rather then simply enabling/disabling various blocks. Plus, the 1070 and 1080 will have the market to themselves for about 9 months, there's no need for anything faster.
Ehhh... one point of contention, P100 is reserved to Tesla line, nothing has been shown about a GP100.

But yeah, consumers will probably only ever see the GP102.
Posted on Reply
#75
beck24
Assimilator, post: 3493125, member: 7058"
Jaysus. 40% more CUDA cores, 40% more TMUs, 50% more ROPs, 50% more memory bandwidth, 50% more memory... this thing is gonna be a monster. And that means 1080 Ti will be even faster.

Seems NVIDIA is out to crush AMD absolutely this round. They could probably have got away with only releasing the GTX 1080 and 1070, but we also have the 1060 that rains on RX 480's parade, and now the TITAN X Pascal which is even faster than the 1080. If Vega is as much of a disappointment as Polaris was, AMD is in big trouble.
Excited to see reviews, oc, etc.
Posted on Reply
Add your own comment