Honestly, a bit surprised to see them releasing the Titan this early. 1080s are still hard to find and they could have waited another month or two until supply met demand. With this move they are missing out on some people that would have double dipped, I think.
It makes me wonder if the Vega rumor for Oct had some merit, and NV had to move up their timeline. Release the Titan now, and then in a few months release the Ti as a competitor to Vega.
VEGA has nothing to do with this launch.Honestly, a bit surprised to see them releasing the Titan this early. 1080s are still hard to find and they could have waited another month or two until supply met demand. With this move they are missing out on some people that would have double dipped, I think.
It makes me wonder if the Vega rumor for Oct had some merit, and NV had to move up their timeline. Release the Titan now, and then in a few months release the Ti as a competitor to Vega.
P100 has been in the works for awhile. We've known about it at least since April. The earlier than anticipated launch could mean all sorts of things:You have me wondering now. I don't understand it either.
As far as I can tell you're right that it would take the 1080ti to compete with Vega and the current 1080 is inadequate for that purpose. I think they just release the Titans on schedule to extract some gravy from those who can't or won't wait for the Ti.
P100 has been in the works for awhile. We've known about it at least since April. The earlier than anticipated launch could mean all sorts of things:
- Yields are better than expected.
- Nvidia hasn't sold as many as they thought to the HPC market.
- Paper launch.
We'll see how it pans out.
Yeah I heard the same story not long ago...Volta for consumers in 2017 at Vega's time-frame? Are you nuts?
We will see in time what will happen with Volta but Vega won't be there in October, not even for Xmass. You can quote
Before I get too excited - Titan cards are only ever released as reference cards right, so there's no aib pcb improvement possible. Have they traditionally made sense for custom watercoolers looking for an extreme build?
My thoughts exactly were along the same lines. There's also a talk of refresh with the HBM2. Now, with that in mind, the real kicker is that either Volta proper the new architecture chip is delayed, or Volta is just further tweaked Pascal, which itself was tweaked Maxwell.It's not even the fully enabled GP102 chip, it's only 40% more cores than GTX 1080. Which means the 10xx series will be fairly short lived like the GTX 600 series was, and Pascal will get a rebadge / refresh with faster vram and different core clocks.
EDIT: 34% higher theoretical will probably translate into 30% better performance than GTX 1080. JHH said it is 60% faster than the previous Titan X, which lines up exactly.
Sell them.
Slightly used and at a great price.
They can't have GDDR5x & HBM2 on the same chip, from what I understand, because you'll need two MC's & at that point it'd be smarter (& cheaper) to just go with HBM2.My thoughts exactly were along the same lines. There's also a talk of refresh with the HBM2. Now, with that in mind, the real kicker is that either Volta proper the new architecture chip is delayed, or Volta is just further tweaked Pascal, which itself was tweaked Maxwell.
I knew it!:biggrin:I'll take two please.
Who do you think its targeted for? As stated in earlier comment, if HPC and deep learning it could be a bargain. What can AMD do for them here?What a joke of a card. Overpriced garbage like this is what people get for years of people refusing to consider AMD and blindly buying nVidia.
Rip off card from a rip off company.
This is what happens when you let one company get to 80% market share. Record revenues and record profits fuel even more R&D and thus here we have a GP102 for consumer with G5X and a GP100 with HBM2 for HPC/workstation. AMD does not have the resources to compete. Plain and simple. I think AMD will never again be able to bring back competition like the HD 5000 / HD 6000 /HD 7000 / R9 200 days. The gap which started with Maxwell is going to grow further and AMD are going to have a hard time getting back to historical market share levels. As for the >300 market Nvidia has no contest there for atleast 6 months. I doubt Vega has a chance to compete given the enormous time to market lead of Pascal GP104/ GP102. Nvidia will rake in the money and be ready for price cuts if needed when Vega launches. Kudos to Nvidia on their GPU dominance.
Who do you think its targeted for? As stated in earlier comment, if HPC and deep learning it could be a bargain. What can AMD do for them here?
https://blogs.nvidia.com/blog/2016/07/21/titan-x-deep-learning/
That wont prevent gamers from snapping them up, but they dont think of bargains in that price range.. and AMD has nothing in this area either.
This is what happens when you let one company get to 80% market share. Record revenues and record profits fuel even more R&D and thus here we have a GP102 for consumer with G5X and a GP100 with HBM2 for HPC/workstation. AMD does not have the resources to compete. Plain and simple. I think AMD will never again be able to bring back competition like the HD 5000 / HD 6000 /HD 7000 / R9 200 days. The gap which started with Maxwell is going to grow further and AMD are going to have a hard time getting back to historical market share levels. As for the >300 market Nvidia has no contest there for atleast 6 months. I doubt Vega has a chance to compete given the enormous time to market lead of Pascal GP104/ GP102. Nvidia will rake in the money and be ready for price cuts if needed when Vega launches. Kudos to Nvidia on their GPU dominance.
Come on man, the article you are quoting has it being built in 2017 and not finished til 2018.
Volta for consumers in 2017 at Vega's time-frame? Are you nuts?
Look at the last flagships, lots of you guys said Fiji wouldn't be possible at sub 300W, some people thought 4096 SP would be 450W based on Hawaii.. etc. It's just bad prediction.
Fiji wasn't far behind GM200 and that's before DX12/Vulkan took off.
2 years? Don't even need that long. In 3 months, the next wave of DX12 titles are here. In 6 months, most of the AAA games are DX12/Vulkan. How many games are in most review benchmarks? Only a handful. Unless you feel Metro or Crysis 3 is going to remain relevant in 2017, there's every chance Vega will be more than competitive vs GP102.
Its amazing to me that people are more willing to shell out tons of money for new hardware instead of pushing developers to optimize for existing hardware. Its insane.
The funny thing is Vega could end up being $800 and it will look like a bargain. How long until Titan is $2000? But hey if people keep buying them good for Nvidia.
FWIW, I expect a 16nm consumer Volta lineup in 2017 then 10nm shrinks in 2018.