Thursday, April 6th 2017
NVIDIA Announces the TITAN Xp - Faster Than GTX 1080 Ti
NVIDIA GeForce GTX 1080 Ti cannibalized the TITAN X Pascal, and the company needed something faster to sell at USD $1,200. Without making much noise about it, the company launched the new TITAN Xp, and with it, discontinued the TITAN X Pascal. The new TITAN Xp features all 3,840 CUDA cores physically present on the "GP102" silicon, all 240 TMUs, all 96 ROPs, and 12 GB of faster 11.4 Gbps GDDR5X memory over the chip's full 384-bit wide memory interface.
Compare these to the 3,584 CUDA cores, 224 TMUs, 96 ROPs, and 10 Gbps GDDR5X memory of the TITAN X Pascal, and 3,584 CUDA cores, 224 TMUs, 88 ROPs, and 11 GB of 11 Gbps GDDR5X memory across a 352-bit memory bus, of the GTX 1080 Ti. The GPU Boost frequency is 1582 MHz. Here's the catch - the new TITAN Xp will be sold exclusively through GeForce.com, which means it will be available in very select markets where NVIDIA's online store has a presence.
Compare these to the 3,584 CUDA cores, 224 TMUs, 96 ROPs, and 10 Gbps GDDR5X memory of the TITAN X Pascal, and 3,584 CUDA cores, 224 TMUs, 88 ROPs, and 11 GB of 11 Gbps GDDR5X memory across a 352-bit memory bus, of the GTX 1080 Ti. The GPU Boost frequency is 1582 MHz. Here's the catch - the new TITAN Xp will be sold exclusively through GeForce.com, which means it will be available in very select markets where NVIDIA's online store has a presence.
143 Comments on NVIDIA Announces the TITAN Xp - Faster Than GTX 1080 Ti
But people had to be special and invent their own name instead of just calling them the Geforce TX and Nvidia TX, and now that's causing issues.
Its like getting mad at your car dealer for releasing a new version of the car you purchased last year. Even though its the same fucking name too. You think i get mad at Audi for releasing an A4 every year? Get over it.
Luckily for AMD fans, patience is a virtue.
I pointed out the fact that Pascal is still a greater improvements than any of the GCN revisions. AMD has not produced a new GPU architecture in many years.
And btw if you can read, I said it should be $400 if it is a little weaker than the 1080 Ti.
-If it's the same strength or 1-2% stronger, it should cost between $550 and $650.
-If it beats the 1080 Ti by a sold 10% or more, if should be $700.
What matter here is its happening either way. Its not going to be much different in performance than the previous Titan XP or new 1080ti. I do wish they would distinguish the names a bit more since it can be confusing especially on the second hand market but whatever.
"Who is buying this card"?
If you bought the Titan X Pascal (2016), then of course you are not going to replace it a few months later, just like Skylake owners are not going to upgrade to Kaby Lake. It's a product refresh with a slightly better product, which is normal in many product segments. If you go buy a 2017 BMW, then you know there is going to be a 2018 model with minor upgrades, and no one goes crazy because of that!
"Why are Nvidia selling this?"
The demand for Titan cards is huge in the professional market, and they want all the performance they can get. The yields now are much better than last year, so Nvidia refreshes the product to provide a faster model. It's pure common sense, and anyone failing to understand this is really incompetent.
This joke claims Nvidia even has a "Titan Ultra" in the works, because it's possible to make bigger 16 nm chips. This kind of argumentation is just crazy.
That being said I'll be first in line for one if it has a short pcb to replace my current cards for a single of equal performance.
AMD does also make multiple SKUs from single chips, and it's been the norm for decades. Please stop acting childish.
Come on, you're smarter than this. GP100 is an even more complex chip and its been around how long now?
Also any comparison to post-Kepler > today has no bearing on the 780ti story because AMD has yet to surpass the 980ti performance wise. There was never any reason to push harder on Maxwell, so we got a Titan X'm', a 980 that almost nobody wanted because grossly overpriced/perf, and a 980ti as a cut down Titan. Maxwell's clocks were enough to stay ahead. Same with Pascal, until AMD drops the Vega Bomb and now we have three versions of GP102.
I see a pattern, you? :P
Even the GP100s used in Teslas have disabled cores and even lower clocks than GP102, and still there is no "perfect" chips out there. Anyone who claims Nvidia is just keeping back chips is simply stupid.
The 780ti is effective proof that Nvidia was definitely holding chips to remarket them as a different product 'when the time was right'. They had serious trouble back then with AMD's offerings, very much unlike today until vega launches.
I do NOT contest that process and yields do improve over time. But it is also business 101 to make the buyer believe that the product at the top of the stack is the 'unique and best' option - scarcity sells. That is entirely what Titan releases are about. And you don't convince buyers with 'we're holding these chips back so we can release another version to eclipse our current halo product'. Again, I point you towards the outrage among Titan owners when 780ti was released.
You DO convince them with 'look at the most powerful and complicated chip in the world, crafted by elves at moonlight, a super delicate process that often fails'. Remember, even their 'FE' coolers are 'meticulously crafted' for 'optimal performance'... meanwhile throttling on 90% of the product stack.