• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Pascal Titan-X Launch

Not if you already had 12GB of it before, no.


12GB of 10Ghz GDDR5X on a 384-bit bus *is* overkill.

It's not 16GB of HBM2 overkill, but still overkill nonetheless.


Well we will have to agree to disagree. You think this is the real full fat Titan and I think this card could have very well of been the Ti re-badged as a Titan.
 
GP100 is not suited to gaming.

That second link is incorrect in assuming only difference between GP100 and GP102 is how it links to motherboard.[/Q


There are 2 versions of GP100 1st (Tesla P100) 2nd(GP100) the 1st is not suited to gaming as part of the chip would be sat there doing bugger all, but the second one is.

Both chips are have 15.3 billion transistors ok which is 600mm BUT the Tesla one has stuff not needed for gaming hence why it it has 3584 FP32 CUDA Cores

The second version has the tesla crap removed and is filled with more gaming cores to make it 3840 FP32 CUDA Cores . Therefore its a gaming GPU and a full fat one at that, yet to be officially released.
 
What I meant is that there isn't much of a performance increase considering the price so it's a bit of a rip-off. The 1080 is extortionate, but this takes it to a new level as I was expecting at least 40% gain over a stock 1080 if Nvidia want to charge this much.


This is what I meant I now expect to see :D

now feel like my 1080 is a bargain :eek:
 
as much as i love Nvidia cards . now we need AMD to save us all .

How will they do that? The Titan brand is a bit of a niche market and then you have the Ti and below which is more mainstream. and the last big AMD launch pre RX 480 was the Fury X and that came in at a whopping £570 for me to buy in the first week. If Vega is a beast, expect that to be expensive also. Also if AMD can't get the power requirements nailed down, Vega is facing an uphill struggle.
 
That second link is incorrect in assuming only difference between GP100 and GP102 is how it links to motherboard.[/Q


There are 2 versions of GP100 1st (Tesla P100) 2nd(GP100) the 1st is not suited to gaming as part of the chip would be sat there doing bugger all, but the second one is.

Both chips are have 15.3 billion transistors ok which is 600mm BUT the Tesla one has stuff not needed for gaming hence why it it has 3584 FP32 CUDA Cores

The second version has the tesla crap removed and is filled with more gaming cores to make it 3840 FP32 CUDA Cores . Therefore its a gaming GPU and a full fat one at that, yet to be officially released.

No, the P100 just has 256 cores disabled because it's not the full part. the GP100 is just the full official core. They both have lots of FP64 standalone cores.

If they made an FP32 focused chip, they could fit 4864 cores in that die size.
 
I said "at least not until AMD's Vega", if there using failed chips they will also be slower.

Also, these very well could already be failed Workstation chips.


I understood that the new GP102 is going in to these cards and not the GP100 as it is not optimised for gaming with too much of the GP100 aimed at double precision.

If the GP102 yields are anything like the GP104 then it is likely that there will be few very good chips, just look at the supply issues still on the 1080(104) yet we have plenty of 1070(104). I could see Nvidia putting the not so good chips in this Titan X with the best chips which they won't have many of yet being saved for a Titan XX in 4-6 months that may even have HBM2. This way there will be no bad GP102 left for a TI and we may not see one. With no competition from AMD it makes more sense to take this sort of path for Nvidia this time round.
 
Last edited:
So which is it you want to know why its not doubled in one post then in another say you dont expect it.

Titans havent always had Vram jumps,, original titan was 6gb titan black was.. 6gb.
There is no point putting more ram on if its not needed


Typo, I meant increased VRAM on titan. I mentioned 16GB earlier. Titan and Titan Black were the same generation.


This card has the same amount of vram as the last titan the fact that it is GDDR5X is irrelevant. When has a new gen top end card been released with the same amount of vram as the last?

What is more believable? That the new 1080Ti would release with 12Gb of vram and the Titan 16Gb, or that we would see a new Titan with the same amount of Vram as the 18 month old Titan X?
 
Last edited:
The Titan brand is an ongoing farce. Nothing but Halo products. The ti product if it ever arrives is what we're really interested in, but I'll echo what others are saying - if the clock speeds are right, it's hard to see these being more than 20-25% faster than 1080.

Don't expect the 1080TI to be less than £800 when it launches dude.

All AMD have to compete is Polaris, a card that barely manages to compete with years old 970's - there is NO competition, so NVIDIA can do as they please.
 
I understood that the new GP102 is going in to these cards and not the GP100 as it is not optimised for gaming with too much of the GP100 aimed at double precision.

If the GP102 yields are anything like the GP104 then it is likely that there will be few very good chips, just look at the supply issues still on the 1080(104) yet we have plenty of 1070(104). I could see Nvidia putting the not so good chips in this Titan X with the best chips which they won't have many of yet being saved for a Titan XX in 4-6 months that may even have HBM2. This way there will be no bad GP102 left for a TI and we may not see one. With no completion from AMD it makes more sense to take this sort of path for Nvidia this time round.

Remember what they did first time round with this Titan naming scheme?

The original Titan were failed Quadro cards, 2560 Shaders, then AMD introduced the 290X, which beat it, Nvidia's response was to use healthy Quadro SKU's to launch the 780TI, 2880 Shaders.

IMO cheaper 1080TI's depends very much on AMD at least matching these.

If reviewers follow Nvidia's review instructions again we will never see a 1080TI :D
 
Remember what they did first time round with this Titan naming scheme?

The original Titan were failed Quadro cards, 2560 Shaders, then AMD introduced the 290X, which beat it, Nvidia's response was to use healthy Quadro SKU's to launch the 780TI, 2880 Shaders.

IMO cheaper 1080TI's depends very much on AMD at least matching these.

If reviewers follow Nvidia's review instructions again we will never see a 1080TI :D

I agree, I don't expect to see a cheaper TI card unless if AMD can challenge at least the 1080 performance.
 
Back
Top Bottom