New Titan X announced, 12GB GDDR5X, $1200

Status
Not open for further replies.
I honestly have no idea what you just wrote lol. I would love to read up on some documentation/whitepaper if you could link me to any thanks.

That was a refreshing post, I salute you.

Unlike a lot of posts on this forum were the posters do their best to display their ignorance, you actullly admitting to being at the end of your knowlegede and asking for links to read up and educate yourself.

A lot of posters could learn from you.
 
I honestly have no idea what you just wrote lol. I would love to read up on some documentation/whitepaper if you could link me to any thanks.

I believe those are internal designations,, but Ryan Smith from Anand confirmed the gp104/102/100 configuration
 
GP102 also launched as the Quadro P6000 with 24GB GDDR5X and a full working GP102 die. P5000 launched as a full GP104 die with 16GB GDDR5X.

Another example that HBM2 is only for the highest of the highest highend.
 
GP102 also launched as the Quadro P6000 with 24GB GDDR5X and a full working GP102 die. P5000 launched as a full GP104 die with 16GB GDDR5X.

Another example that HBM2 is only for the highest of the highest highend.
Can hardly call it a launch, more like announcement. But yeah, HBM2 probably won't come to enthusiast tier until after Volta.
 
Can hardly call it a launch, more like announcement. But yeah, HBM2 probably won't come to enthusiast tier until after Volta.

I don't blame them, keep the higher end low volume stuff for the higher margin professional market and then as the process tech improves and the yields increase bring it to the plebeian masses.
 
GP102 also launched as the Quadro P6000 with 24GB GDDR5X and a full working GP102 die. P5000 launched as a full GP104 die with 16GB GDDR5X.

Another example that HBM2 is only for the highest of the highest highend.

Mainly for DP, it'll be interesting to see if they also produce a smaller version of GP100 for Tesla, would that also use HBM ?

Or just overrated..

Not really, just people conflating raw bandwidth with performance
 
Or just overrated..

I personally don't think it's overrated if the footprint reduction alone meant you could potentially have Titan X performance in a Nano package. If nothing else it'll certainly be very popular with SFF builders.

Oh and it'll be an absolute blessing for high performance laptops (you might laugh, but they're very much a thing, and there's a market for them).
 
When is the 1080 Ti out which will spit in the faces of people purchasing the Titan X?

Should be out by late November is my guess. That's the card I'm waiting on or possibly Volta since nvidia pushed it up to May 2017. I was considering the new Titan X but $1200+ tax just doesn't sit right with me for an air cooled thermally constrained card.
 
If it helps it's also a cut down chip. :D (only 3584 cores out of a possible 3840)

NVIDIA Announces Quadro Pascal Family: Quadro P6000 & P5000

On a quick technical note, as this announcement comes just 4 days after NVIDIA announced the GP102 GPU used on this card, this Quadro announcement does confirm a few more things about GP102. Quadro P6000 ships with 3840 CUDA cores (30 SMs), confirming our earlier suspicions that GP102 was a (or at least) 30 SM part. Meanwhile this also confirms that GP102 can be outfit with 24GB of GDDR5X. Finally, NVIDIA has confirmed that there’s no high-speed FP64 support on GP102, which is why we’re looking at a 1/32 rate for even the top Quadro card.

So not only is this Titan X not a full chip, it doesn't even have twice the vram either. And on top of that Leatherman wants to charge $200 more for it. :rolleyes: Talk about a triple fail.

If Volta is indeed coming in May 2017, I would just skip Pascal entirely, because you just know even GV104 is going to beat the shit out of GP102.
 
If Volta is indeed coming in May 2017, I would just skip Pascal entirely, because you just know even GV104 is going to beat the shit out of GP102.

For a Geforce Volta you are most likely into August. There is always better performance if you wait. But you dont have any better performance in the meanwhile. Just wait for Volta successor on 10nm, this will beat the shit out of GV104 etc ;)
 
If it helps it's also a cut down chip. :D (only 3584 cores out of a possible 3840)

NVIDIA Announces Quadro Pascal Family: Quadro P6000 & P5000



So not only is this Titan X not a full chip, it doesn't even have twice the vram either. And on top of that Leatherman wants to charge $200 more for it. :rolleyes: Talk about a triple fail.

If Volta is indeed coming in May 2017, I would just skip Pascal entirely, because you just know even GV104 is going to beat the shit out of GP102.

Like others have said, the new Titan X comes off as a Ti part. I think we are being gouged much more than $200. Somewhere around $400 sounds about right.
Nvidia will continue to offer high end variants to milk the high end. Don't fall for it and skip this generation.
 
For a Geforce Volta you are most likely into August. There is always better performance if you wait. But you dont have any better performance in the meanwhile. Just wait for Volta successor on 10nm, this will beat the shit out of GV104 etc ;)
Pascal isn't fully using AC and not receiving noticeable performance increases when using new APIs, along with being priced at a premium makes Pascal a loser. I can understand coming from Kepler and older gens that Pascal maybe exciting. But for Maxwell users, there's no reason to upgrade. Even OG Titan X owners can keep their cards and wait it out. Up to 60% over the previous Titan isn't enough. The resell value is around $500, which means it will take $700 upgrade. Paying 140% more ($700) to receive an increase of "up to" 60% in performance is insane.
 
Pascal isn't fully using AC and not receiving noticeable performance increases when using new APIs, along with being priced at a premium makes Pascal a loser. I can understand coming from Kepler and older gens that Pascal maybe exciting. But for Maxwell users, there's no reason to upgrade. Even OG Titan X owners can keep their cards and wait it out. Up to 60% over the previous Titan isn't enough. The resell value is around $500, which means it will take $700 upgrade. Paying 140% more ($700) to receive an increase of "up to" 60% in performance is insane.

AC is not part of the DirectX 12 API, funny how a feature get so much focus, but I guess it takes away the focus from AMD's terrible DX11 overhead...which begs the question:

What would give AMD more performance: DX12 AC or fixed CPU overhead under DX11...my money is on the latter.

But since ignorance is the new back on techforums, the focus is on how much better AMD does compared to DX11 and attributed to DX12 + AC...pure win.
While more technical people look at the DX12 performance and thinks (Now this suit the theoretical SKU performance, so what is the BUG under DX11 holding these cards back there?!).
 
Guys do you think Nvidia will do most likely the same thing that happened with the 980's? After the Titan a 1080Ti will arrive?

Thanks
 
Guys do you think Nvidia will do most likely the same thing that happened with the 980's? After the Titan a 1080Ti will arrive?

Thanks

Only if NVIDIA feels AMD has a competive card....look at the past...Ti SKU's have always been in response to AMD competition...and have never just been relased to release something.
 
AC is not part of the DirectX 12 API, funny how a feature get so much focus, but I guess it takes away the focus from AMD's terrible DX11 overhead...which begs the question:

What would give AMD more performance: DX12 AC or fixed CPU overhead under DX11...my money is on the latter.

But since ignorance is the new back on techforums, the focus is on how much better AMD does compared to DX11 and attributed to DX12 + AC...pure win.
While more technical people look at the DX12 performance and thinks (Now this suit the theoretical SKU performance, so what is the BUG under DX11 holding these cards back there?!).

How did AMD become apart of your argument? We already know AMD has poor DX11 performance and remained competitive, but that isn't the trend moving forward with DX12.
You are making it seem like you just stated some new information. It still doesn't take away the fact that this $1200 card will be an afterthought in months time, obsolete before it leaves the factory.
 
Should be out by late November is my guess. That's the card I'm waiting on or possibly Volta since nvidia pushed it up to May 2017. I was considering the new Titan X but $1200+ tax just doesn't sit right with me for an air cooled thermally constrained card.
Source on pushing Volta up to May 2017? I hope you have something other than WCCFTech or Fudzilla.
Pascal isn't fully using AC and not receiving noticeable performance increases when using new APIs, along with being priced at a premium makes Pascal a loser. I can understand coming from Kepler and older gens that Pascal maybe exciting. But for Maxwell users, there's no reason to upgrade. Even OG Titan X owners can keep their cards and wait it out. Up to 60% over the previous Titan isn't enough. The resell value is around $500, which means it will take $700 upgrade. Paying 140% more ($700) to receive an increase of "up to" 60% in performance is insane.
Pascal works with async compute just fine. NVIDIA will never use AMD's async shaders because their architecture wouldn't benefit from it. Just because NVIDIA doesn't see 30+% performance gains from async compute doesn't mean that the architecture is somehow deficient.

You're also under the assumption that everyone is purchasing a video card as an investment. I do not purchase video cards (or any PC hardware, for that matter) with resell value in mind. Also, a 60% bump in performance over the previous generation is pretty damned good. I think most people are forgetting that the Mitan X was only around 45% faster than the Titan Black.
 
Once AC becomes a more mainstream thing perhaps Nvidia will bake it in as a hardware feature instead of doing it in software.
 
How did AMD become apart of your argument? We already know AMD has poor DX11 performance and remained competitive, but that isn't the trend moving forward with DX12.
You are making it seem like you just stated some new information. It still doesn't take away the fact that this $1200 card will be an afterthought in months time, obsolete before it leaves the factory.

They are behind the FUD about AC.
Muppets will gobble it up and not ask the inconvenient question (Why AMD's pipeline isn't utilized efficiently without it?) and instead treat AC as something "new and cool" when in reality it only fixes AMD's inefficient pipeline.

AC will never matter to an NVIDIA GPU in the same manner, the pipeline is utilized to near max already...if people do not understand this important bit, people so stop posting and read up.
 
Re: AC it's still very new and games are only just now using it. Nvidia already has the perf lead in the high end and wins more battles than they lose on the lower end of the price spectrum so it's likely a non issue.

I keep thinking that AMDs hardware is more than capable and yet it's their drivers or other software keeping the cards back. The amount of sheer compute from the Red team is amazing. That is to say that compute doesn't mean better FPS just that it's a capable card and I think there's more they could squeeze out of them
 
For a Geforce Volta you are most likely into August. There is always better performance if you wait. But you dont have any better performance in the meanwhile. Just wait for Volta successor on 10nm, this will beat the shit out of GV104 etc ;)

True, but person in question has a Titan X SLI setup, which is plenty of performance already. We also have no idea when 1080 Ti would drop, could be soon, could be much later. If SLI is working fine and he doesn't have the need/desire to go back to single GPU, I see nothing wrong with waiting to see how things pan out.

SAlso, a 60% bump in performance over the previous generation is pretty damned good. I think most people are forgetting that the Mitan X was only around 45% faster than the Titan Black.

To be fair, a 45% bump while remaining on the same node is absolutely nothing to scoff at. In fact in my eyes that's even more impressive because it highlights just how much more efficient nVidia made the architecture.
 
So to upgrade I just need $1200 for a gpu, I'd like the 10 core intel 6950x, thats $1650, a new power supply, case, and motherboard... so about $3500 give or take, and then the time to use it (nonexistent)

Oh wait, forgot ram and a new 4k monitor... Hmmm
 
Last edited:
So to upgrade I just need $1200 for a gpu, I'd like the 10 core intel 6950x, thats $1650, a new power supply, case, and motherboard... so about $3500 give or take, and then the time to use it (nonexistent)

Oh wait, forgot ram and a new 4k monitor... Hmmm

By the time it gets to you, it will be obsolete.
 
Intel revealed the 6950X sold quite well already. So the buyers are out there for the Titan cards as well.
 
So this thing will be 30-40% faster than 1080s?

Slower. 25-30%. But depending on the game you could see 30-40%.
If it was near 50% I would definitely consider Pitan because it would be near 1080 SLI numbers if OC from there.
 
Last edited:
Slower. 25-30%. But depending on the game you could see 30-40%.
If it was near 50% I would definitely consider Pitan because it would be near 1080 SLI numbers if OC from there.
30% faster than 1080 would still be good. This was the performance delta between 980 and Mitan X.

Need to find out if NZXT knows the G10 will fit the Pitan X... They say it fits the 1080, so here's hoping.
 
Can't wait for people to complain about the cooler
There seem to be a few custom Maxwell Titan X cards like from Gigabyte so makes me wonder if the Pascal Titan X will also get a custom version. The Gigabyte site has conflicting photos showing both a reference cooler and a custom Xtreme version like my 980Ti.

This guy has some interesting research about gfx cards: AdoredTV
 
There seem to be a few custom Maxwell Titan X cards like from Gigabyte so makes me wonder if the Pascal Titan X will also get a custom version. The Gigabyte site has conflicting photos showing both a reference cooler and a custom Xtreme version like my 980Ti.

This guy has some interesting research about gfx cards: AdoredTV

When the card was first announced, I read that NVidia will be the only ones making and selling Pascal Titan X's, in an effort to control quality.
 
When the card was first announced, I read that NVidia will be the only ones making and selling Pascal Titan X's, in an effort to control quality.

More like in an effort to bank all the profits! Quality is a non issue with after market card.
 
Status
Not open for further replies.
Back
Top