nVidia X80, X80ti and X80 Titan rumoured Specs

[dalekphalm]

Just now, thekeemo said:

Then your VGA monitors are useless..

...

what?

http://www.newegg.com/Product/Product.aspx?Item=N82E16812225063&cm_re=dp_to_vga-_-12-225-063-_-Product

$12 on NewEgg

[thekeemo]

Just now, dalekphalm said:

...

what?

http://www.newegg.com/Product/Product.aspx?Item=N82E16812225063&cm_re=dp_to_vga-_-12-225-063-_-Product

$12 on NewEgg

Do those actually work or are they like the passive HDMI to VGA adapters?

[dalekphalm]

2 minutes ago, thekeemo said:

Do those actually work or are they like the passive HDMI to VGA adapters?

Can you please elaborate when you say "Do those actually work...?"

 

You connect it to a VGA monitor, and it gives you a video signal... what else would you need? DP can natively output a VGA, HDMI, or DVI-SL signal.

 

EDIT: And just to be sure, to satisfy you, we're using ~20 or so DP-to-VGA Adapters at work. All with zero issues.

[thekeemo]

Just now, dalekphalm said:

Can you please elaborate when you say "Do those actually work...?"

 

You connect it to a VGA monitor, and it gives you a video signal... what else would you need? DP can natively output a VGA, HDMI, or DVI-SL signal.

Did not know that.

I thought it was digital only like HDMI.

Why do we even have other GPU ports now??

[dalekphalm]

Just now, thekeemo said:

Did not know that.

I thought it was digital only like HDMI.

Why do we even have other GPU ports now??

Because people don't like using adapters. It's another thing to purchase, and it confuses people who don't know any better.

 

Realistically, a gaming grade GPU could get away with nothing but DP1.3 ports, or perhaps a single HDMI 2.0 port as well. But most people like having connection variety, because when they break out the their new GPU and realize it only has DP when they have VGA monitors, they're gonna be pissed. They might not have known better to go buy the adapter ahead of time or check the specs. Maybe it's Dad who went to Best Buy to grab a new GPU because his died. Not everyone knows better.

[thekeemo]

Just now, dalekphalm said:

Because people don't like using adapters. It's another thing to purchase, and it confuses people who don't know any better.

 

Realistically, a gaming grade GPU could get away with nothing but DP1.3 ports, or perhaps a single HDMI 2.0 port as well. But most people like having connection variety, because when they break out the their new GPU and realize it only has DP when they have VGA monitors, they're gonna be pissed. They might not have known better to go buy the adapter ahead of time or check the specs. Maybe it's Dad who went to Best Buy to grab a new GPU because his died. Not everyone knows better.

Wait so you dont want DVI because DP can adapt to it but you say we should have other ports because people dont like adapters?

[dalekphalm]

1 minute ago, thekeemo said:

Wait so you dont want DVI because DP can adapt to it but you say we should have other ports because people dont like adapters?

I'm not sure what you mean.

 

I personally would be happy with only DisplayPort ports on new GPU's.

 

I just understand why GPU manufacturers still put legacy video ports on their GPU's.

[thekeemo]

Just now, dalekphalm said:

I'm not sure what you mean.

 

I personally would be happy with only DisplayPort ports on new GPU's.

 

I just understand why GPU manufacturers still put legacy video ports on their GPU's.

Ah ok. Misunderstanding on my end.

[Sakkura]

12 hours ago, Daring said:

Am I seeing that right, or are they saying that the x80 Ti has a greater bus width than 384-bit? No way. Everyone knows Nvidia is allergic to 512-bit buses.

 

Also, the Titan and X80 Ti have different RAM types, despite being based on GP100. According to this guy, that's impossible for the same GPU core to have two different RAM types. As usual with any rumor, take this with a grain of salt.

While the different memory certainly is suspicious, you cannot use a random redditor as a source. I might as well say that, according to this guy, there's no problem.

[dalekphalm]

3 minutes ago, Sakkura said:

While the different memory certainly is suspicious, you cannot use a random redditor as a source. I might as well say that, according to this guy, there's no problem.

I'd be pretty skeptical about the memory config. But of course, in either case, we should take everything as pure rumour until solid details/benchmarks/reviews are released.

[Sakkura]

2 hours ago, NumLock21 said:

Nvidia is dumb to start with such a high number. After 8, it's 9 and they're out of number again. 

 

2 hours ago, LAwLz said:

I am pretty sure the X is just a placeholder, because we don't know if it will be called 1080, or some completely different numbering scheme.

According to the source, this series will be X80 Ti, X80, presumably X70 and so on down the line. Then the next line will be X180 and such, and then X280 etc.

[Samfisher]

With the introduction of the P100, I think I can safely say that all the rumoured specs are wrong.  The P100 has 3584 CUDA cores, and has 56 SM units.  Big Pascal has 60 SM units, so that's a further 256 CUDA cores.  The Pascal architecture uses 64 CUDA cores per SM unit compared to 128 CUDA cores per SM unit for Maxwell.

 

This means Big Pascal with 60 SM units will top out at 3840 CUDA cores, each more powerful than a Maxwell CUDA core.  Based on teraflop numbers, the P100 is 43% faster in FP32 performance than GM200 (Titan X) which is the most important metric for game performance.

[Big-G]

21 minutes ago, Samfisher said:

With the introduction of the P100, I think I can safely say that all the rumoured specs are wrong.  The P100 has 3584 CUDA cores, and has 56 SM units.  Big Pascal has 60 SM units, so that's a further 256 CUDA cores.  The Pascal architecture uses 64 CUDA cores per SM unit compared to 128 CUDA cores per SM unit for Maxwell.

 

This means Big Pascal with 60 SM units will top out at 3840 CUDA cores, each more powerful than a Maxwell CUDA core.  Based on teraflop numbers, the P100 is 43% faster in FP32 performance than GM200 (Titan X) which is the most important metric for game performance.

P100 specs will most likely be 100% Big Pascal -> Titan ?? For the new generation of GeForce GPU's.

[Lethal Seraph]

Meh. I'd rather wait for a HBM Ti than get the one that cones out right away.

[Agosto]

12 hours ago, Samfisher said:

With the introduction of the P100, I think I can safely say that all the rumoured specs are wrong.  The P100 has 3584 CUDA cores, and has 56 SM units.  Big Pascal has 60 SM units, so that's a further 256 CUDA cores.  The Pascal architecture uses 64 CUDA cores per SM unit compared to 128 CUDA cores per SM unit for Maxwell.

 

This means Big Pascal with 60 SM units will top out at 3840 CUDA cores, each more powerful than a Maxwell CUDA core.  Based on teraflop numbers, the P100 is 43% faster in FP32 performance than GM200 (Titan X) which is the most important metric for game performance.

Specs were obviously fake from the beginning, they listed both GP100 variant with different memory technologies.

[Samfisher]

Just now, Agost said:

Specs were obviously fake from the beginning, they listed both GP100 variant with different memory technologies.

What is fake from the beginning?  https://devblogs.nvidia.com/parallelforall/inside-pascal/

 

Posted 3 days ago.

 

Quote

GP100 achieves its colossal throughput by providing six GPCs, up to 60 SMs , Each SM has 64 CUDA cores and four texture units, for a total of 3840 CUDA cores and 240 texture units.

 

[Agosto]

14 minutes ago, Samfisher said:

What is fake from the beginning?  https://devblogs.nvidia.com/parallelforall/inside-pascal/

 

Posted 3 days ago.

 

 

I was obviously referring to the ones in the OP, just like I wrote in my first comment on this thread

[Arty]

On 3/18/2016 at 11:37 AM, dalekphalm said:

Can you please elaborate when you say "Do those actually work...?"

 

You connect it to a VGA monitor, and it gives you a video signal... what else would you need? DP can natively output a VGA, HDMI, or DVI-SL signal.

 

EDIT: And just to be sure, to satisfy you, we're using ~20 or so DP-to-VGA Adapters at work. All with zero issues.

My school even uses usb to vga adapters for thier 1080p monitors at 60hz,  on about half the computers. (obs won't be gaming on that but , for school purposes it works)

[Samfisher]

4 minutes ago, Agost said:

I was obviously referring to the ones in the OP, just like I wrote in my first comment on this thread

I dono.  You quoted me instead of tagging me, which makes it obvious you were referring to my post  But eh, I never really cared for rumour specs cos you know, they're rumours.

[Agosto]

1 minute ago, Samfisher said:

I dono.  You quoted me instead of tagging me, which makes it obvious you were referring to my post  But eh, I never really cared for rumour specs cos you know, they're rumours.

I usually quote because it's faster.

If big pascal is coming only in 2017, Polaris has a good chance IMHO.

[Delicieuxz]

22 minutes ago, Agost said:

Specs were obviously fake from the beginning, they listed both GP100 variant with different memory technologies.

Why do you take that to mean they're not legit? All that would suggest to me is that one card would release earlier than the other.

[Agosto]

12 minutes ago, Delicieuxz said:

Why do you take that to mean they're not legit? All that would suggest to me is that one card would release earlier than the other.

They were listed with the same GP100 chip but with different memories. No one would design and produce a GPU chip with two different controllers, it's wasted space and the P100 is already pretty big.

[Delicieuxz]

5 minutes ago, Agost said:

They were listed with the same GP100 chip but with different memories. No one would design and produce a GPU chip with two different controllers, it's wasted space and the P100 is already pretty big.

Are you certain the memory controller would be integrated, or non-interchangeable?

[Kunter]

1 minute ago, Delicieuxz said:

Are you certain the memory controller would be integrated, or non-interchangeable?

Yes, because the smaller variant (like 950, 970, 980ti) are the chips with one bad block. So 980ti is originally produced as a titan x. It is the exact same GM200, so it is simply not possible to change the controller AFTER the chip is already made.

[Delicieuxz]

1 minute ago, Kunter said:

Yes, because the smaller variant (like 950, 970, 980ti) are the chips with one bad block. So 980ti is originally produced as a titan x. It is the exact same GM200, so it is simply not possible to change the controller AFTER the chip is already made.

I see, interesting point.