• 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.

** The Official Nvidia GeForce 'Pascal' Thread - for general gossip and discussions **

Simon Belmont said:
I know GTC in April. People waiting for Pascal High end, you guys have a long wait. I expect GP104 to be faster then both 980Ti and Titan X.
Click to expand...

Maybe not.

Neither the TitanX or GTX 980 Ti have compute functions which was one of the reasons they perform so well.

Pascal will have the compute functions back which will reduce performance for gaming.
 
The mid range may well be faster but I can't see it being by a great deal and being worth the switch.

I'll not be going for the Pascal Titan this time though, so my wait may be even longer as the TI will likely be the final card high end card again.
 
Ayahuasca said:
The mid range may well be faster but I can't see it being by a great deal and being worth the switch.

I'll not be going for the Pascal Titan this time though, so my wait may be even longer as the TI will likely be the final card high end card again.
Click to expand...

All depends what putting the compute function back into the mid range does to gaming performance.
 
Kaapstad said:
Maybe not.

Neither the TitanX or GTX 980 Ti have compute functions which was one of the reasons they perform so well.

Pascal will have the compute functions back which will reduce performance for gaming.
Click to expand...


I don't know? These cards have to be faster otherwise the new cards wont sell. So They have to be faster one way or another. I think neither of us know what's coming? We'll know soon enough that's for sure.
 
Past experience has shown that Nvidia are only too happy to release their new mid-range GPUs at high end prices then release the true high end quite a bit later at even higher prices.

I feel Nvidia have had a good return on this strategy and unless AMD execute their Polaris release perfectly (I know lol) I see no reason for Nvidia to change it. At least from a business perspective though not so good for consumers.
 
Last edited:
ICDP said:
Past experience has shown that Nvidia are only too happy to release their new mid-range GPUs at high end prices then release the true high end quite a bit later at even higher prices.

I feel Nvidia have had a good return on this strategy and unless AMD execute their Polaris release perfectly (I know lol) I see no reason for Nvidia to change it. At least from a business perspective though not so good for consumers.
Click to expand...

Yeah this is my worry as well, especially with it basically being a 2 generational leap in one (Skipped 20nm, so 2 die shrinks + Finfet). I.e the new midrange is very likely to outperform the older high end 980 Ti, but will be priced in that same 'high end' bracket, rather than replacing the 970 tier. With the hook being lower power consumption. Meanwhile we'll have to wait another 6 - 12 months for the real deal 980 Ti replacement which offers much higher performance which we should have got at the start..

The responsibility is AMD's also, because if they have nothing to counter basically Nvidia can just get away with it lol. Maybe AMD / Nvidia are happy to drip feed us...

Meh, get ready for the milking !! :p:D
 
Boomstick777 said:
Yeah this is my worry as well, especially with it basically being a 2 generational leap in one (Skipped 20nm, so 2 die shrinks + Finfet). I.e the new midrange is very likely to outperform the older high end 980 Ti, but will be priced in that same 'high end' bracket, rather than replacing the 970 tier. With the hook being lower power consumption. Meanwhile we'll have to wait another 6 - 12 months for the real deal 980 Ti replacement which offers much higher performance which we should have got at the start..

The responsibility is AMD's also, because if they have nothing to counter basically Nvidia can just get away with it lol. Maybe AMD / Nvidia are happy to drip feed us...

Meh, get ready for the milking !! :p:D
Click to expand...

Yep, I have been very critical in the past about the fact that AMDs ineptitude allows Nvidia to release new GPUs without any viable alternatives. 780 and 980 were all released without an AMD equivalent and this allowed Nvidia to price their midrange at high-end prices. Nvidia even took their advantage to extreme by adding a whole new price segment for Titan and Titan X.
 
Considering Volta will be shipping to one of sierra/summit supercomputers sometime in 2017 it makes you wonder what their approach to big pascal might be if to be so soon superseded by volta. Although with those contracts to meet, gaming consumers can wait at the back of the queue ofc.

1. Middle sized (but on the larger end of middle) 16nm pascal die to much neededly replace their aging kepler GK210 for fp64 in a somewhat ASAP way. Soon to be followed by their biger compute die volta a year later. All available in some form to gaming consumer, professional and HPC markets priority depending.

2. Or whether they will bifurcate gaming/fp32 and HPC/compute more cleanly. So perhaps overlap of HPC/compute dies to the professional market but not to consumers in any meaningful way (ie gaming) as a largish non-fp64 could comprehensively outperform it (repeat a GM 200-980ti/TitanX). Their gaming/fp32 chips will span consumer and professional products as they do now.

In the second scenario I can envisage the compute pascal maybe appearing in the consumer market as a brief halo product, end of 2016, simply due to the large performance increase from the smaller process making it viable. The bifurcation manifesting after. Otherwise they could be planning on simply scaling up their gaming arch to whatever large die size they expect the 16nm process supports, for a beefy gaming chip for late 2016/early 2017 and leave compute pascal for the pro market.

Anyway just some ponderings.
 
ICDP said:
Nvidia Pascal over a year ahead of 14/16nm competition

http://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

Sensing some sarcasm but gotta love that Nvidia learn from past mistakes. :D
Click to expand...

I was like:
472.gif
 
Rroff said:
Lol at the sarcasm.
Click to expand...

Blatant sarcasm and the fact that Charlie is clearly in the "Nvidia unfriendly" camp aside, I do find it perplexing that Nvidia are obviously telling porkies again. IMHO it hurts them where most consumers are concerned and even Nvidia diehards must cringe with embarrassment when they see this crap. This is right up there with the hilariously (in hindsight) idiotic choice of describing Fiji as an "overclockers dream" BS from AMD.

Maybe it fools the stockholders etc but anyone even a bit GPU tech savvy would be thinking "FFS Jen are you having a giraffe".
 
Last edited:
eddyr said:
Considering Volta will be shipping to one of sierra/summit supercomputers sometime in 2017 it makes you wonder what their approach to big pascal might be if to be so soon superseded by volta. Although with those contracts to meet, gaming consumers can wait at the back of the queue ofc.

1. Middle sized (but on the larger end of middle) 16nm pascal die to much neededly replace their aging kepler GK210 for fp64 in a somewhat ASAP way. Soon to be followed by their biger compute die volta a year later. All available in some form to gaming consumer, professional and HPC markets priority depending.

2. Or whether they will bifurcate gaming/fp32 and HPC/compute more cleanly. So perhaps overlap of HPC/compute dies to the professional market but not to consumers in any meaningful way (ie gaming) as a largish non-fp64 could comprehensively outperform it (repeat a GM 200-980ti/TitanX). Their gaming/fp32 chips will span consumer and professional products as they do now.

In the second scenario I can envisage the compute pascal maybe appearing in the consumer market as a brief halo product, end of 2016, simply due to the large performance increase from the smaller process making it viable. The bifurcation manifesting after. Otherwise they could be planning on simply scaling up their gaming arch to whatever large die size they expect the 16nm process supports, for a beefy gaming chip for late 2016/early 2017 and leave compute pascal for the pro market.

Anyway just some ponderings.
Click to expand...

It could be that Volta is for commerical use and Pascal for consumer use?
 
ICDP said:
Yep, I have been very critical in the past about the fact that AMDs ineptitude allows Nvidia to release new GPUs without any viable alternatives. 780 and 980 were all released without an AMD equivalent and this allowed Nvidia to price their midrange at high-end prices. Nvidia even took their advantage to extreme by adding a whole new price segment for Titan and Titan X.
Click to expand...

From a consumer perspective, mid and high end merely refer to performance. Why does a consumer care if a chip is 300mm^@ or 500mm^2? They don't, they care about performance. The 980 may have been the mid-sized chip but it outperformed the previous fastest GPU and thus got priced accordingly.
People knew that at the time, the chip size was public information, as was the code name. People didn't care, they knew exactly what they were getting, a GPU that was the fastest on the planet yet used less power and made less heat than the predecessor, and cold be clocked very well.


The TitanX was stupidly priced but that is just market economics. a Lamborghini is stupidly priced but there is a market for it. That is why Nvidia tend to do so well - they realize there is a market segment at TX levels and have the brand recognition to be able to sell in that space. People knew the price-performance was poor and that Nvidia would have a massive profit margin but the people that bought a TX simply don't care because ultimately they have more disposable income and price is less of a factor.

I have a friend who loves wine but know nothing about it, they worked for a start-up that went public so he has a pile of cash. when he buys wine he doesn't look at anything less than $50, simply because he thinks he has a good chance of getting a good wine if he pays more and can't be bothered looking for the cheaper bargains. I know a tiny bit more about wines but have a fraction of the cash so I never spend more than $20 and i'm looking for <$10 bargains that knock the socks off his $50 bottles.
 
Last edited:
ICDP said:
Nvidia Pascal over a year ahead of 14/16nm competition

http://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

Sensing some sarcasm but gotta love that Nvidia learn from past mistakes. :D
Click to expand...

Well, according to the AMD Polaris thread AMD were shipping Polaris chips in January 2015. All just BS.



I do know that Pascal based PX2 will be in developers hands by around April. What sits behind glass in a CES both is largely irrelevant. A lot of the demo products are from what the manufacturer actually claims. I know a certain auto OEM that was touting an autonomous car but the showfloor model was mostly just a shell. For the Nvidia PX2 customers aren't expecting production until Q4 so nvidia have most of a year to turn engineering samples into production.
 
You must log in or register to reply here.
Back
Top Bottom