How about we agree that you scored on the black curtain point, and that I scored on being right about AMD not only failing miserably at execution, but also doing so very publicly by prancing around with it and by crowing about how much ahead they are and then not deliver.
Using this kind of argument penalizes those who with the best engineering: you're able to make a 300mm2 die with better perf/W, perf/mm2 and better absolute perf than a 360mm2 die (as was the case at GTX 680 launch) ?
Sorry, buddy, you're mid-end, while your less able competitor is allowed to call himself high-end.
The maximum die size doesn't change over the life time of a process. It's simply defined by the stepper size. It should be somewhere in the 700+mm2.
D
Deleted member 13524
Guest
You can claim whatever score you want by the end of Q3. Until then you're simply speaking out your wet dreams about AMD failing and nvidia getting a nobel prize on corporate execution.
I hope those dreams include having to pay $2000 for a mid-range graphics card that is barely any better than the mid-range from 2 years ago, in case they come true. Otherwise you're in for some disappointment.
So if I get these two, I'd be set for Battlefield, eh?
Now you are turning it into a comparison of competitors.
Fact remains that the 1080, while being a great achievement, doesn't stretch the design parameters to the max.
Their is clearly headroom for a higher end card that will appeal more to people having currently a 980Ti, Titan-X or Fury-X.
DuckThor Evil
Legend
From the Vega thread...
I am very sceptical of that chip ever becoming a Geforce. A 450-480mm2 GP102 with the same SM count as the GP100, dropped FP64 baggage with 384bit GDDR5X should be far better Geforce than GP100 would ever be.
I am very sceptical of that chip ever becoming a Geforce. A 450-480mm2 GP102 with the same SM count as the GP100, dropped FP64 baggage with 384bit GDDR5X should be far better Geforce than GP100 would ever be.
Last edited:
Indeed that would make perfect sense.
+1000
I think the same. IMHO, the perfect range will be:
GP106 : 200mm2 / 20 gaming SM / 128bit GDDR5 / For 1050 and 1060
GP104 : 320mm2 / 40 gaming SM / 256bit GDDR5/X / For 1070 and 1080
GP102 : 460mm2 / 60 gaming SM / 384bit GDDR5X / For 1080Ti / Titan and quadro (deep learning)
GP100 : 610mm2 / 60 HPC SM / HBM2 / For Tesla
D
Deleted member 13524
Guest
I thought the GP100 was reportedly lacking in ROP count and other consumer stuff, in order to make room for more computational resources.
No-one knows as they only reported the co-processor/accelerator aspects of the Pascal architecture in context of Tesla.
I thought they did say the GP100 die was never going to be a consumer product though.
But they could still/intend to manufacturer a GP102 *shrug*
It would make sense.
Cheers
According to what reports?
Remember, when they announced GF100 back in .. 2010 (?): Compute architecture only, but of course it had ROPs and TMUs and everything.
Yes there will be there will be factory overclocked cards form AIB partners, which will be more expensive. Just like they did in the past. Either nV could have pushed the clocks up or let their AIB partners do it, it doesn't matter who does. That tells ya the 1080ti or titan versions will not be GP104's.
D
Deleted member 13524
Guest
To be honest, if nvidia manages to put a GP104 running at 2GHz base / ~2.4GHz boost clocks, coupled with 12GT/s GDDR5X while keeping power consumption under 275W, then they don't really need to make a GP102.
Get an AiO liquid cooler to make it silent and more premium, sell it for $750 and there's your 980 Ti.
Back in 2008, nvidia got through an entire year using only G92 from upper-mid to top-end ranges, and they were very successful at it.
Get an AiO liquid cooler to make it silent and more premium, sell it for $750 and there's your 980 Ti.
Back in 2008, nvidia got through an entire year using only G92 from upper-mid to top-end ranges, and they were very successful at it.
I'm sure the AIB partners are already sketching up their custom boards with this in mind. The
DuckThor Evil
Legend
I'm sure there will be plenty of exotic SKUs and GP104 could have a nice lengthy run as the top dog, especially if AMD continues to be MIA. However GP104 is not and will not be the Pascal Titan and typically the Titan chip gets a cut SKU as well. Maybe Titan is 2017 or Christmas 2016, but it's coming and it won't be GP104.
D
Deleted member 13524
Guest
Why not? Besides memory amount and price, Titan as a brand is a moving goalpost.
The first Titan was supposed to be the top offering for the Kepler generation along with fully enabled FP64 units (per SMX) so it was useful for DP compute too.
Half a year later the 780 Ti came out with higher clocks and more SMX units enabled and Titan wasn't the top offering anymore.
Then came the Titan Black with all the SMX units too and higher clocks than the 780 Ti, but with higher-clocked custom 780 Ti boards the Titan Black still wasn't the top offering for gaming.
- It was then decided that Titan would be the brand name that carried a premium for larger memory amount and unlocked DP compute.
Them came Maxwell and Titan X was again the top performance offering compared to the 980 Ti, but this time its chip came with a very low DP-SP ratio (there were no DP units to unlock this time around).
- It was then decided that Titan would be the brand name for top performance and more memory.
Then came the custom 980 Ti versions with higher clocks that easily surpassed the Titan X.
- It was then decided that Titan would be the brand name for.. erm.. more memory.
Who knows what nvidia decides to call a Titan this time around...
And until (if ever) AMD decides to launch a GPU that directly competes with the GTX 1080 in performance, there's little reason for nvidia to launch a more powerful "prosumer" chip. Especially if the GP104 can clock as high and as easily as nvidia seems to claim. GP104 + >2GHz clocks + 16GB GDDR5X 12MT/s + AiO liquid cooler + $1200 price tag and there's your Titan.
Last edited by a moderator:
