NVIDIA Pascal Thread

[airfathaaaaa]

Pascal was not put in the place of Volta, since Volta didn't have a place (time wise).

Pascal when it was introduced was given a date of 2016, at that point in time (2014) Volta hadn't been given a date so there was no place for Pascal to take.

It is quite possible that Volta was originally planned for earlier than 2018, and perhaps even 2016 (where Pascal ended up), but we have no real way of knowing this without access to Nvidia's internal roadmaps.

Anyways this is getting a bit off topic, so I won't bother belabouring the point any further.

last post for this

volta at first was going to be the successor of maxwell cards..
http://www.geeks3d.com/20130319/gtc-2013-nvidia-volta-post-maxwell-gpu-platform/
then it got renamed into pascal
http://wccftech.com/nvidia-generation-pascal-gpu-feature-3d-stacked-memory-nvlink-1-tbs-bandwith/
now the volta we see on 2015 roadmaps was meant to be the pascal we saw on early 2014
but because of all the huss and fuss about dx12 volta kicked in again into being a new uarch into the future

 

[Silverforce11]

Er, the source:

It's just someone editing the bios tables for boost states of GM204/980.

If you had LN cooling and unlimited TDP sure.

 

[nvgpu]

http://videocardz.com/59558/nvidia-geforce-gtx-1080-3dmark-benchmarks

1.8GHz overclocked core clock(?), GDDR5X 10GHz 320GB/s bandwidth only for these card samples, 368.03 driver is the latest known driver.

 

[Head1985]

Stock reference 980Ti have 8000points in Firestrike extreme.1080 is 26% faster
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_980_ti_review,24.html

 

[Sweepr]

Stock reference 980Ti have 8000points in Firestrike extreme.1080 is 26% faster
http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_980_ti_review,24.html

Nice find. Exactly what I was expecting.

 

[linkgoron]

http://videocardz.com/59558/nvidia-geforce-gtx-1080-3dmark-benchmarks

1.8GHz overclocked core clock(?), GDDR5X 10GHz 320GB/s bandwidth only for these card samples, 368.03 driver is the latest known driver.

Those 1080 scores seem low to me.

According to kitguru they get a 19859 on 3dmark11 with their "ASUS STRIX Gaming GTX 980 Ti DirectCU 3" (unless I'm missing something?) vs ~19005 of the 1080.

The zotac 980ti from guru3d gets 8891 on firestrike extreme 1.1 vs 8959 of the 1080.

So basically almost equivalent (at stock at least) to some after-market 980tis.

 

[railven]

Those 1080 scores seem low to me.

According to kitguru they get a 19859 on 3dmark11 with their "ASUS STRIX Gaming GTX 980 Ti DirectCU 3" (unless I'm missing something?) vs ~19005 of the 1080.

Looking at graphics score, 1080 is ~2K higher, and 980 Ti used is ~20% OC'd over stock 980 Ti.

CPU Physic score is also higher on the other test, with the OC'd 980 Ti.

The GTX 1080 still beating the OC'd 980 Ti is good to see.

EDIT: Same applies to the Guru3D scores. GPU score is higher for 1080 vs OC'd 980 Ti, CPU score is lower.

 

[rainy]

I wish to understand logic of some Nvidia fans: GTX 980 Ti was released less than year ago (2 June 2015) at 649 dollars price.
If 1080 is 25 percent faster and will cost 600-650 dollars, it's really worth to get it?
Nvidia for sure will release significant faster card in the next 9-12 months - personally, I would prefer 50-60 percent more performance having such strong card like 980 Ti.

 

[railven]

I wish to understand logic of some Nvidia fans: GTX 980 Ti was released less than year ago (2 June 2015) at 649 dollars price.
If 1080 is 25 percent faster and will cost 600-650 dollars, it's really worth to get it?
Nvidia for sure will release significant faster card in the next 9-12 months - personally, I would prefer 50-60 percent more performance having such strong card like 980 Ti.

Not an NV fan, per say, but I look at my upgrade money like a government contract. If I don't spend it, the GF claims it and I won't get it again until the following year. She's already eyeing a new bed frame (EDIT: Might I add when I wore a red cape, I regularly upgrade my GPU with every new card AMD put up. I skipped 290X due to price inflammation and Fury X under performed at launch when compared to the same priced GTX 980 Ti).

Also, some of us make enough income where this isn't a financial question.

 

[linkgoron]

I wish to understand logic of some Nvidia fans: GTX 980 Ti was released less than year ago (2 June 2015) at 649 dollars price.
If 1080 is 25 percent faster and will cost 600-650 dollars, it's really worth to get it?
Nvidia for sure will release significant faster card in the next 9-12 months - personally, I would prefer 50-60 percent more performance having such strong card like 980 Ti.

It looks (imo) to be much less than 25% if you bought a decent after-market 980ti.

I really think that those scores are a bit low, unless the 1080 overclocks well (i.e. clocked to basically edge the 980ti OC scores at stock, and will OC as well as the 980ti), as otherwise I don't see the incentive to upgrade from a decent after-market 980ti.

 

[Timmah!]

Some last minute rumours/leaks before the event tomorrow (thanks to user PolyMorph @ Chiphell):





www.chiphell.com/thread-1579388-1-1.html

500 MHz clock boost compared to Maxwell. 10-15% better IPC.

Reference Geforce GTX 1080 supposedly costs 3999 CNY. Probably U$599?
Custom models between 4500-5000+ CNY.

Thats what i got, when i tried to calculate the perf difference between GM200 and P100 based on their proclaimed peak FP32 performance.

 

[Sweepr]

http://videocardz.com/59558/nvidia-geforce-gtx-1080-3dmark-benchmarks

1.8GHz overclocked core clock(?), GDDR5X 10GHz 320GB/s bandwidth only for these card samples, 368.03 driver is the latest known driver.

Here's the full article:

NVIDIA GeForce GTX 1080, 8GB RAM GDDR5X at 10 GHz

I think the most interesting detail about GTX 1080 is the new memory. NVIDIA has finally used 8GB ram for its flagship card. It’s no longer an exclusive to mobile solutions. Additionally, the GDDR5X modules are clocked at 2500 MHz (which is shown as 5000+ MHz in 3DMark). However the effective clock is 10000 MHz, which means the bandwidth is somewhere around 320 GB/s (assuming it’s 256-bit wide).

On the GPU side we have a huge improvement in terms of frequencies. It is said that GTX 1080 can boost up to 1.8 GHz, with base clock around 1.6 GHz. At the time of writing we are not able to confirm the exact reference clock. For such reason I decided to avoid making comparison charts, so this post will essentially tell you what GTX 1080 is capable of and nothing more.

NVIDIA GeForce GTX 1080 8GB — 3DMark11 Performance
Like always, we are only looking at graphics score. 3DMark11 Performance preset is rendered at 1280×720 resolution. GeForce GTX 1080 scores 27683 points, which is still above overclocked GM200 cards (~23-25k). Worth noting 3DMark11 is not showing correct GPU clocks, however the new driver already supports GTX 1080 by its name.

NVIDIA GeForce GTX 1080 8GB — 3DMark FireStrike Extreme
The second benchmark we have is from FireStrike Extreme, which as you probably already know is rendered at 2560×1440 resolution. This is actually the first benchmark where you will see 1.860 GHz clock. In this scenario GTX 1080 is faster than typical overclocked GTX 980 Ti (~8700 points). However if we compare it GTX 980 Ti running at almost the same frequency (1.8GHz with LN2 cooling), GTX 1080 is actually much slower clock to clock (link).

Benefit of using higher frequencies will definitely make GTX 1080 faster than most high-end cards out of the box. However is this a better product in all scenarios? Will be just as fast in games? How does it perform at 4K? Will VR performance be satisfying? Well, we don’t have answers to these questions yet, but we will keep digging.

http://videocardz.com/59558/nvidia-geforce-gtx-1080-3dmark-benchmarks

Hopefully GP104 has similar (or better) OCing potential/headroom compared to Maxwell.
People comparing it to OCed Geforce GTX 980 Ti should wait for non-reference / OCed Geforce GTX 1080 results.

Ps: First post updated.

 

[Det0x]

Those 1080 scores seem low to me.

According to kitguru they get a 19859 on 3dmark11 with their "ASUS STRIX Gaming GTX 980 Ti DirectCU 3" (unless I'm missing something?) vs ~19005 of the 1080.

The zotac 980ti from guru3d gets 8891 on firestrike extreme 1.1 vs 8959 of the 1080.

So basically almost equivalent (at stock at least) to some after-market 980tis.

Yes i full heartedly agree with you.. but earlier leaks have shown the very same low performance delta between 1080 and aftermarket 980ti. (like pretty much all sold 980ti are)

Its just that some people are so blinded by their own bias that they refuse to believe anything to the contrary.

So it seems that this unknown card is ~4% faster then the fastest aftermarked 980 TI.

Unknown card:
Graphics score: 27271
GT1: 127.63 FPS
GT2: 132.23 FPS
GT3: 170.97 FPS
GT4: 80.08 FPS

GTX 980 Ti Waterforce:
Graphics score: 26112
GT1: 121.41 FPS
GT2: 126.93 FPS
GT3: 166.33 FPS
GT4: 76.31 FPS

www.3dmark.com/3dm11/11162175

Score: P20625
Graphics Score: 27271

www.chiphell.com/thread-1565171-1-1.html

For reference:

*Probably different CPUs, the Physics Score suggests a slower chip than the Core i7-5960X @ 4.4GHz used by Guru3D.

You can read the the rest back on page 64 in this thread. ()

If the rumored price of 649+$ is true, i would deem the whole geforce pascal to be a utter failure, possible even GP100 too. :'( Remember we haven't seen any working silicon yet and a recent article on SA could point to why. ()

 

[96Firebird]

Are you purposely ignoring the graphics score or...?

 

[Det0x]

Are you purposely ignoring the graphics score or...?

What ? By your own post:

27271 / 26112 = ~4%

*edit*

No reason to add fuel to the fire

 

[Sweepr]

At stock, it's faster than Geforce GTX 980 Ti overclocked to 1.5 GHz (3DMark 11 Graphics Score).

BTW Chiphell suggests the card in this leak is actually a bit underclocked (Geforce GTX 1080 could operate at 1823MHz - up to 1962MHz). 1 day to go.

 

[96Firebird]

Yes, notice I said unknown card. Could be anything, I don't pretend to know things, unlike you... Deeming the entire Pascal launch a failure before it even gets released, you're a joke.

 

[airfathaaaaa]

At stock, it's faster than Geforce GTX 980 Ti overclocked to 1.5 GHz (3DMark 11 Graphics Score).

BTW Chiphell suggests the card in this leak is actually a bit underclocked (Geforce GTX 1080 could operate at 1823MHz - up to 1962MHz). 1 day to go.

but the 1070 leak shows 2022

 

[Glo.]

Judging by performance delta, it looks like GP104 has 2048 CUDA cores, not 2560.

Water force Extreme Gaming has 7.6 TFLOPs of compute power(7.558) and this would be similar to 2048 CUDA cores clocked at 1.86 GHz(7.618). Performance delta is quite close.

 

[Det0x]

Judging by performance delta, it looks like GP104 has 2048 CUDA cores, not 2560.

Water force Extreme Gaming has 7.6 TFLOPs of compute power(7.558) and this would be similar to 2048 CUDA cores clocked at 1.86 GHz(7.618). Performance delta is quite close.

Lo and behold

http://vrworld.com/2016/04/25/16nm-nvidia-geforce-gtx-1080-leak-ahead-computex-taipei-2016/

While the specs of GP104 are still hidden, we can now say that the chip packs 1920 CUDA Cores and a 256-bit controller supports both GDDR5 and GDDR5X (MSI uses GDDR5X) memory. That should suffice for up to 384 GB/s i.e. easily surpass the bandwidth offered by the GeForce GTX 980, 980 Ti, and even TITAN X. Expect GeForce GTX 1080 and 1070 to make an appearance on Computex Taipei 2016, while GP106 based 1050/1060 will debut in the fall.

 

[Sweepr]

- Some stock (and OCed) Geforce GTX 980 Ti and Fury X numbers, for reference:

- Geforce GTX 1080 (?):

~30% faster than Geforce GTX 980 Ti / ~68% faster than Geforce GTX 980 (stock clocks).


BTW, the event tomorrow will be livestreamed:

Tune in to an NVIDIA Special Event livestream May 6th at 6pm PST, exclusively on @Twitch! http://nvda.ly/4nrpFT

https://twitter.com/NVIDIAGeForce/status/728222996138168320

 

[renderstate]

Unconfirmed perf numbers without TDP numbers and people already deeming it a failure. Just wow..I am speechless.


Sent from my iPhone using Tapatalk

 

[Head1985]

Lo and behold

http://vrworld.com/2016/04/25/16nm-nvidia-geforce-gtx-1080-leak-ahead-computex-taipei-2016/

With 1920SP Gp104 should be only 200mm2.
If true then will Gp104 have Fp64 like Tesla.This is pretty bad for Nv to have Gp104 100% gaming SKu with half FP64 units.

 

[railven]

With 1920SP Gp104 should be only 200mm2

Isn't that what Polaris 10 is at in terms of die size? Does that mean Polaris 10 can hit Fury X+ performance?

The plot thickens!

 

[R0H1T]

Isn't that what Polaris 10 is at in terms of die size? Does that mean Polaris 10 can hit Fury X+ performance?

The plot thickens!

Of course, they only need high(er) clocks & GDDR5x, the IPC gains & 14nm node will take care of the rest.