Nvidia Pascal Announcement

[Genotypical]

this is the problem with that doom vid. But some comparisons could be possible. 390x can do 60 fps at 1440 (59 minimum). 980 does around 35fps at 4k. Should be around 120-140 fps at 1080p. Though not sure the test done for these numbers is as demanding, but gives some idea. Unfortunately the beta was capped at 60 fps, which I guess is why they are using doom so there is no direct comparison (plus the vulkan). Under vulkan the 390x at least should do even better, probably the 980 as well.

source http://www.pcgameshardware.de/Doom-2016-Spiel-56369/Specials/Open-Beta-Benchmarks-1192393/

 

[smw]

Flops aren't the only thing that come into the performance metric

PS, for more increases in performance

http://deliddedtech.com/2016/05/07/hidden-information-found-in-source-code-of-nvidias-gtx-1080-page/

they measured out the graph lengths and calculated from them.

80% increased performance for ROTR with a 10% increase in power consumption. That it higher than 2.5x perf/watt

70% increased performance for Witcher 3 which will end up around 2.5x per per watt.

Percentages, man! This is not how they work
In essence you're saying that 1080 in ROTR is 80% faster => 1.8 perf of 980. Even if they have the same TDP, 1.8 is the ceiling that you can have for perf/watt of 1080 vs 980. But you also say that 1080 draws 10% more power than 980. That means you also divide by a further 1.1 and arrive at about 1.64 perf/watt. For TW3 it is actually 1.7/1.1 = 1.54

 

[Deleted member 13524]

I think it's obvious today that AMD isn't ahead with the FinFET transistor even though they claimed they were.

AMD has shown working silicon in early January, with live demos for the public to see.
5 months later, nvidia showed a card on stage that might or might not have been assembled with woodscrews and some demos through a projector that might or might not have been rendered by the new cards.
Come May 27, you might or might not see a paper launch.

Don't get ahead of yourself.

80% increased performance for ROTR with a 10% increase in power consumption. That it higher than 2.5x perf/watt

And this, my children, is why you shouldn't skip math class.

 

[silent_guy]

Come May 27, you might or might not see a paper launch.

It's easy to check.

After Friday's event, what will happen with the sales of any Maxwell GPU that's a GTX 970 or higher?

If your paper launch hunch is correct, we should see a dramatic drop in GeForce revenue for the financial quarter that ends in July.

 

[Deleted member 13524]

If your paper launch hunch is correct, we should see a dramatic drop in GeForce revenue for the financial quarter that ends in July.

First off, it's not a hunch, just an hypothesis that is still on the table. Meaning it's too early to reach the conclusions you stated in your previous post.

Second, I doubt it would mean a dramatic drop in revenue, because not many end-users keep up with GPU announcements and the GTX970 is still nvidia's best offer for the ~$300 segment (which the new cards don't cover, BTW).
Graphics cards don't get the same Osborn-effect as for example smartphones (whose announcements tend to appear on mainstream media).

 

[CSI PC]

Nobody tell you that they are not in production.. But looking that they only use 10GB/s ( the 1080 with GDDR5x have a lower bandwith than the 980TI ), can make think that it was due to the production .. But it can mean too ( for me ) , that they keep a bit of "overclock" margin for non reference gpu's on the memory side.
( at contrario of AMD gpu's, most non reference gpu's maxwell was sold with higher speed on memory and a large difference in some case, it was even a big part of their marketing )

What I mean is Micron themselves explicitly state the 11 and 12Gb/s memory is in sampling status, their status for 10Gb/s went into production status quite awhile ago when I checked their site.
So AIB/manufacturers/etc will still need to wait for the upper two memory products to be placed into factory orders.
Micron is not hiding this type of information.
This makes sense because the one time the detail was shown with a Polaris card using GDDR5X, that also was 10gb/s.
Both AMD and NVIDIA are limited to this until I am sure Micron complete the sampling phase (which I agree means some of these memories are given to the various manufacturers interested so it is in limited production).
Cheers

 

[xEx]

Do we know the size of the chip of the 1080?

 

[xpea]

reviewers got their cards after the event.

We should see real numbers this week

Edit: and Zotac has already the 1080 on their website:
https://www.zotac.com/gb/product/graphics_card/geforce-gtx-1080

 

[xpea]

and interesting infor from gurud3D:
http://www.guru3d.com/news-story/geforce-gtx-1080-photos.html

So at last nights event there where a couple of setups showing the GeForce GTX 1080. The cards where available in large quantities alright.

Does it mean that Nvidia won't have any supply issue or do they give everything they have to reviewers

 

[Kaotik]

AMD has shown working silicon in early January, with live demos for the public to see.
5 months later, nvidia showed a card on stage that might or might not have been assembled with woodscrews and some demos through a projector that might or might not have been rendered by the new cards.
Come May 27, you might or might not see a paper launch.

Don't get ahead of yourself.

Actually AMD showed working silicon already in December for the press, January was only first for public eyes

 

[McHuj]

I hope they are getting the 1070's as well.

 

[Razor1]

AMD has shown working silicon in early January, with live demos for the public to see.
5 months later, nvidia showed a card on stage that might or might not have been assembled with woodscrews and some demos through a projector that might or might not have been rendered by the new cards.
Come May 27, you might or might not see a paper launch.

Don't get ahead of yourself.



And this, my children, is why you shouldn't skip math class.

Well I can see how they got 2.5 perf per watt now they were comparing, titan x which has similar transistor amounts to gp 104 and going from that.

 

[Razor1]

Percentages, man! This is not how they work
In essence you're saying that 1080 in ROTR is 80% faster => 1.8 perf of 980. Even if they have the same TDP, 1.8 is the ceiling that you can have for perf/watt of 1080 vs 980. But you also say that 1080 draws 10% more power than 980. That means you also divide by a further 1.1 and arrive at about 1.64 perf/watt. For TW3 it is actually 1.7/1.1 = 1.54

I was trying to fit a number to what I thought was something from the presentation which it wasn't it was from a titan X to a gtx 1080 they were comparing they used those two since the transistor counts are similar.

I should have done the math myself, but was too lazy

 

[DavidGraham]

5 months later, nvidia showed a card on stage that might or might not have been assembled with woodscrews and some demos through a projector that might or might not have been rendered by the new cards.

So you claim that the ID guys lied about running their DOOM Vulcan piece on a real hardware? Let alone dozens of reviewers who got cards as well?

AMD has shown working silicon in early January, with live demos for the public to see.

Showing silicon and live demos doesn't mean real production (ie, in large quantities) has started, Silentguy's has a solid hepothesis as well, if they are ready indeed then why the delay?

 

[gamervivek]

AMD has shown working silicon in early January, with live demos for the public to see.
5 months later, nvidia showed a card on stage that might or might not have been assembled with woodscrews and some demos through a projector that might or might not have been rendered by the new cards.
Come May 27, you might or might not see a paper launch.

Don't get ahead of yourself.

Ridiculous.

nvidia have announced a release date, have cards out, AMD otoh haven't. Who showed off earlier is a moot point.

 

[CSI PC]

So you claim that the ID guys lied about running their DOOM Vulcan piece on a real hardware? Let alone dozens of reviewers who got cards as well?


Showing silicon and live demos doesn't mean real production (ie, in large quantities) has started, Silentguy's has a solid hepothesis as well, if they are ready indeed then why the delay?

And this is a key point.
NVIDIA are giving 1080 samples out now to reviewers albeit under NDA.
We are yet to hear any solid news when Polaris is being given to a broad selection of reviewers.
So at a minimum NVIDIA looks to be in a good position with the 1080/1070, and at a minimum at the same phase as AMD (being generous lets say they scheduled their news a couple weeks later for flexibility) and quite possibly NVIDIA is actually a bit ahead now.
All depends how AMD responds, but NVIDIA is looking good either way for having a product available.
Cheers

 

[Razor1]

First off, it's not a hunch, just an hypothesis that is still on the table. Meaning it's too early to reach the conclusions you stated in your previous post.

Second, I doubt it would mean a dramatic drop in revenue, because not many end-users keep up with GPU announcements and the GTX970 is still nvidia's best offer for the ~$300 segment (which the new cards don't cover, BTW).
Graphics cards don't get the same Osborn-effect as for example smartphones (whose announcements tend to appear on mainstream media).

There is a around a 2 to 2 and half month lead time for nV once they EOL products, for the channel to clear, as they try to keep their inventory around 60 days. This is what Silent_guy is getting at. If it was a paper launch, and we have to wait another two or three months for proper stock, yeah q2 of 2016 will have a pretty large decline as one month will have no sales (20-30% drop). We don't have exact days of the EOL of current products but we know it was around a month back so the March 27th launch date comes to the 60 days of inventory.

Any case NDA expires well before the 27th

 

[silent_guy]

reviewers got their cards after the event.

How do you know? That box could be full of wood screws.

 

[xpea]

Any case NDA expires well before the 27th

May 17th is when the official reviews under NDA will go online. But we can already expect leaks this week from Asia...

BTW, I'm surprised to see people so bitter about Pascal coming first to market. Looks like for some folks, it's not possible and AMD should be so much ahead because they showed very early silicon in closed doors few months ago..

How do you know? That box could be full of wood screws.

Wood screws review

edit: oups finally we won't get our wood screws review:
http://www.guru3d.com/news-story/geforce-gtx-1080-photos.html

 

[Razor1]

Well its not like nV missing anything for not showing stuff off earlier, it actually creates a bigger buzz the way they just waited and let AMD have a small ray hope and get people excited, then jump on the bandwagon and start creating a frenzy.