"Nextgen" GPUs

[azarhal]

I kinda miss when test results were presented in FPS and not %. See saying that a card output 20 more FPS is a lot more informative than this card does +30%...especially when +30% is 5 or 10FPS more.

 

[JDR13]

I think I'll be sticking with my 980 Ti for now

 

[Kordanor]

Never saw a flood of the same news like that.

15:00 Youtube Notification - Linus TechTips
https://www.youtube.com/watch?v=2-61Zn4Sb8Q&feature=em-uploademail

15:02 Youtube Notification - Paul's Hardware
https://www.youtube.com/watch?v=kYdzZzSzF2I&feature=em-uploademail

15:02 Youtube Notification - JayzTwoCents (about SLI though, which was also under NDA)
https://www.youtube.com/watch?v=2wBDt9tN5-c&feature=em-uploademail

15:20 Youtube Notification - Digital foundry
https://www.youtube.com/watch?v=RqK4xGimR7A&feature=em-uploademail

And then I checked the news...
Gamestar:
http://www.gamestar.de/hardware/gra...x-1080/test/geforce_gtx_1080,987,3272373.html

PC Gamer:
http://www.pcgamer.com/gtx-1080-review/

Tweaktown:
http://www.tweaktown.com/reviews/7703/nvidia-geforce-gtx-1080-review-hail-king-baby/index.html

WCCFtech:
http://wccftech.com/nvidia-geforce-gtx-1080-review/

TomsHardware.co.uk:
http://www.tomshardware.co.uk/nvidia-geforce-gtx-1080-pascal,review-33557.html

TomsHardware.de:
http://www.tomshardware.de/nvidia-geforce-gtx-1080-pascal,testberichte-242111.html

Guess you can spend the rest of the day watching test videos and reading reviews.

 

[joxer]

I think I'll be sticking with my 980 Ti for now

GTX 1080 is not supposed to "remove" previous Ti card.
Only someone crazy would buy 1080 (non-Ti) if already has 980Ti.

I'm going to bet nobody tested with Async compute either.

All published Ashes tests were removed fast. Not sure why, maybe were not done properly, but also maybe nVidia is hiding something. So wait a bit.

.

There is maybe a bit paranoic article on GTX1080, but with some valid concerns:
http://www.overclock-and-game.com/news/pc-gaming/46-gtx-1080-what-s-not-being-discussed

Some concerns there are, at least to us here, irrelevant.
For example dx12 benchmark for Tomb Raider 2 is broken? Like we care. We need here TW3, FO4 and DX:MD benchmark that shows frames per sec with everyhing maxed and turned on.
The article is well worth reading though as it's about PR bullshit on stuff noone cares about and hiding the actual fact on default clock while keeping mouth shut on exact things people want to know BEFORE buying.

 

[azarhal]

There is no reasons for anyone to upgrade to the 1080 unless you have a psychological need to make the highest FPS. Its biggest boost in performance is at 1080p where the high ends are already sufficient and cheaper cards (1070/Polaris) will release in a few months that will make those high FPS cheaper to "afford" for everyone else.

All published Ashes tests were removed fast. Not sure why, maybe were not done properly, but also maybe nVidia is hiding something. So wait a bit.

I saw some after I posted that comment. The GTX1080 gets a small 3FPS improvements which could just be FPS fluctuation. At 1440p in DX12 with Async Compute the Fury X comes at 5FPS of it. Going by the clock difference between the two cards, the 1080 is struggling…

Everything point at Pascal not supporting hardware DX12 Async Compute like Maxwell did. I'm wondering if Nvidia thinks they can get away at making DX11 video cards for a few year still or if this is just a throw away generation and next year they are going to release a new arch.

There is maybe a bit paranoic article on GTX1080, but with some valid concerns:
http://www.overclock-and-game.com/news/pc-gaming/46-gtx-1080-what-s-not-being-discussed

Some concerns there are, at least to us here, irrelevant.
For example dx12 benchmark for Tomb Raider 2 is broken? Like we care. We need here TW3, FO4 and DX:MD benchmark that shows frames per sec with everyhing maxed and turned on.
The article is well worth reading though as it's about PR bullshit on stuff noone cares about and hiding the actual fact on default clock while keeping mouth shut on exact things people want to know BEFORE buying.

Tomb Raider 2 DX12 is just a wrapper over DX11, it's basically adding overhead instead of removing it. Hitman might be a good indicator of DX12 performance until DX:MD though.

 

[azarhal]

Some German did a comparison of all the reviews data.

 

[Ripper]

It seems an odd move if Pascal really is weak in DX12. It could let AMD claim an easy win in DX12 with their Polaris cards, even if they don't match the brute force power of Pascal. We really need a more major DX12 benchmark game to get a better idea.

 

[Kordanor]

As long as there isn't really anything using Dx12 it's not that much of a deal though...

 

[azarhal]

It's not an odd move as much as Nvidia being unable to predict the future 4 years in advance.

1. Nvidia probably didn't expect DX11->DX12 to get adopted faster than DX7->DX9 and DX9->DX11 were (aka 3-5 years), but DX12 ended up very popular among developers and with it being backward compatible with 3-4 years old cards (even if not all the features are supported) and Windows 10 being free acceptance rate is fast (previous DX required buying a new OS and video card).

2. Nvidia used their superior driver team to get more FPS out of games. DX12 remove a lot of the driver overhead where that happened, Nvidia lost one of their advantage. Driver overhead happened to be one of AMD weakness too, so it helps them a lot…Most of the work is now in the game dev hands, which will probably bring other kind of issues.

3. Making a new architecture takes years, to have a DX12 card now they would have needed to start designing it 3-4 years ago. AMD basically had DX12 cards before DX12 spec were finalized because it is Mantle+.

4. The "last minute" 20/22nm die size skip probably had an impact in internal R&D. Although, that affected AMD as well.

 

[Gorath]

Some German did a comparison of all the reviews data.

That' actually a very credible source for detailed infos about graphics cards.

 

[Deleted User]

GPUs. Pay $600+, get to experience hair physics like you've never experienced before!!!

 

[GothicGothicness]

Well, maybe we should wait until AMD has actually released some cards before we assume they'll be better ?

 

[Kordanor]

I don't think that there is anyone who expects AMD Cards to each the 1070/1080 performance.

They probably have a 970/980 performance while being cheaper.

 

[azarhal]

I don't think that there is anyone who expects AMD Cards to each the 1070/1080 performance.

They probably have a 970/980 performance while being cheaper.

Some actually believe that the biggest chip (Polaris 10) will compete with the 1070 or even surpass it performance-wise, because of the rumor mills of the cards specs and the demo last January (even if we can't really validate that performance).

Then AMD always said they were going after the mainstream market (sub-$300) and that they wanted to make VR affordable which is very vague thing to say and could land them anywhere, performance-wise, above a 390/390x…for under $300.

Regardless of what happens, there is a catch: AMD is focusing a lot on DX12 and their card performance was always talked in relation to it.

Edit:
Seems like AMD themselves are kinda saying with this slide that Polaris 10 is replacing the 390/390x/Fury X.

 

[joxer]

AMD will apparently reveal and present their ace about the same day 1080 hits stores.
I'm eagerly waiting to see it, they don't have the "fastest card ever" but know how to give more bang for the same buck, so if their answer is cheaper than 1070 with the same performance, great news for us - graphics whores.

GPUs. Pay $600+, get to experience hair physics like you've never experienced before!!!

What are you talking about?
GTX 970 on newegg is cca $300. That card is strong enough to grow hair on the back of my characters without a sweat. If possible of course. In phonegame ports that simply isn't possible, those games have characters with plants and cardboard boxes on their heads instead of hair.
http://www.newegg.com/Product/Produ...ics Cards&Order=PRICE&Pagesize=30&isdeptsrh=1

GTX 1070, rumor is, although stronger will have even lower price when released.

 

[Deleted User]

I have a GTX 970 and I play a bunch of games in 4K. Apparently, though, I don't actually play them in 4K, because most people say you can't do 4K gaming with a GTX 970. Strange paradox, I know. Pretty sure I had Oblivion/Skyrim modded to the brim with upgraded everything @ 4K resolution and 60 FPS. But I also may have been dreaming.

 

[azarhal]

I have a GTX 970 and I play a bunch of games in 4K. Apparently, though, I don't actually play them in 4K, because most people say you can't do 4K gaming with a GTX 970. Strange paradox, I know. Pretty sure I had Oblivion/Skyrim modded to the brim with upgraded everything @ 4K resolution and 60 FPS. But I also may have been dreaming.

You were dreaming or playing very old games. All the benchmarks I see have the GTX 970 at around 35-45FPS at 4K on high settings in recent (post 2013) games.

That's totally playable FPS though.

 

[Deleted User]

You were dreaming or playing very old games. All the benchmarks I see have the GTX 970 at around 35-45FPS at 4K on high settings in recent (post 2013) games.

That's totally playable FPS though.

I played Oblivion and Skyrim fully modded with every graphical enhancement, which is still pretty taxing.

But you're probably right. More modern games at 4K wouldn't hit 60 FPS. Good thing I don't play many modern games and can enjoy my older games in 4K just fine.

 

[azarhal]

Looks like Nvidia didn't put that much effort in their GTX1070 marketing slides.

 

[joxer]

Seems so, but what if those are not incorrect?
It's not the same anyway, VR gain on x70 is one % (or 2?) less than on GTX 980/1080 comparing graph.