Nvidia GeForce GTX 1000 Series (Pascal) MegaThread: FAQ and Resources

Page 76 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar


No need for that many models. Probably they just thought of a way to make more money.
 


Wow, so the GTX 1050 will equal GTX 970 performance. That's pretty sweet!
 


Thank goodness. That means the 750ti/950 will get an actual successor that should be a large improvement over their previous models.

The 750ti has seriously been one of the best bang for your buck cards out there, able to max out the popular cheap/free games that people are fond of, but also letting your run the fancy high end ones like GTAV and Witcher 3.

If I was more cash strapped before i built my 750ti (mk2-m in my sig) system and was building my 770(mk2-r) system I would have been completely satisfied with only the 750ti. Having both and using both, the differences between them were minor enough that if you're not a graphics snob you wouldn't be hunting for a new GPU.
 
it makes sense really. a 1050 would have to be so weak to not get in the way of a 3 gb 1060 model it would not be worth releasing. but a 1050 model roughly a 970 in performance is right in line with the others nvidia has released.
 
So if the 1050 will replace the 970, what price should we expect? Sweet $150 and less, or nVidia inflated $200+? For $150 it could potentially take a big bite at 480's market share, me thinks.
 


I don't think they will be that worried about interfering with the 900 series sales. They didn't worry about hurting the 700 series when they released the 900 series. The 900 series cards have likely been out of production for a while and everyone knows that the 1000 series cards are here to replace them.

$200 may very well end up being the price but I don't think it will have much to do with the 950 and 960. Something in the $175-$200 range would be expected since all of the 1000 cards are more expensive than their 900 series counterparts.
 
So I thought I read that Pascal wasn't really using Async Compute and that they were using some other proprietary feature instead. However, Rise of the Tomb Raider was patched last week to add support for Async Compute and multi-GPU support in DirectX 12 mode. According to the patch notes thread on Steam Rise of the Tomb Raider now supports Async Compute on AMD GCN 1.1 and later cards and Nvidia Pascal cards but not Nvida Maxwell and older.

So, what's the deal? Does my new 1070 support Async Compute or not? If it does why are the weaker AMD cards still performing so close to it in Ashes of the Singularity?
 
it supports async but not as well as amd. that's all it is. amd uses it so well, that cards get a big boost in performance if it is enabled. this is why you see lower end amd cards catching and beating "better" nvidia cards when async is enabled fully.

amd banked on this a long time ago and built it into the cards while nvidia ignored it. now that bet is paying off the more games using async. if it does not use it, then of course nvidia still has the edge elsewhere.

nothing you can do about it. get amd and hope they use async or get nvidia and hope they don't. of course this only applies if you care how other cards do vs yours. i personally am just happy with what i got and don't look for a reason not to be. you're 1070 is still a great card and performs well :)
 


Yeah, don't get me wrong. I'm completely satisfied with the way my 1070 is performing on my 1440p monitor. I just didn't think it supported Async Compute. That's why I was pleasantly surprised to read those Tomb Raider patch notes yesterday and see that Async Compute was added for Pascal cards. It's a bummer that it isn't supported for people with Maxwell cards though.
 
Also there's like, no real a-sync compute games anyways, and AOTS is like the only example of a game with it working and not being a patched in mess like tomb raider.

So it's really just a matter of adoption, no reason to buy(or make) an async compute card if there aren't programs that actually use it.
 


Actually since Async Compute was patched into Tomb Raider last week people are reporting some nice performance gains in the Steam forums. It sounds like the DirectX 12 performance is finally better than DirectX 11.


 
reviews i have seen show 30% improvement for amd cards!! that's pretty significant. nvidia saw a bit as well with pascal but not near as much.

and maxwell won't see any gains cause the support is not in the gpu itself. software can't change the hardware that it runs on. it's clear pascal has added better support but it is not as good as amd uses for sure. each has its strengths and that happens to be amd's.
 


Could you please post a link to the "reviews" you mention? I cannot seem to find any that include the GTX 1070/80 and RX 480. Thanks.
 
For those interested in seeing some numbers from 3DMark's new DX12 benchmark tool, Time Spy, here are PC World's results. Unfortunately the 1080 wasn't tested.

http://www.pcworld.com/article/3095301/components-graphics/3dmark-time-spy-tested-we-pit-radeon-vs-geforce-in-this-major-new-dx12-benchmark.html

time-spy-results-100671645-large.png


Edit: This is interesting. They also have a comparison with Async Compute enabled and disabled. It looks like on the Nvidia side the Pascal card does gain some performance with it enabled. As expected the Maxwell cards do not.

time-spy-results-async-vs-no-100671644-large.png
 


i'm trying to remember which new thing i made that remark about. 🙁

been a long couple days. was this in reference to the doom vulkan add-in or the tomb raider dx 12 patch? was gonna link a review for you but then i forgot what i was supposed to be talking about there.
 


Ashes was build with GCN hardware in mind from the get go. before they transition the game development to DX12 the game was build using Mantle.
 


more like only AMD the only one pushing for it instead nvidia ignoring it. remember tessellation? AMD has been pushing them for years but when tessellation finally including in direct x they need to be compatible with shader model 5 hardware making AMD early investment did not pay off on their older hardware. good thing for DX12 was you don't strictly need new hardware for it. if DX12 strictly need new hardware then not even GCN1.2 will be compatible with DX12 despite having the hardware for async compute. also AMD got the advantage of being in console. as Crytek dev mention right now there is no standard way to implement async compute by GPU maker but since AMD hardware in the console most developer want async compute to work like it did on AMD hardware. though by the time volta arrive nvidia might have similar solution to AMD ACE unit. i still remember nvidia troubles when they first adopting GDDR5 with Fermi. they fix it with Fermi 2.0 and then later being the first to use 6Ghz GDDR5 with Kepler. and there is thing with the upcoming shader model 6. i think i would not upgrade my gpu unless we know more about SM6
 


No worries, I found the Vulkan API for Doom which gives a nice boost to nVIDIA and AMD cards. I'm not sure if the Tomb Raider patch has been benchmarked yet.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom