Ashes of the Singularity Benchmark
http://www.ign.com/articles/2015/08/17/fir...nchmark-results
http://wccftech.com/ashes-singularity-alpha-dx12-benchmark/

nvidia says this game is not matured enough to be considered to be a proper benchmark for dx12... so.... just take the results for now with a grain of salt.

This post has been edited by Moogle Stiltzkin: Aug 28 2015, 12:23 PM

from what i heard dx12 makes multi gpus much more compatible.

DirectX 12 Will Allegedly Allow Multi-GPU use Between Nvidia and AMD Cards

http://www.maximumpc.com/directx-12-will-a...-and-amd-2015/#!


if i'm not mistaken this also would mean 2 gpus with 4gb vram each would effectively be 4+4 = 8vram, rather than 4gb vram prior dx12


This post has been edited by Moogle Stiltzkin: Aug 28 2015, 12:37 PM

yeah multi gpu more bang for your buck. but i still wouldn't go multi gpu :} expensive. not to mention i don't think this will change multi gpus being more dependent on driver updates compared to single gpu solutions.

oh that sucks :{

so.... if wont work with amd gpus... then what about other nvidia gpus ?

when i attend this event, will see if they mention on that

are you refering to the one in singapore ? or was there alrdy one in my :/ ? i must have missed that one then

well it was speculated that nvidia does have async compute, but that most likely it's emulated, considering this recent issue by oxide, then that could explain their current results.

but that is still speculation, though a rather logical one judging by the results so far.

looking into this async compute issue.... regarding how nvidia cards would perform in dx12 games has got me thinking twice about pascal.

i'm just gonna have to wait and see what the reviewers say on that matter.

when it comes out and they test on ash singularity or another newer dx12 game, then can know whether pascal will last me a few years (with being able to play dx12 games well in mind) or not

if not then choice would be go for amd (which apparently was not a total flop because they actually did async compute the right way despite being a power guzzler), whereas nvidia probly did some emulation method that could do it albeit poorly

nvidia telling ash to disable async compute ... does not inspire much confidence

This post has been edited by Moogle Stiltzkin: Sep 2 2015, 03:23 AM

https://www.reddit.com/r/nvidia/comments/3j...nvidia_cant_do/


:/

dx11 nvidia clearly wins, but come dx12 with all sorts of performance tricks you can do, this may change needs closer look when benchmarks are out. but judging by all the technical mumbo jumbo, it's not looking good for 980ti

sure hardly any dx12 games out yet, but most people buy a card to last 4 maybe 5 years. so you wouldn't want to change out your card lesser than that just to play a dx12 game as well it should have been

maybe the performance penalty isn't too bad, or better yet pascal fixes this ? so have to wait for benchmarks

This post has been edited by Moogle Stiltzkin: Sep 2 2015, 11:07 AM

not sure about others. only cared about the best single gpu nvidia has atm

even with this dx12 concern when it comes down to async compute which a game like rts genre that spawns many units like AOTS was demonstrating how rts could be like by increasing efficiency/performance via dx12 is possible. the gains from async compute if the developer uses it, seems to be quite good on paper :]

will come down to fps and latency, whether the difference is huge to warrant switching camps






http://www.pcper.com/reviews/Graphics-Card...abs_block_tab=1


from the benchmark it just clearly shows the amd cards have huge jumps in performance in dx12 compared to dx11. so although the difference in end result is by a small margin between the 2 cards, it brings up the question, could nvidia performance had been much better had they used better async compute like amd ? nvidia async compute now being scrutinized for being inferior to amds judging by this result.

might be that in pascal the dx12 leveraging would be the same as maxwell, and this is what people waiting to be answered :/

but if the results is more or less same, dx 11 will be better than amd, and dx12 only slightly worse than amd, so nvidia may still be a better choice albeit disappointing they didn't fully leverage on dx12 async compute (cause the benchmark shows it does make a difference, if a game decided to use it e.g. rts especially).

for VR, people may prefer amd for the lower latency compared to nvidia. not sure the latency will be that big to be a problem for non vr games in regards to async compute

the real question is what gpu can play the games you want at what quality settings.

if you don't care about medium or low graphics, you can get the cheapest gpu on the latest platform e.g. maxwell etc.

i think the cheapest is a 950 gtx ?

but considering dx12 async compute, amd might be a safer bet for better scaling performance for dx12... but this is not conclusive just yet, but it seems that way so far.

i heard volta is the real big change in design compared to pascal. anyone know anything about this ?

i use OBS. so not really sure why would use the nvidia one

This post has been edited by Moogle Stiltzkin: Feb 2 2016, 10:26 AM

how much fps performance can i expect going from a 680 to a pascal ? playing Dragon age inqusition my 680 max setting cannot even handle 1080p resolution on 24'' lcd..... time for upgrade....


if not mistaken i heard the timeframe from pascal to volta may be somewhere between 1-2 years






http://www.anandtech.com/show/7900/nvidia-...ecture-for-2016








This post has been edited by Moogle Stiltzkin: Feb 7 2016, 12:40 PM

confirmation pascal async compute is hardware based?

in 8:32 he also implies they made changes in how async works compared to previous generation.

go to 11:37 in the video where somebody asked the presenter this question to direct confirmation if it's hardware async.



This post has been edited by Moogle Stiltzkin: May 18 2016, 02:52 AM

it also has 10bit hevc encode/decode.


i didn't see any mention for 10bit h264 i'm assuming it handles that too (decode at least)? anyone know for sure?

http://www.tomshardware.com/reviews/nvidia...cal,4572-4.html



problem with this though, how much would the 1080ti be? 750usd?

even if it was more reasonably priced, i doubt it will be offered very soon after 1080..... also by the time it comes out, volta will be just over the horizon.....

and even after all this, it's just speculation that there will be a 1080ti o_o; we don't know for sure.


also architecturally the pascal seems to be a maxwell on steroids. i suspect volta will be the biggest step when they finally go hbm2. there was never originally suppose to be a pascal.

that said the pascal on overview seems to solve most of my issues

1. 10bit support
2. hevc decode/encode
3. async compute hardware for dx12 games
4. better performance than a 980ti 20% ish difference ?



i'd probably get a pascal, but most likely an AIB model because i'm not gonna pay 100usd more for founders.... especially if i am most likely going to go water cooling. founder is pretty much paying for the fan cooler which was well made, but i most likely won't be using. Not to mention AIB customs will be clocked higher, have more cable to power the card for overclocking people speculated AND 100 usd cheaper...

This post has been edited by Moogle Stiltzkin: May 18 2016, 07:36 AM

looking at the recent performance charts, seems that a 1080ti would mostly benefit 4k fps performance using more higher settings.

but if i'm not mistaken, the 1080 can also do 30-40fps+ with high graphics settings at 4k. but if you can hit 60 fps+ that is always going to make a difference.

you can still do 4k with lower settings or leave aa disabled (which most people often do for aa for 4k anyway). 4k may seem doable as such on a 1080 without needing to wait on a 1080ti.

my point is, at most 1080ti will mostly be of most benefit for 4k gaming. but as far as lower res below 4k is concerned, the 1080 seems to get the job done just fine. so you'd be waiting on a 1080ti if it shows up, and you'd be paying presumably much more, presumably more than a 980ti, considering that the 1080 is priced higher than 980 at launch wasn't it?

speaking from the viewpoint of someone who buys their cards to last 4-6 year before next gpu upgrade also i'm using a 680gtx which is was released in 2012, so over due for an upgrade.


PS: i think pcper pretty much summarized the pascal as being not too far off from the maxwell architecture, especially that performance wise it's a 1:1 ratio if you had the same number of parts. But they increase performance by adding on more stuff ontop of the other tweaks. (in video scroll to 1:57 he explains it better)


as such could speculate that the bigger real change will be with volta perhaps? in terms of fps performance boost, also more bandwidth with HBM2 which 4k will most likely need (right now with 1080 their using compression to deal with the narrower bandwidth, also using gddrx5 for that slightly more bandwidth needed).

but even if thats the case, i already pointed out for ultra 1080p and other resos below 4k the 1080 and even 980ti will be good enuff. Also don't forget 8gb vram is now the new standard with the intro to 1080 and this is good enough even for the highest end games. 1080 edge over 980ti is that the former may have gotten async compute done right this time around. as for image quality 1080 also has 10bit HEVC encode which will work well with my Dell U2413 10bit monitor to reduce banding when watching my 10bit videos

my 50cents

This post has been edited by Moogle Stiltzkin: May 18 2016, 09:47 AM

i'm still waiting on reviews. seeing if async compute was done right. once thats settled then wait for founders edition to fuk off before i get the cheaper aib models

early reviews seem good so far though for my specific requirements

This post has been edited by Moogle Stiltzkin: May 18 2016, 06:35 PM

ultra settings 4k resolution tom clancy's the division at 42 fps... playable? yes, but it's not 60fps lets put it that way. but at ultra settings at that reso o_o; hm nice.


1080 is 25% faster than a 980ti if we were to roughly it up in terms of performance.




This post has been edited by Moogle Stiltzkin: May 18 2016, 07:46 PM

another reason to wait for AIBs. or at least wait for them to compare



Source:
http://www.tomshardware.com/reviews/nvidia...cal,4572-2.html

This post has been edited by Moogle Stiltzkin: May 18 2016, 08:25 PM

but u need to remember, that is FOUNDERS edition. which is the 100 usd marked up early adopter tax.

Or maybe becauz founder edition comes with a good fan so cost more. But AIB custom that comes slightly later, the cooling shouldn't be much worse than that would it? Also they would tend to do stuff like oc the cards for slightly more performance too.

also since i personally will most likely go down water cooling route, doesn't make sense to pay more for the stock fan i won't be using anyway



anyhow of interest to me was this analysis of ashes of singularity (which is one of the earliest dx12 games to come out)



so in comparison to ATI's did the ati have any judder at all or is this a nvidia specific issue only? or is it that Nvidias async compute didn't really solve the issues fully ?



4k gaming at ultra settings. can it do it yet?



http://www.tomshardware.com/reviews/nvidia...cal,4572-6.html





This post has been edited by Moogle Stiltzkin: May 19 2016, 11:23 AM

what is the difference between nvidias founder edition vs aib reference?

would the aib reference have as good a oc potential as the aib custom designs? But to my understand the waterblocks will be less likely to be compatible with the non reference models.

any tips

PS: fan of ekb

This post has been edited by Moogle Stiltzkin: May 19 2016, 08:13 PM