>/v/
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
>/v/
>>53414576
nvidia is so good, looks like they win dx12 and dx11
>>53415266
Check the price of a 390, then the price of a 980 non-Ti then come back. Pascal is still like 6 months away and they're already gimping all their cards but the most expensive.
>>53415457
AMD's 8GB card can't even beat nvidia's 6GB. This is just sad.
>>53415500
>b-but async compute
>>53414576
>When using Directx 12 a Fury is so much faster than a 980 TI it's not even on the list anymore.
>>53417772
AMD can't run dx12
>>53414576
Also the new NFS
>>53415500
Nvidia's 6GB card can't even beat AMD's 4GB. The gimping is only begining
>>53418071
all I see is nvidia green at the top
>>53418071
>970 SLI 93 FPS above Crossfire Fury X 94 FPS
When will the shilling stop and the killing start?
>>53415266
>>53415500
I hope that you're just trolling
>>53418117
They base it on minimums.
>>53418117
Probably because crossfire didn't work
>>53415457
>Pascal is still like 6 months away and they're already gimping all their cards but the most expensive.
do you even have a shred of evidence to back up not only this but the gimping of kepler?
>>53418258
>gimping of kepler
Kepler's AMD counterpart (290/x) being Maxwell's counterpart now gives it away. Nvidia cards relies heavily on driver optimization and it's clear they're not putting enough work on Kepler
>>53418258
The post-patch FO4 benches were terrible for Nvidia. Go look them up. Some 700 series cards had double digit FPS losses. Nvidia banks on nobody seriously benching games a few months after they are released.
>>53418340
what are you saying? nvidia paid bethesda to ruin performance on all of their cards but especially kepler?
>>53418180
I've seen examples when even an nvidia card that has lower minimum framerate its above an amd card in these charts
![dvd_snapshot_00.02.03_[2016.02.11_02.25.39].jpg](https://i.imgur.com/VJa0cUi.jpg "dvd_snapshot_00.02.03_[2016.02.11_02.25.39].jpg")
Did I fuck up when I bought 980TI
>>53418526
I dunno, you tell me.
>>53418526
Check back in 3-4 months
>>53418526
No, AMD is shit. Nvidia cards are better.
>>53418569
You wish. AMD cards just mature slower and better, like fine wine.
Nvidia is like cheese, quick to mould and tastes like shit
>>53418548
>390 in the same price bracket as the 970, beating the 980
>390x in the same price bracket as the 980, creeping up to 980ti levels
>Both using an architecture from more than 2 years ago to compete with cards that they no longer bother to put in benchmarks in order not to make Nvidia look so bad.
Oh lawd, it's going to be pure comedy by this summer
>>53418291
Stop talking about things you know absolutely nothing about. The reason why kepler gets shit performance sometimes is because these benchmarks are using gameworks features which rely mostly on tesselation, which Kepler is very poor at. If you turn them off, the architecture will behave normally. It has absolutely fucking nothing to do with drivers.
>>53418340
>bethesda made patch = nvidia driver update guyze!!!
Retard. Besides, the issue was promptly fixed right after Bethesda's outstanding display of competence fucked performance with their garbage patch.
>>53418599
No, AMD abandons everything they touch.
>>53418963
Sure, that's why Pitcairn still gets performance improvements while Kepler fades into obscurity.
>>53418963
That explains driver support for my gtx 670 on windows 10!
>meanwhile Nvidia stocks are skyrocketing and AMDead are so cheap that nobody even buys them
Topcucks
>>53418358
>>53418960
>ultrafaggots
Nvidia doesn't give a fuck about optimizing games for Kepler once nobody is paying attention. Deal with it.
>>53414576
Do you GPU shills all get together and hate fuck each other? There's always a strange homo-erotic undertone to these threads.
>>53419897
>this much projection
When are you coming out, bro?
>>53418960
AMD cards are not good at tesselation too, so why they have better performance? Also benchmarkers normally turn gameworks shit off when doing tests so this excuse is wrong
>>53419918
>When are you coming out, bro?
When Pascal does. <3 8==D,----
>>53419959
don't forget project cars and physx
>>53418042
>mfw Nano's performance
What the fuck is the point of the Nano then?
>>53419996
for people with small dicks
>>53419996
Small builds
>>53415266
not hard to tell its a troll when the price difference is that large. better work on dat bait
>>53418548
Graphs like this always makes me sad that Linux users are locked into nvidia cards.
AMD, please release Linux drivers for your cards ;_;
>>53418664
No it won't. If something runs bad on NVidia hardware, NVidia will pay the devs not to use it. Since consoles are/will make use of async compute, expect Gameworks console ports to run even crapier.
>>53419996
Manlets with small dick.
>>53418664
Anyone can see NVidia literally pays developers to gimp AMD performance by including "NVidia-only" features that just happen to run on AMD cards and destroy their performance (lol, they can't even get those features to work well on THEIR GPUs). How else would they even remain competitive? They wish they were Intel in 2016, but in reality they're Intel when AMD was making the superior processors...
dx12 is shit
save we pray son of opengl
>>53420083
>using linux
>buys nvidia
thats retarded
>>53420870
Let's hear it then, anon. What vendor's GPU would /you/ buy for use in a Linux system?
Matrox?
>>53420911
Why the fuck would you need a gpu in linux?
>>53419996
Nano is fucking cute and based.
Faster than a 980 but same power consumption, especially at 4k.
>>53420936
High quality video playback on a 4K monitor?
>>53420969
you can do that on integrated
>>53420983
You literally can't. You're full of shit if you think you can run GPU debanding and EWA convolution at 4K 60 Hz on integrated graphics.
Also, Intel sucks absolute ass at writing drivers. I'd rather pay extra to get nvidia's driver quality than use i915.
>>53420950
>Faster than a 980
>>53418042
>>53421122
you are retarded, I am sorry for you.
>>53421197
>cherrypicking
I can do that too, lol.
>>53421409
>390X is still better than a Nano
>>53421224
>backpedaling this hard
I've spent a considerable amount of time working on a cross-platform OpenGL program for high quality video playback. I'm *very* familiar with all the fun driver crashes, kernel panics and render errors you get when trying to use various OpenGL features on various driver versions on various operating systems.
On Linux, pretty much nothing beats nvidia when it comes to stability / features / performance. They support pretty much the full OpenGL spec out of the box, they've never crashed on me, and the only time they fuck up performance-wise is when you accidentally hit a bad code path in the GLSL compiler.
Among the rest, you have your choice of:
>Shit performance
radeon (foss)
noveau
>Shit stability
fglrx
nouveau
>Shit featureset
all mesa drivers
>All of the above
intel (i915)
And for the reference, I've benchmarked a good number of cards and studied their performance trends carefully. For 4K 60 Hz realtime playback with 3-tap EWA (or higher) + debanding/CMS/etc, you need about a GTX 970 or R9 380.
>>53421409
game?
>>53421546
Forgot to add fglrx to “Shit featureset”
>>53415457
Any idea what price will pascal and polaris be at?
>preparing my butthole
>>53421546
you are talking about thing that have nothing to do with playing video
>>53421695
>high-quality video playback
>have nothing to do with playing video
okay then
>>53421637
The first Polaris card will be between $125-200 (It replaces the 390/X in performance, but at a much smaller die size).
The first Pascal card will be $300-500 and $1000 (I think both GP100 and GP104 will be launched within the same month, with the GP100 rumored to be about twice the performance of a GM200).
>>53421827
How powerful will they be compared to something like GTX 980 TI?
>>53421827
>$125-200
>replaces the 390/X in performance
does that mean I should sell my 290 now while it still has value and just get a Polaris? I don't game much anyway
>>53421867
You won't gain much in performance, if at all. The Polaris 10 will have about 90-95% of the transistor count as the 390X, so it might actually come under the 290 in raw performance. The main thing going for it is power efficiency. The Polaris 10 will probably only require a single 6-pin PCI power connector, if at all.
>>53421859
Polaris 10 will be about 80-90% of the 980 Ti (given that it's roughly identical to the 390X). The GP104 will be either at least above 10% of the 980 Ti's performance, probably more. There is no formal word of the GP100, but it might be nearly twice the 980 Ti's performance at stock speed. But at a $1000 price tag.
>>53421719
It does make sense that a retard like you would use linux.
>>53421503
>330w 2 year old card outperforming one year old 170w card
WHO WOULD HAVE GUESSED
