TITAN X Pascal

>>55838509
>Synthetic CuDNN Benchmarks
Oh baby. Can't wait to pwn some n00bz at CuDNN with my new Titan P!

>>55838509
kek

What is CuDNN?

>>55839079
ML library by nvidia that's significantly faster than anything else on the market (everyone uses it).

>>55838541
>using gpus for games
Waste of resources, back to /v/ with you

>>55838509
>>Synthetic CuDNN Benchmarks Show Up To 200% Performance Increase

shit, gonna have to buy one of theirs for my self-driving car project

File: NVIDIA-Auto-Drive.jpg (99KB, 1574x985px) Image search: [Google]

99KB, 1574x985px

>>55839987

>>55838509
>No DX12
>No Async
>No Vulkan

>>55838509
AMDfags on suicide watch

>>55838509

nvidia is just sending off everything they have from paxwell to have 1080ti and a possible HBM2 version to counter vega, next architecture is built ground up

File: 1455686940528.gif (2MB, 500x282px) Image search: [Google]

2MB, 500x282px

>>55838509
The perfect waifu2x card

File: 1Ashes_01.png (274KB, 3984x1720px) Image search: [Google]

274KB, 3984x1720px

>>55841569
>No DX12
>No Async

>>55842556
Pascal is just tiny maxwell, it added some pre-emption stuff, but AMD hardware is built for A-sync, which is why it does so well when it's properly implemented ala DOOM

File: 1469825895018.gif (1020KB, 500x281px) Image search: [Google]

1020KB, 500x281px

>>55842577
we do good in a single game. WWWEEEEEEE

File: 1469846964388.png (498KB, 1920x1080px) Image search: [Google]

498KB, 1920x1080px

>>55842577
>DOOOOOOOOOOOM

It's all about the CPU...

>>55838509
>http://wccftech
I hate /g/ so much and I can't get out. Wait until the damn thing releases. Fuck. This probably isn't the full Titan, they're probably holding that down just in case AMD pulls a Hawaii again. That's probably the reason they went with the retarded Titan X name again.

/v/

>>55842752
>...
>generations old i5 at low clocks
Dumb shitposter. Everyone knows AMD has more overhead. They also have superior hardware but it has been suffering with this problem for years now.

>>55842752
63 vs 81

61 v 71

Looks like async works even in older CPU.

Hello guys, I dont want to make a new thread for this.
Can a 1060/480 upscale my chinese cartoons smoothly with madvr on 4k?

>>55842790
>>55842812
>you need the latest cpu if you want vulkan work well

I though the RX480 was for poorfags.

just wait for the gtx 1160 theres no games worth playing right now anyways

>>55842852
The RX460 is for poorfags. Anyone not running a at least a quadcore is mad.

>>55842884
The i5 is a quadcore...

>>55842071
>waifu2x
cancerous artifact-inducing meme

you have to be blind to think it looks good

>>55842752
so what about a real world test with like an i3 or FX 8 core? my FX chip at 4.4ghz can easily push past 120fps with my Fury X in it, obviously bottlenecked, but still great for my 144hz display

>>55842934
>...
Stop.
Let me clarify then. Anyone not running at least a Sandy Bridge quad core is mad. Even the FX83** can do some work when the APIs take advantage of its numerous cores.

>>55842752
>cpu bottlenecks
https://www.youtube.com/watch?v=TzdaOzxIsyI
it's okay when nvidia do it right?

>>55842970

>>55842852

>>55842981
>talk about vulkan
>post the witcher 3

>>55842577
>AMD hardware is built for A-sync

Why the 290/390 is faster than Polaris on Vulkan, DX12, A-sync and DX11 games?

>>55842994
hey i'm doing you nvidia shills a favour by using a gimpworks title aren't i

>>55843030
Tons of driver optimization

>>55843056
I will tell you a secret my son, the GlobalFoundry/Samsung 14nm is a mess

>>55843089
except it isn't

>>55843089
Not really, but it is their first run, the 5xx series cards should be more capable of hitting 1400/1500mhz

>>55843098
>>55843125
>14nm vs 16nm
>very poor performance per watt
>draw more power
>hotter
>low overclocking performance even with more voltage regulator module

>>55843159
Maxwell was already efficient, moving it to 16nm let it hit even higher clocks and maintain efficiency

GCN has never been good at overclocking, and is relatively power inefficient, the nano proved AMD could do it though, and the 480 is a lot better than the 290 or 390

Vega should be pretty solid if it's following fiji and has HBM2. plus they'll be more familiar with the process

File: fasc.png (52KB, 249x238px) Image search: [Google]

52KB, 249x238px

>>55842970
Why SB and not zambezi ;^) I thought amdrones don't want to give money to jewtel.

>>55843159
the "poor performance per watt" is hardware a-sync.
nVidia tried hardware a-sync in the gtx400 series, le housefires.

>>55842577
No Rajesh, Async increases performance in Doom but it was not the one that propelled it high, it was Shader instrinsics, read more on your AMD Shilling guide next time

>>55843319
>hardware a-sync
amdpajeets need it because moar coars stay idle and their sheduler is shit compared to nvidia

>>55843319
>nVidia tried hardware a-sync in the gtx400 series
Source?

File: 03hardwaref480.png (25KB, 630x576px) Image search: [Google]

25KB, 630x576px

>>55843319
wut

>>55838509
Even Nvidia's cards have too much tessellation.

File: lie.png (175KB, 353x318px) Image search: [Google]

175KB, 353x318px

>>55843319
WHAHAHAHAHAHAHAHA

>>55843381

The greatest victory is that which requires no battle.