https://youtu.be/xtely2GDxhU
>Starts at 2:13:00
NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
>NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
>NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
>NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
>NVIDIA BTFO WHEN ASKED ABOUT ASYNC COMPUTE
>>54614530
>BTFO
How so? The guy just said that he doesn't feel competent enough to talk about it, which is a perfectly honest and valid answer.
>>54614571
>Tom has been Director of Technical Marketing with NVIDIA since 2005
https://blogs.nvidia.com/blog/author/tom-petersen/
this guy is one of the best guys at nvidia, he answered every question perfectly and was comfortable enough to talk about anything and everything.... EXCEPT async compute
I WONDER WHY
>>54614624
Because he doesn't know enough technical details about it?
>>54614655
are you this naive anon?
apparently nobody at nvidia knows enough technical details about it
>>54614655
That really doesn't reflect well on him then. It's quite literally his job to know this stuff.
>>54614530
And?
It's well known that Pascal doesn't support it
It also won't be needed in the next few years since no one uses DX12 and Vulkan isn't really finished yet
>>54614720
He knows
He just isn't allowed to say it
>>54614772
>It also won't be needed in the next few years
thanks to plenty of TWIMTBP-shekerus
>>54614530
how can you not hate nvidia? they're literally dirty fucking kikes.
I guess I'll finally move to AMD gpu at least for y next upgrade.
nvidia is 2x faster than amd even without "async compute" lmao
>>54614772
>It's well known that Pascal doesn't support it
Which makes Nvidia liars. Pre-emption is not Async, and Async it is NOT built into their architecture. We know this because Pascal and Maxwell are identical.
>It also won't be needed in the next few years since no one uses DX12
Go away
>>54614571
LMAO
jesus wtf is this company doing
>>54616117
Needed 1500-2100mhz to do it
That's some shitty ass lazy engineering I think the 3garillion spent on r and d for this POS is a meme
>>54614789
he knows................too much. this is the last time he was seen alive.
>>54616185
Over clocking the same card over and over again since 2008
Its technically just a maxwell die shrink get over it.
That means that there are almost no architectural improvements, just higher clockspeeds.
So it btfo out of every benchmark and its literally the best card you can buy right now.
Out performs every single card and its not even expensive.
When amd releases a better card let me know.
>>54616350
>and it's not even expensive
>>54616350
It is nvidia's fault.
They hyped to hard and thus performance/price is not even close to expectations.
Nobody believes them anymore (3.5 and all other shenanigans). I bet my ass this Founders Edition bollocks is because they are selecting the best dies out there. 3rd party ones will perform worse.
>>54616390
>poorfag
Hey if you don't want the best performance available by all means don't buy it.
doom benchmark says suck it amd
>>54616434
The fuck? Every benchmark I've seen is around 20% over a 980 ti and that's exactly what they said it would be.
Did you think their presentation about VR "2x performance" was for non-VR or something?
>>54616461
970 - 71 fps
390 - 80 fps
Okay.
>>54616434
I dont understand. Did you preorder a card based on expectations? No.
The card is the best you can buy. I thought they would give me hookers and cocaine with a card purchase but they didnt omg that makes the card bad! No.
Best card available. Let me know when a more powerful card is released ok?
>>54616508
1080 - 173 FPS
>>54616461
AMD just put out new drivers that fixed it's performance issues
>>54616511
Vulkan isn't even out yet
>>54616511
irrelevant until next gen from both sides
>>54614530
>>Starts at 2:13:00
Why do this and not just include the timestamp in the link? I can't be bothered to look at it now mate.
>>54616511
Damnnnn the 1080 btfo doom. It really is the best card on the market. No other card will come close for a long long long time. Well you get that sort of performance when you buy the best I guess.
>>54616553
>2:13:00 one click away
>bothered to type 2 sentences
>>54616553
maybe he done know how... not everyone is perfect like you.
>>54616494
>>54616509
See >>54610739
At this point I do not even trust any review that does not come out from a poor guy that had to buy his GPU unit instead of having Nvidia sent one out to them.
okay
>>54616579
One click if I'd click correctly on a scale that is pretty large. Mousing over it to find it is also an extra effort that I do not want to do. I fully realize it's not a lot of effort.
Posting on 4chan is something that comes pretty naturally, and I'd say that typing a post out is easier than needing to find a correct timestamp in a video, when I could've just clicked the link and be done with it.
>>54616592
You don't need to be perfect to be able to understand how fucking youtube works, you just need to be not retarded. I assumed the people that would post youtube links on /g/ would know basic shit like that.
Can a GPU even go any faster without blowing up your monitor?
Async compute is irrelevant
GDC 2016 slide shows 5% gains and not worth the amount of work developers need to do for minimal gains
Pascal is FASTER than AMD meme async GPUs
>>54616627
WOW YOU TOOK THE FUCKING EFFORT TO TYPE SUCK A BIG FUCKING MESSAGE BUT YOU CANT FUCKING READ AND DRAG A SLIDER OVER. man stfu and stop being a woman.
>>54616603
Okay. We've seen video benchmarks as well as charts. I don't understand your point.
>>54616657
>WOW YOU TOOK THE FUCKING EFFORT TO TYPE SUCK A BIG FUCKING MESSAGE BUT YOU CANT FUCKING READ AND DRAG A SLIDER OVER.
Why do you think that? There's no reason to think that.
>>54616658
My point is that those card are selected by nvidia you fucking paid shill.
>>54616603
Doesn't matter who you trust.
Nothing will change the fact that the 1080 is the best card on the market. The best performance.
That's what it comes down to. Wake me up if a more powerful card is released, it will be from nvidia anyway.
>>54616680
All of the founders edition cards are selected by nvidia. You mad poor FAGS? Can't even afford the Best performance on the market haha lol
>>54616631
not much
>>54616680
Aren't all nVidia cards selected by nVidia?
>>54616753
Yes. Its called Quality Assurance/Control. When you pay for the absolute best card on the market you are guaranteed quality.
Poor amd FAGS don't know what that feels like, they just keep returning cards until they get a good one, that still isn't as good as a 1080 haha lmao
>>54616211
can someone tell the best one between EVGA, msi, Gigabyte and asus. can i use a msi card with an asus motherboard?
>>54616576
>256bit memory interface
>Current year
>699usd
No wonder it cant do 4K
its literally a 1080p memecard anything over that or 1440 and it chokes
i guess we will have to wait for the GTX3840ti with HMB3 and 900gb of ram
>>54616840
EVGA and yes you can.
>>54616840
The cheapest.
>can i use a msi card with an asus motherboard?
Yes.
>>54616882
It's the fastest card on the market. Nothing beats it. You have to pay more to have the best card on the market.
Sorry you cant exchange your food stamps for one lmao
>>54616882
>yfw it really is the GTX 1080p
>>54616645
It's not really that hard to implement. Most of the big studios are already doing it for the current gen consoles. It's not an exact comparison but it's thanks to async compute that current gen consoles with fucking 7670 tier power are able to look as decent as they do.
>>54616576
How much do you want to bet this is the same shill that always responds with TOPPEST LELKEK
Async compute really isn't the be all and end all. AMDfags are just fluffing it up because it's a rare occasion when their cards do one highly specific thing better than Nvidia, and they're going to keep reminding everyone because it's all they've got.
STOP
>>54616631
there was that nvidia driver that made card go 1000fps in menu of some game, card blew up
>>54616350
>so it blown the fuck out out of every benchmark
Nvidia buyers everyone.
Can this go 60+ fps at 1440p in the latest Tomb Raider or Witcher 3, with all settings at ultra?
>>54617142
no. ROTR is mad hard to drive with ultra settings which increase resolution of all effects to unpractical amounts
and you can't drive witcher with hairworks it eats away more than 20 fps
>>54616494
So that graph where they said a 1070 gets 200% faster performance than a Titan X was just a dream?
>>54617059
You sounds very salty anon.
Are you upset that nvidia couldn't rig the race this time?
Pascal is GeForce FX all over again.
I'd laugh my ass off and even buy one of them 4chan passes if Valve released HL3 and was Async compute heavy.
>>54616509
No they only gave the shill reviewers hookers and blow, that's why people on here are believing it's worth it to buy one.
>>54616576
>I'm a paid marketer pretending to be a 12 year old mexican - The Post.
>AMD Bundles Total War Warhammer With Radeon R9 390(X) and FX processors
is this AMD year? can you feel it? will we see greatest comeback in tech history?
>>54616882
Wait? Are you kidding me?
It doesn't seriously have a 256bit bus width still does it?
HOLY FUCK lilllllll
and people are buying these? wooooooooow.png
>>54617214
>people with jobs that can afford the Best card on the market make me jealous
>>54617251
Polaris10 will be best card on the market though...
>>54617242
>the memory is so fast so fast(c) huan
He forgot to mention that it's bottlenecked, again.
>>54617242
>ARE YOU KIDDING ME
>YOU'RE A WHITE MALE
Nvidia privilege is when you can afford the Best card on the market while poor FAGS suffer amd frame rates lmao
>>54616553
Is typing-start 2:31:00too hard?
>>54617266
LMAO
Nvidia perpetually BTFO. 1080 confirmed for total flop. Wow what a generational leap lmao
>>54617315
It would not if that was easy to do, however I would need to open the link in a different way to do that. Same problem really.
>>54617332
Wow it's the highest performance card you can buy. Poor FAGS btfo.
Best. Card. Available
>>54617332
No leap at all.
0% IPC increase.
0% New Features
0% Arch improvements
Pascal is literally a die shrink of Maxwell - which isn't bad in and of itself, but dodgy as fuck for nVidia to lie about it.
>>54617339
Well normally you play youtube videos likempv LINK, right?
Just play them asmpv LINK -start 2:31:00instead. not rocket science
>>54617350
Best single card available is PowerColor Devil13 16GB.
>lol at the shitty ProDuo though, wtf were AMD thinking.
>>54617350
Technically its not the best - according to TPU at times its outperformed by the 295x2.
>>54617362
>Well normally you play youtube videos likempv LINK, right?
Technically, yes. However, I do not have easy access to that piece of text. Assuming that I do, and moreover assuming that I play youtube links with mpv is a baseless assumption.
>>54617171
ROTR and Witcher 3 both can go 60+ at 1080p with a 980ti. This card is shit.
>>54617376
>>54617376
It actually is the best because it outperforms every other card in 9/10 benchmarks. But hey if you want worse frames in every game just don't buy it.
This card is for people who want the most fps and can afford it. Poor FAGS can stay with amd and keep playing in windowed 800x600.
>>54617385
>assuming that I play youtube links with mpv is a baseless assumption.
Is it wrong to assume a base level of technological competence from /g/ - Technology?
>>54617445
No. Assuming somebody plays youtube links with a specific media player is not assuming a base level of tecnological competence though. There are many valid ways to play those, not all of which rely on your specific way accessing mpv.
>>54617360
That old roadmap from like 2010 that adored showed had no mention of Pascal. It's basically stopgap: the gpu.
>>54617360
Jesus Christ can nvidia recover from this
>>54617360
There are several new features, such as simultaneous multi-projection.
The GDDR5X is insanely fast.
The power efficiency, temps, and clock speeds are a miracle of engineering.
>>54617518
Rofl get out of the thread Tom
>>54617518
lmao, go away shill, you're not even trying to look like one of us
>Nvidia
>releasing cards
Enjoy waiting for october nvidiots :)
>>54617518
>The GDDR5X is insanely fast.
That isn't a Pascal feature though, that's just specifying Pascal boards with nice memory.
>There are several new features, such as simultaneous multi-projection.
Software features that could be ported over to Maxwell if nvidia wasn't too busy cucking Maxwell owners.
>he power efficiency, temps, and clock speeds are a...
Product of TSMC's 16nm process - like I said Pascal is a Maxwell die shrink.
That isn't bad, die shrinks are great.
What is bad is that nVidia feels they can't just be honest about it just being die-shrink when most people would welcome it even without the lies.
>>54617581
Whats bad is that its just an oc'd 980ti and it wont be available for MONTHS
>bought a nice 10bit 4k monitor on a sale couple months ago
>holding out for new gen of cards
>1080 can't do 60fps at 4k on current games
>intothetrash.jpg
I guess I'll just wait 'till Computex for AMD and buy whichever card is the cheapest with the outputs I need.
>>54617662
No game can do 10 bit anyway
>>54617662
>reviewers enable 8xmsaa
>wonder why performance is shit
Doesn't help that nvidia is fairly shit at msaa in general, but thats a different matter.
>>54617844
I tend to keep my tech for a long time without replacing. Usually until it shits the bed and dies.
>>54617858
All I want is to play Witcher in it's maxed out 4k glory. And maybe Cyberpunk 2077, whenever they decide to release it. Is that too much to ask...
>>54617986
> Is that too much to ask...
If you are retarded with your AA settings, yes. Another good one is metro LL at max settings at 4k - max includes SSAA and you'll cripple cany single gpu for a looooooong time with it enabled.
Pic sort of related.
>>54617858
>Doesn't help that nvidia is fairly shit at msaa in general, but thats a different matter.
Because their texture bus is stupidly slow. Nvidia cards are pain and suffering when you're bound by the texture fill rate
How many years until amd catches up? Will they release a 4k 60fps capable card in say 5years?
Enjoys those Maxwells now guys
You know, before Nvidia makes their yearly 'optimizations' for you.
>>54618566
>Will they release a 4k 60fps capable card in say 5years?
Will Nvidia?
>>54618652
They already have.
>>54618687
Not according to most reviews of the card at 4k.
>>54616682
you're not wrong. The point of contention is that it's not THAT much of an improvement. It's certainly not worth upgrading from a 980ti, questionable if it's worth upgrading from a 980.
If you have say a 970 or a 780ti, then yeah, probably worth it.
>>54618566
>>54618652
>>54618687
>>54618775
Cards aren't “4K-capable” or not. Cards just provide X amount of raw performance.
How many FPS you get out of that X amount of performance depends solely on how demanding the games are.
As long as 1080p is the dominant resolution, and people continue demanding higher graphical quality in video games, video games will continue to bump up their graphical settings to the point where it will still run smoothly on 1080p.
If you want GPUs to magically become 4K-capable, then you need to make 4K the dominant resolution among consumers so developers have a reason to target that resolution.
>>54618808
Well if you want the increased performance right now you need to pay for it. I mean 980ti or fury x isn't the cheapest of cards so if you already shelled out big on those it might not be worth it but we need to see what price non reference cards will be, msrp is 100 less than foundry.
>>54614772
>>54616645
>>54617059
>6 months ago
>nvidiots tell me dx12 wont matter, its years away amd will die
>async and dx12 start to matter like everyone said they would
>dx12 and async are memes, they wont matter for years, nvidia is good enough even without async
i cant fucking believe this shit
>>54619035
>Cards just provide X amount of raw performance.
Yes and not all settings are computationally equal - especially things that rely on rendered resolution and take multiple samples of it. Funnily enough like msaa, or ssaa and a host of other settings.
Driving 4k at playable framerates has been doable for a while for high end cards, but idiots refuse to understand some things should be disabled as they provide virtually nonexistant IQ increases but have enormous performance penalties.
>>54619194
Do MSAA and SSAA provide a greater relative performance hit at higher resolution?
I would naively expect that, say, 2x SSAA just makes rendering 4x slower.
>>54619052
1080 is the fastest card on the market. Nothing can beat it. How does it feel to own cheap trash like amd?
>>54617858
msaa is shit anyway (textures, transparency). your card either has the power to do ssaa, or you don't use any aa.
>>54619323
SSAA basically exists so that 2K monitor owners can feel a bit of what it's like to have a 4K monitor
If you have a 4K monitor, why would you enable SSAA?
>>54614624
Not to be an Nvidiot but
>Technical marketing
>MARKETING
>>54619388
>If you have a 4K monitor, why would you enable SSAA?
depends on your screen size and viewing distance. if you have something like 4k 40" and sit no more than half a meter away from the screen, then you'll notice aliasing. with something like 4k 28" it's more or less useless.
>>54619263
Note what they mean:
Multi sampling anti aliasing
Super sampling aliasing
MSAA will render a scene multiple times and take the "best" (in laymans terms) of the factor you are rendering by (2x, 4x, 8x) etc. As screen resolution increases that means rendering many, many times more pixels. Note MSAA doesn't cover everything (its complicated to explain as you need to have some basic knowledge of the rendering pipeline).
Super sampling renders everything at four times the base resolution and then scales down to fit. SSAA at 4k is rendering an ungodly amount of pixels which will kill any gpu attempting it. SAA is the best method of AA for straight up removing jaggies but equally it has the highest performance hit and is somewhat archaic when you get down to it.
Both of these types of AA are why post process AA (such as fxaa) are so popular - they do a good job of removing jaggies and are nearly computationally free.
>async meme
what for and what game ?
geforce doesnt really benefit from async compute the same way gcn does. its practically useless. 2-3% at most.
gcn gets around 10% boost, which is neat.
>>54619821
>MSAA will render a scene multiple times and take the "best" (in laymans terms) of the factor you are rendering by (2x, 4x, 8x) etc
But that's false. MSAA just averages out the pixels in order to prevent moiré aliasing.
>As screen resolution increases that means rendering many, many times more pixels.
But that's a contradiction
>its complicated to explain as you need to have some basic knowledge of the rendering pipeline
I have basic knowledge of the rendering pipeline
>SSAA at 4k is rendering an ungodly amount of pixels which will kill any gpu attempting it.
It's still a constant factor..
Did you even read my post or are you just quoting random shit? Explain how SSAA and MSAA some how get “less” efficient at 4K compared to 1080p.
Geez AMD cucks, I didn't want THAT much salt to my fries
>>54616882
It supports 4K...
>>54622852
>>>redd*t
>>54623016
There are cancer containment boards for cancerous people, you belong there.
>>54623134
>There are cancer containment boards for cancerous people, you belong there.
this entire post, including its poster, belong in the cancer disposal bin.
