Nvidia:

Nvidia:

>1060
>1070
>1080
>now a fucking new Titan X Ultimate

AMD:

>RX480


Are AMD even trying? Are they giving up?

Other urls found in this thread:

anandtech.com/show/10433/nvidia-announces-pci-express-tesla-p100
youtube.com/watch?v=dTJPp7WdEBs
anandtech.com/show/10325/the-nvidia-geforce-gtx-1080-and-1070-founders-edition-review/9
youtube.com/watch?v=O7fA_JC_R5s
twitter.com/pellynv/status/702556025816125440
computerbase.de/2016-07/geforce-gtx-1060-test/4/
techpowerup.com/reviews/NVIDIA/GeForce_GTX_1060/24.html
radeon.com/radeon-wins-3dmark-dx12/
youtu.be/PF5j7KDb2Oo?t=303
docs.google.com/spreadsheets/d/1Q4VT3AzIBXSfKZdsJF94qvlJ7Mb1VvJvLowX6dmHWVo
twitter.com/AnonBabble

my thoughts exactly

Nvidia also owns the professional gpgpu sector.

they are bankrupt and finished. period

>More paper launches
>Massive shill campaign
Nvidia is scared shitless by Polaris.

really fires up your neurons, huh

NVidia:
>1080/1070 announcement.

AMD:
>Oh shit. Get as much power out of the 480 as you can. I don't care if we cannot get the final version through QA until three weeks from now. DO IT!

Nvidia:
>1080 a smash hit in power and performance.
>1070 a great success for the midrange cards.

AMD:
>480 a middling mess of PCIe overcurrent, missing its TDP, and uninspiring performance.

Nvidia:
>1060 released is an acceptable low cost card in price performance especially for its low power draw.
>TITAN X announced

Now I imagine it is likely at AMD:
>We need Vega to produce. I don't care if you have to turn it into our version of Fermi and use angel's blood for TIM and have to use 3x8pins get it up to at least somewhere near Titan X.
>Fuck I hope the 1080Ti doesn't actually have HBM2 memory.

>Make vega good, no matter what! Also do it fast enough so I can go home and sell my shares in AMD

AMD headquarters at the moment

When is Nvidia's earnings report? We'll see if they actually moved GPUs around or was it really a paper launch.

You know whats fucking weird?

This isn't the full chip

Pascal has a granularity of 640 shaders by design, but they can bin down by 128 shaders at a time, 1060 has 2 GPC, 1070 has 4 GPC - 5 groups of 128 shaders, 1080 has 4 GPC, this thing has 6 GPC

but if you do the math, this thing is missing 2 groups of 128 shaders

Thats actually the same number that was cut off the Titan X for the 980 Ti


It's going to be another fucking original Titan scenario, where they'll come out with something like the 1080 Ti which is fully enabled then a Titan X Black or some shit which is fully enabled with 24gb of memory or something retarded.

just wait for ____

- AMD

>but they can bin down by 128 shaders at a time,

SM size was changed to 64 with pascal, P100 and the new titan x both have 4 of the 60 SMs on the die disabled.

>dumb invidiot
>you

This is what I hate about nvidia. They make fantastic hardware, but their jewish business practices suck hairy balls.

wow you told him!

>thinking any other company is different
lel just lel kid

Sure

No it's still 128, not sure where you got 64 from.

I know others companies do it, but none do it quite so bad as nvidia. If nvidia were doing circumcisions, they wouldn't stop with the tip, they'd cut at the base.

>HBM2 memory.

nice meme

Can Nvidiots please just be banned from this board already?

Still won't help AMD's sales though, even if AMD has 100% shill marketshare on Sup Forums.

If Nvidia had real competition they wouldn't be able to get away with treating consumers the way they do.

Then again Intel doesn't have real competition either and they're not nearly as bad.

(You)

AMD is actually bankrupt and finished this time

so antisemitic, can't believe someone would say such a thing

(You)

I think even if nvidia had good competition, they'd still be using ultra-jewish tactics

AMD owns gayman consoles. And the mobile market.

Nvidia is a joke.

...

nvidia is rushing their cards because they know they can't dx12 and vulkan and async. They need to sell these non dx12 paperweight so they can build their Volta, which hopefully can async and dx12/vulkan

anandtech.com/show/10433/nvidia-announces-pci-express-tesla-p100

OY VEY

>consoles
AMD was the lowest bidder, they make cents on each console.

>Mobile market
What? You mean the other big aggressively razor-thin margin market?

>nvidia is rushing their cards because they know they can't dx12 and vulkan and async.
Why do you perpetuate lies? 1060 beats the 480 in Time Spy (inbefore nvidia biased benchmark) and gains 5.4% or so with async on in test.
youtube.com/watch?v=dTJPp7WdEBs

>AMD reports first quarterly net income since 2014
>Revenue up
>Cash reserves up
>No debt due till 2019
>beating analyst expectations again and again
>tfw owning AMD stock

>AMD reports first quarterly net income since 2014
>It's like $30,000

In the meantime, intel is making hundreds of millions a quarter

>not in stock for 5 more days
>$100 over their own fucking MSRP
you fucking tell me.

thats the beatuy of actually starting with the lowest cards at first...

you can literally make your top ones faster than them instead of gimping the low ones with no sli in order for you not have an internal competition..

you mean emulated async/pre emption? True async is nonexistent on nvidia hardware

AMD is the Lowest Bidder thats why consoles go for it.

they would lowball the as much as possible just to have that market.

AMD investors at this stage are real easy to impress, so he might be on to something here.

This might be the final hump in their stock before they fade off forever.

Depends which accounting method you choose to use. GAAP shows a net profit, whereas non-GAAP shows a net loss.

still beating analyst expectations
still on track for the non-GAAP profitability AMD targeted for Q3

oh so that's why AMD announced the 470 and 460 months ago lmao

>you mean emulated async/pre emption?
No I mean async.

>True async is nonexistent on nvidia hardware
>muh no true scottsman
> I’d consider the changes to work scheduling to be the other great change to the overall Pascal core architecture. With Pascal, NVIDIA has significantly improved their ability to allocate and balance workloads, which in turn has ramifications in several difference scenarios. But for the AnandTech audience the greatest significance is going to be in what it means for work concurrency when using asynchronous compute.
anandtech.com/show/10325/the-nvidia-geforce-gtx-1080-and-1070-founders-edition-review/9

>1070
>"mid range card"

yeah sure. unfortunately, for poorfags like me, I don't have the money for the so called "mid range card". the real mid range card for me that I'd go for is the 1060 or the RX 480. but since the rx 480 shows promising performance under Vulkan, I'll go with that.
Nvidia will make the 1060 slow soon, like they did with the 7xx series.
>daily reminder that your card will be good only for the next 1.5 years:
youtube.com/watch?v=O7fA_JC_R5s

>still on track for the non-GAAP profitability AMD targeted for Q3
Definitely. The majority of 480's are being sold as OEM to HP, Dell etc. Q3 earnings should be significantly higher.

do you see any top end amd gpu yet? no the only info we have about vega is that they will be from 15 to 18 billion transistors that is the only thing we actually know so far

AMD is simply incapable of making a top end gpu
>lol just glue two cards together and call it fury

twitter.com/pellynv/status/702556025816125440
Sean Pelletier from nVidia even says there is no async

You're right but he is talking about how the SMs behave like linked dual-64-unit clusters.

i dont know about that seems like fury x is far away nowdays than 980ti.. especially when you put dx12 and vulcan..

>is not enabled on the driver side.
What does this have to do with hardware support, ie. it is not on the driver side because it is on the hardware side?

>Mobile GPU Market
Nah, that would be various Chin Chongs, Qualcomm, Samsung and Apple with bongs' PowerVR

nvidia way of doing async is what time spy does..

literally making the paraller nature of dx12/vulcan into single path render..
thats why in every game you see maxwell cards regressing and on time spy they gain..

and thats why nvidia is saying time spy is the perfect example of async

Butthurt AMDshit

>no debt due til 2019
So why the fuck aren't they running as massive bill as possible on their hardware R&D sections?

Marketing goes so far when you don't have products on shelves and the competitors are cutting your fingers off.

>nvidia way of doing async is what time spy does
I'd suggest you actually read the article rather than make wild claims that are debunked in it.

>thats why nvidia is saying time spy is the perfect example of async
I don't recall nvidia saing TS is the perfect example of async. It does not take advantage of specific Nvidia optimizations just like it does not take advantage of AMD specific optimizations. That's what you get when you have both vendors working with FM to make a vendor neutral benchmark which both vendors had to approve of before it was launched.

And no FreeSync support. g-sync monitors are pricy as fuck.

What's the better brand for stuff like madVR/MPV video players?

>nvidia way of doing async is what time spy does
I almost missed this.
>inafternvidiabiasedbenchmark

They've either done two things:
>use basic compilers and programming schemes which might have code that does favor one hardware/ISA type
most likely
>go through every call and function to make sure nothing has a corresponding native instruction on a certain ISA
very unlikely

Truly "agnostic" code without the same time and effort to make fully optimized code is a logical fallacy

True Async requires either a hardware scheduler or Asynchronous Compute Engine (ACE). ACE is AMD tech however, so nvidia is left with the hardware scheduler. The last nvidia card to use a hardware scheduler was the Fermi series (GTX400 series). Pascal doesn't have it, but it has something a little different that works very similar to async, but is far less efficient.

Both brands are pretty good if you look at it from cost/watt/performance consideration if you don't plan on gaming. The mid-range to low-range cards handle video playback without using much power and emptying out your wallet. The 1060 is a pretty good example of that. There was a review which gave watts used in non-gaming circumstances but I cannot remember which one it was now.

is NNEDI3 exclusive to Nvidia? Yeah I know 1060 has better power consumption but if they are mostly "idle", they will be almost about even?

>hardware scheduler
You mean a dynamic scheduler which can concurrently load idle computer or graphics cores?

>Getting back to Pascal then, Pascal finally fixes the resource allocation issue. For Pascal, NVIDIA has implemented a dynamic load balancing system to replace Maxwell 2’s static partitions. Now if the queues end up unbalanced and one of the queues runs out of work early, the driver and work schedulers can step in and fill up the remaining time with work from the other queues.
Did you not read the article link provided?

>if they are mostly "idle", they will be almost about even?
I'm not talking only about when they are idle but also during simple web movies and shows. I'll look it up.

computerbase.de/2016-07/geforce-gtx-1060-test/4/

For example I am looking at this review, but barely understand german.

Found it. It was TPU.
techpowerup.com/reviews/NVIDIA/GeForce_GTX_1060/24.html

Alright, thanks.

Really want to try AMD for shits and giggles

I'm just stoked that the Titan X will take pressure off 1080 sales and reduce prices with higher supply. Will be getting a 1080 in September.

holy fuck the 1060 SMASHES poojet's card

>Titan X Ultimate
>Costs as much as 2 1080s
>Is slower than 2 1080s

Before rshit 480
>oooo rshit 480 going to be so cheap strong and good
>its expensive and shit

After rshit 480
>ooooo dx 12 and vulkan going to kill nvidia haha xd


Amd autists are so deluded and retrded

>Amd autists
I'd suggest you just call them AMDrones if only to avoid confusing them.

AMDrones can't read, they can only eat shit procreate and shill

You can't SLI 4 1080s

what the fuck is this game list

TPU is an nvidia shill site so they dont test vulkan or dx12 games

What's it look like? A collection of AAA games people are currently playing?

No AotS? How dare they!

What did they say about the new titan?

Where is my 1080ti?

radeon.com/radeon-wins-3dmark-dx12/

>they dont test vulkan
Why would they test Vulkan? For the niche market that is no way representative of the much wider gaming industry?

>or dx12 games
What is RotTR and Hitman (though they use DX11 since the DX12 version of Hitman is buggy on Nvidia hardware)?

True, but with no 4K 120Hz monitors and without 5K or higher really being an option either, there isn't much reason to go above 1080 SLI to Pascal Titan X SLI. Maybe if you have 3 4K monitors, but that's like the niche of a niche and 2 of these things probably won't be fast enough in that case either.

>radeon.com
I'm sure that won't be biased. . .

>the Radeon maker banked a positive GAAP net income and non-GAAP operating profit.

At this point it doesn't even matter on the high-end because the 1070/1080/new Titan X are so far ahead of AMD's cards it doesn't even matter if DX12 perf gain isn't that big. AMD needs to actually compete, 1080s appear to be selling like hotcakes.

No it's more

No Doom
No XCOM 2
No Dark Souls 3
No Total Warhammer
No Mirrors Edge Cat
No Star Wars Battlefront

but hey, fucking World of Warcraft, AssCreed (which is still broken, it wasn't just Unity), both Battlefield 3 and 4, the broken Bamham game, and Crysis 3

What the hell are you smoking? In some cases one 1080 unable to hold solid 60 in 1080p:

youtu.be/PF5j7KDb2Oo?t=303

1080 SLI barely enough for 4k@60. And you need to carefully adjust settings to get solid fps.

docs.google.com/spreadsheets/d/1Q4VT3AzIBXSfKZdsJF94qvlJ7Mb1VvJvLowX6dmHWVo

1060 vs 480 meta-analysis based on 240 benchmarks across 51 games.

Conclusions:

1060 is about 14% faster than 480 overall.
1060 is roughly equal with 480 in DX12 games.

>No Doom
>No XCOM 2
>No Dark Souls 3
>No Total Warhammer
>No Mirrors Edge Cat
>No Star Wars Battlefront
Why should they be added to the gaming suite?

ok

what

Vulkan works on Windows 7 and above, it's got double the market as DX12

>No Dark Souls 3

This isn't new, Nvidia released 970 and 980 first then Amd release the 390 and Fury the year after last time

I wonder what the difference is between his set up and Anandtech's where they were getting 114fps @1080p.

Take the R9 Fury X performance and double it. That's what you'll be getting with the new Titan X Ultimate coming on August 2nd.