http://www.bitsandchips.it/52-english-news/6785-rumor-pascal-in-trouble-with-asyncronous-compute-code
Waitfags BLOWN THE FUCK OUT
>No HBM2 or GDDR5X
Just pick up a 390 today guys.
DELETE THIS
2016 year of AMD
>async compute
le dx12 meme, not this again
>>53697283
>he shills older technology for free
>>53697283
Nvidiots stuck in the past lmao
To be fair, AMD won't have HBM2 either.
the 680 was roughly 30-40% (40% on only a few) faster than the 580.) (http://www.anandtech.com/bench/product/1348?vs=1350)
we are comparing successors here. I.E. the 680 was the 580 successor
remember the 1080 is the successor to the 980. NOT the 980 ti. that's the 1080 ti's job
1080 being roughly 30 - 40% faster than the 980 (which would put it equal to or 10% greater than the 980 ti.
the "x70" has traditionally been 10 - 20% lower than the "x80" counterpart. this goes all the way back to the 6xxx series. I.E. the 6800gt was about 15% slower than the 6800ultra. just like the 970 is roughly 20% slower than the 980 (but the 980 has 30% increased price for that extra 20%! which has made the 980 a ****ty value)
if we take the 1080 being a full 40% faster than the 980
take the max 20% reduction for the 1070
that would put the 1070 around 20% faster than the 980 and 10% slower than the 980 ti
that's practically the 670 over the 580.
>>53697535
i think a lot of people forget about the last die shrink is that there wasn't a "ti" card in the mix. the 580 was the fastest single gpu solution on the market for nvidia. it was nvidia's single solution flagship, top dog. yes they had the 590 but it was NOT a single gpu card. it was a dual gpu card on a single pcb.
the 600 series? same exact thing. 680 was the top dog for a single gpu solution. not only that but both the 570, 580, 670, and 680 were the "big chips." they were the full fermi and keplar gpu's of their series.
starting with the 700 series this changed. the 780 was not nvidia's top dog. a new tier was added. the 780 ti followed by the titan was. maxwell 2 continued this trend but took it one step further with the 970 and 980 (gm204) not being full maxwell 2 chips. gm200 is the full maxwell 2 "big chip" and were exclusive for the 980 ti and titan x. the 970 being a cut down gm204 chip and the 980 ti being a cut down gm200 chip.
we went from "570 and 580" to "970, 980, 980 ti, and titan."
the "1080" is aimed at taking over the 980 spot. NOT the 980 ti spot. that's the "1080 ti" job.
you cannot compare the 1080 to the 980 ti since both cards are on different tier's.
another thing to note is what nvidia said: "pascal built on 16nm finfet process will offer 2x the performance per watt compared to maxwell 2." the 980 ti is 250 watts. half of 250 is 125. so that means, with no gimping, a full big pascal chip, at 125 watts, should offer the same performance as a 980 ti. the full pascal chip at 250 watts? that should offer the performance of two 980 ti's in sli. that's incredible.
FUCK ALL YOU JEW FAGS WITH NVIDIA AND INTEL.
AMD IS UNDERDOG AND WILL RISE TO THE TOP YOU WILL SEE. AMD IS SUPERIOR AND BETTER YOU WILL SEE.
>>53697555
seeing as the "1070" and "1080" are not the full pascal chips, that's being reserved apparently for the ti and titan, similar to maxwell 2's setup, the 1080 having performance near a 980 ti is perfectly reasonable. that puts it at roughly 30% increase minimum over the 980 since the 980 ti on average is 30% faster than a 980 with a similar tdp as a 980. also toss in the faster clocked gddr5 and increased memory capacity it will probably be even more faster. near beefy 980 ti overclock levels (extra ~7 - ~10%). rougly a 30 - 40% increase over the previous 980 that occupied its tier level. that's a big increase.
>So, GamesWorks optimizations will play a fundamental role in company strategy.
Pls god no.
>>53697513
amd has no future
>only one game so far relied heavily on async compute
>people act as if this is the future of games development let alone the near future
>one game starts applying overkill tessellation
nVidiafags: THIS IS THE FUTURE AMDTARDS. GET BTFO!
>one game starts applying overkill async compute
nVidiafags: THIS IS NOT THE FUTURE AMDTARDS. GET BTFO!
>>53697527
Async compute is more important and only high-end Pascal will have HBM2 and most people won't be able to afford it, every other Pascal is GDDR5X.
Async compute is a driver issue on gm200+(older chips lack the hardware) chips, new architecture isn't going to fix the lack of proper support for async compute.
>>53695518
honestly, who is surprised?
Nvidia stripped out harware shader scheduling in Kepler and virtually all fp64 in Maxwell.
> native fp64 rates
> GM200 (Maxwell) - 1/32
> GK110 (Kepler) - 1/3
> GF110 (Fermi) - 1/2
> GT200 (Tesla) - 1/8
> Hawaii/Fiji - 1/2
Their newer chips just have more and more functionality stripped out and emulated/hardcoded into host drivers, with the notable exception of tessellation, which they decided to go overboard with.
Async compute is a dynamic hardware control system that is antithetical to their recent design philosophy of stripping down the GPUs and handling everything in software with their TWIMTBP program.
>>53697555
Are you retarded? 680 was not the big Kepler chip. It was gk104. 780/ti were the big Kepler chips, aka gk100.
>>53695518
Indians won once again
AMD cares more about supporting marketing gimmicks then supporting their cards.
/thread
>>53699022
You're mixing up AMD with Nvidia.
>>53695518
390 nitro, already rocking it since Jan 2016, feelsgoodman
>>53699067
hey man, I totally appreciate all those millions of tessellated triangles in witcher hair and batman capes.
you can really tell the quality from the pixels
>>53699022
Worthless shill, why do you still come here. Don't you have some other agenda to fill?
>>53697929
18 PERCENT
8
P
E
R
C
E
N
T
>>53698829
not op but i think his point was that the 680 was the biggest chip for the 600 series which is true. the 680 was the 600 single gpu like op said flagship. the 690 was the fastest but was a dual gpu.
>>53695518
DELETE THIS
>>53699210
>1.7%
>>53695518
Well, Nvidia said that Pascal was pretty much a die-shrinked Maxwell with some added new features.
What were you expecting? A vast redesign for a stop-gap die-shrink generation?
The only 14/16nm cards worth buying are the ones following Pascal/Polaris
>>53697929
Buy low, sell high. Now is actually the best time to invest in AMD. I bought 10000 shares at 2.65. AMD has a lot of opportunities to turn things around. Moving from 28nm to 14nm.Zen with its 40% more instructions per clock. Polaris is shaping up to be a decent card.
And I'm no fanboy. I haven't purchased an AMD product since phenom II. But I' strongly hope AMD gets shit right this time; competition is good for the consumer. As long as AMD is shitting the bed, Intel can dick around with 5-10% performance improvements and bend us consumers over. If you want to see technology advance, then you should be hoping AMD stays in the game.
Fanboys don't understand economics and the importance of competition.
>>53699518
You don't understand physics.
>>53699518
>buy low, sell high
Cancer
Good stocks are worth keeping over time and reaping in the growth after ten years or more. A company that can not only stand the test of time, but grow faster than the pace of the market is the backbone of any good, dependable nest egg. AMD could be one of those stocks that will grow by about 3-5 times its value in ten years IF AND ONLY IF they make a significant breakthrough in the mobile market, the server/enterprise market, and automation. Relying on consumer and OEM sales for a contracting desktop market will make AMD go up and right back down again to its current value (except the price will be justified then rather than its grossly inflated state right now).
Nvidia is one of the major tech players that is putting their eggs in the right basket in automotive and industrial markets that rely on raw computing power with minimal power consumption. They may not grow as quickly as AMD in the same time period, but they'll at least have a stable increase in value as they gobble up market share from the likes of Intel and VIA.
Alright might as well post here
Dark souls 3 is forcing my hand to upgrade and I have no idea if I should buy a 390 or a 970. There's some pretty good deals for the 970 right now but the 390 seems pretty good especially because 8gb vram.
Size factor is also an issue for me, I don't have a lot of space and my current 760 barely fit into my case. So a smaller card is preferable.
Dark souls 3 is the only game I'm buying this year so I don't give a shit about running the witcher 3 4k at 144fps. So I care more about which card is more likely to better run that particular game than anything else.
Wat do
>>53699893
>meet the minimal specs of dark souls 3
>forced to upgrade
You don't even know how it will run yet. I played Dark Souls 2 with the exact minimum specs it listed and yet it ran perfectly with graphics maxed 1920x1080.
>>53695518
Why get a 390 when Polaris 10 is coming soon?
>>53700666
Why wait for Polaris when Vega is coming soon?
>>53700517
Yeah I kinda figured it's sort of a wild guess. Still, they updated the requirements this week and they list the 970 as recommended. 750 is listed as minimum so my 760 isn't much better.
Dark souls 2 was kinda running old tech but ds3 seems more up to current graphical standards so I don't really want to take a chance because I wanna play day 1.
Anyway so that's why Im kinda being a baby about it. 390 is really nice but the 970 is cheaper and nvidia usually has solid day 1 drivers for new games, thats why I'm on the fence.
>>53700687
What is Vega?
>>53700702
Fury successor, equipped with HBM2, allegedly.
>>53700687
Why get Vega when Navi is coming soonish?
>>53699893
>>53700692
Both cards will do fine. The 970 requires less power and if it's cheaper it may be the better buy.
Both the 390 and 970 are fucking huge cards though.
>>53700770
hmm when is this coming around the same time as the Polaris cards? That would probably be out of my budget as I'm looking for a 390/390x tier successor, or maybe a 980ti once it drops in price
>>53700924
>when is this coming
Early 2017, probably by the end of the first quarter
>>53700692
>ds3 seems more up to current graphical standards
Then get a 390, it has better support for dx12 and the like.
>>53700940
Yeah im not interested in waiting that long personally. But ill still be looking forward to seeing its performance.
Vulkan was recently finalized and is new meemee, who the fuck cares about dx12 and this async gimmick?
It's like how Nvidia was bragging about PhysX 24/7 years ago, almost no games actually used it and the ones that did often had horrible performance issues (BL2/TPS) so people shut it off anyway. And god knows Mantle was useless outside of Battlefield games.
Fact remains that devs are fucking lazy idiots and raw power will always remain the deciding factor in GPU quality
>>53701033
Approximately the same as two Fury Xs in Crossfire, but using the same power as one Fury X
>>53700876
Honestly it's probably going to come down to size regardless. I'll measure my 760 and try to find the dimensions of the cards. I know some manufacturers make slightly smaller cards sometimes so that's where I'm gonna go, whatever the card ends up being.
>>53700876
The 970 comes in an ITX format. I think ASUS and Gigabyte have single-fan sub 7-inch 970s for sale.
>>53701309
I'll check this out, thanks for the heads up anon.
wasn't it just a week ago that you faggots were saying directx 12 was going to fail and vulcan will be the best?
>>53695518
I already did this past xmas. A XFX R9 390.
Here's been my experience so far.
>Won't play Fallout 4. Compass doesn't work (gpu fault according to net so someone made a fix). Game shuts down after a few minutes.
>GTA5 completely fucked up. Game maxes the CPU to 100%. Still can't figure out why
These are the first two graphic heavy games I've thrown at it and neither work. Might be the CPU (i5 6600) or Win10 (directx12). Pretty sad this never happened on my Mac Pro '08 w/GTX680. Every game worked perfectly on it (through bootcamp and Win7).
>>53701540
did you install the latest drivers?
>>53701540
Played both of those games just fine on my 290
>>53698522
It isn't a driver issue. Nvidia lacks the hardware for Async compute.
>>53701540
Both those games run fine on my win10/gtx680 box
>>53697929
>amd has no future
They have 2-3 new chips under works, so they'll have products to sell, and they are also powering ALL consoles on the market (including all the confirmed upcoming ones, like PS4K, NX, and the Xbone update). And that 80-whatever % share in VR sets.
They most definitely have a future. They just aren't the big players they used to be.
>>53699893
If you need a small powerful card Furry Nano and FuryX are the only way to go.
>Async Compute Only Boosted HITMAN’s Performance By 5-10% on AMD cards; Devs Say It’s “Super Hard” to Tune
http://wccftech.com/async-compute-boosted-hitmans-performance-510-amd-cards-devs-super-hard-tune/
>>53700702
390/X and Furry replacement.
Polaris is the low/mid end part.
Vega will be out Q4 2016 our Q1 2017.
>>53701540
Get the newest beta drivers.
>>53701563
Already done all the usual stuff. Upgrade drivers, uninstalled/reinstalled, etc. Installed Raptor (ati's "geforce experience").
>>53701623
Yeah, I've noticed popular youtubers playing games just fine on their 290s. Not sure what I'm doing wrong here. Going to try 3DMark when I get home from work and load in Skyrim. Might just be those two games. Older Steam games and Guild Wars 2 work perfectly on it.
>>53695633
>>53697535
>>53697555
You keep fucking copy pasting this:
#1: the 780 Ti was the first time since 2003 that a top model carried the Ti brand
#2: Ti itself hadn't been used since 2003 until the GTX5 series
#3: There are two different types of Ti, where it's just a fully unlocked version, and where it's a cut down version of a card above.
You're absolutely kidding yourself with that assertion that the GTX680 is the GTX580's successor, as you say later there was a "big chip" Kepler. The GTX680 was meant to be the GTX660, until GK100 got delayed for a year, and released as GK110. The 780 Ti was also released later as yields on GK110 improved, as initially it was cost prohibitive, and the fully enabled dies were reserved for workstations. This is what led to the fucktarded naming schema change.
You do have it correct that the "1080", ie GP104, is the successor to the 980, but your performance estimates are off. It's going to have more shaders, but the clock speeds they're hitting on 16FF+ aren't too good. You'll should start seeing that come up soon, unless the new revision fixed it, which I doubt.
>>53700876
>>53701309
Fury Nano / X best chibi
>>53701079
Async isn't a gimmick like PhysX, it's a scheduling thing to more efficiently use that raw power. Think of it as GPU hyperthreading. It has small benefits on any implementation when enabled, but can be tuned per application for more benefits. Secondly it's in both Vulkan and DX12s specifications.
Also just like Hyperthreading, sometimes it can actually have performance impacts if it's done retardedly, like back in the Pentium 4 days when it first dropped.
>>53701772
>10% is the gap between the 970 and 980 roughly
lol
this'll be funny
So how do you feel
when you know
nVidia is GOING to get
BTFO?
>>53701079
Multi-Engine makes sure noting is idling in your system. It doesn't make computation tasks\games GO FASTA, it just makes sure nothing is slacking off. People here usually don't understand what it does.
AotS results are due to elimination of DX11 driver overhead, not Multi-Engine. HITMAN makes a better use of it, it gets from 5 to 10% increase in performance, intrestingly enough Hawaii benefit the most, not newer Fiji or Tonga.
devs also said it's really hard to get right
>>53702493
>Also just like Hyperthreading, sometimes it can actually have performance impacts if it's done retardedly, like back in the Pentium 4 days when it first dropped.
Those were some dark days.
so AMD this year?
>>53699893
just wait for 2 or 3 month, 760 can run DS3 on medium or low just fine. Seriously, next batch of cards will be a bit disappointing, but a lot of people will drop their perfectly good cards for a useless upgrade and retailers will try get them off the shelves.
>>53701481
Just the Nvidia fanboy faggots, they will say DX12/vulkan is ded until GTX cards have a-sync then they will sing it's praises
>>53702605
where did you get this graphic?
the coloring is nonsensical, and the its notions of what a queue and engine are differ greatly from everything ever posted about GCN.
> GCP drives RBEs, rasterizers, and CUs
> ACEs only drive CUs, but can handle like 8 command queues each
>>53702811
https://msdn.microsoft.com/en-us/library/windows/desktop/dn899217%28v=vs.85%29.aspx
>>53702831
OK, that makes more sense since it's just an introductory guide to synchronization/serialization in the abstract through the API.
it's still a retarded graphic given that the engine box labels look like they're items inside the actual engines rather than being the engines engines themselves.
/care
Gt430 masterrace
>>53697283
You say the AMD_ONLY_HYPERTHREADING_FOR_WEAK_CORES_FK_U_INTEL_AND_NVIDIA vulkan 1.0 extension?
>>53697527
Vega will indeed have HBM2
