Rx Vega official bench thread

>>61231078
So it's literally Vega FE performance.

>>61231078

>http://wccftech.com/nvidia-gtx-580-dx-12-benchmarks-geforce-384-76-geforce-400-500-series-direct-x-12/

Bad link
http://wccftech.com/amd-rx-vega-benchmark-leaked-specs-confirmed-1630mhz-8gb-hbm2-484gbs-of-bandwidth/

How does he know it's overclocked if there's no clocks to prove it?

>>61231078
Those are AIB partner cards they're comparing against.

>>61231588
And AMD will take the Vega sapphire xtrem ultra oc vs the 1080FE

>>61231877
I like my GPUs extra crispy, thanks. Vega partner AIB or bust.

>>61231104
Fast

>>61231078
You think amd will overhype again vegafail?

File: smug horror.png (121KB, 354x347px) Image search: [Google]

121KB, 354x347px

>>61231078
>still can't beat OC Ti
>probably comes with water block because FE can't hold 1600 on air for more than 10 secs
>mfw

>>61234447
Reference blowers suck, film at 11. The AIB partner versions will easily be able to handle it.

ITT unverified 3dmark scores are totally legit guise honest despite futuremark explicitly stating anything that isn't identified correctly is an unreliable result.

>>61234520
Obviously everything leak related is just something to speculate on.

>The "Fiji fallback driver" or "Fiji drivers" meme needs to stop. That's not how it works or should be described. Otherwise you should start calling Volta drivers Pascal drivers Maxwell drivers. There is obviously commonality -- that's just how software engineering needs to work for a GPU -- but calling it a Fiji driver or Fiji fallback is wrong.
>Rys, 2 minutes ago
AMD is mad.

File: vega-v-furyx-specviewperf.png (32KB, 808x594px) Image search: [Google]

32KB, 808x594px

Vega is really faster than Fiji.
The drivers are fucked up.

>>61234825
Sauce?

>>61235046
https://forum.beyond3d.com/posts/1990723/

>>61235046
B3D forums, ofc.

>>61234825
>implying that this somehow negates that the current drivers are dogshit for gaymen

>>61235096
It doesn't. He just said that it's not Fiji drivers.

>>61235117
of course they weren't fiji drivers, vega wouldn't even work with those

>>61235153
It works like Fiji right now though. Even TBR is disabled.

>>61234881
>Vega is really faster than Fiji
Do you see how stupid it is you even have to say that? Was anyone wondering if Pascal was faster than Maxwell?

AMD may fix it, but they still fucked up.

>>61235174
>Was anyone wondering if Pascal was faster than Maxwell?
No, people eat whatever shit nVidia sells them.

>>61235195

Sometimes its not even a full turd, only 3.5 of a turd!

>>61235393
You forgot the turd with 1.7% yields or turd that also doubles as a leafblower.
https://www.youtube.com/watch?v=qspdnAYiiug

On 1080Ti, wanted more from AMD than it seems i will get! FUCK! VOLTA WHEN?

>>61235431
>VOLTA WHEN?
You want Pascal refresh? Just buy pascal.

>>61235431
Next year

>>61235431

If you have 20 grand you can get put on the short list for volta.

>>61235452
>headless compute boards
How do i gaym on it?

File: oBKa5EK.png (316KB, 500x500px) Image search: [Google]

316KB, 500x500px

>>61235174
Release of FE was the most retarded thing in the history of retarded things. I start to believe they idea that they had a catastrophic bug (rasterizer?) in the first run of the silicon and this clusterfuck is the best idea they had.

>>61235441
Though Pascal was Maxwell refresh? And it was truly refreshing!

>>61235447
>>61235452
Next year then and with another price bump due to AMD pushing what little RND they have towards HBM.

>>61235487
Pascal was Maxwell dieshrink. Volta brings like nothing new besides memes.

>>61235493
60% performance gain from die shrink makes me have high hopes for those memes though, I do think there are people at Nvidia that knows how get shit done and crank out something atleast 30% faster than Pascal over the board!

>>61235530
Oh naive child.
Buy nvidia, gweilo.

>>61235530
what die shrink? 16 to 12nm is nothing like 28 to 16nm. hell, it is not even real 12nm.

>>61235541
I want AMD to push it though! Make the 1080Ti seem like a bad card for top performance, that would make me happy.

>>61235547
All the more impressive what they did with that generation shift!

1080 +5-10% for $380

Screencap this

>>61236351
I'd be happy with 1070 performance, if it comes at nano size and $300.

>>61236403
Nano a cute!

Especially if they don't price it retardedly like they did fiji nano.

All the fiji cards were pretty good after their initial round of price cuts.

If it's priced like a $350 1070 and has the performance of a 1080 we could have a goat high end GPU

But we know fagcoin miners will fuck up prices.

Can't wait to wait months for stock

>>61236690
they won't, vega is 10% above a 580 for mining

>>61236952
But 580's are expensive and hard to find. If RX Vega is cheaper at launch and easier to get...

well you do the math.

>>61237188
The entire GPU market is fucked for the time being. Perfect time to release a new one!

File: 1467296306101.png (91KB, 653x726px) Image search: [Google]

91KB, 653x726px

>>61231078

>>61237207
>Perfect time to release a new one!
Which will end up at 2x the price in the span of a week!

AMD should sit on RX Vega for another month, get drivers even better and wait for the miners to crash and burn

>>61236690
I doubt they'll get the efficiency anywhere near the nvidia offerings though, which is really a shame.

File: 1464020321394.png (337KB, 1064x698px) Image search: [Google]

337KB, 1064x698px

>>61235530

File: AMD-RX-Vega-GTX-1080-Ti-GTX-1080-GTX-1070-3DMark-Performance-Wccftech.jpg (332KB, 1832x1504px) Image search: [Google]

332KB, 1832x1504px

>wccfshit says RX Vega in "striking distance" of 1080ti

>Literally 20% behind

?¿? Really makes me think

>>61237838
>Engineering sample
>20% behind 1080ti
We saw the same shit with Ryzen at the beginning of the year. The engineering sample that everybody was basing their assumptions on in January/February was at least 10% slower than the final product.

>>61237890
Ultimate sandbag technique? I don't know if they can pull the same shit off twice.

>>61238031
vega FE has the worst gaming drivers known to a gpu right now, but in other workloads it performs better even at the same clockspeeds as fury X

>>61231078

doesn't matter

this thing will sell out day 1 thanks to buttcoins

File: SPECview Vega FE.jpg (115KB, 1280x720px) Image search: [Google]

115KB, 1280x720px

>>61238091
pic related

>>61238102
it won't, the FE performed like dogshit on mining ~30MH/s

>>61238091
>>61238106
Ok so 60% improvement in 3DSMax-05 and Maya-04, but it still actually needs the gaming drivers to be complete in like, a month.

It would be cool if they put anti-mining into the drivers. Sure it would get hacked eventaully but it would mean it would give a chance for 'normies' to grab one at a sensible price for gaming.

>>61238993
>god I wished AMD made this basic, ADDMUL slower, obviously it would never affect other tasks!

>>61238121
well, they clearly gave preference to the prosumer and enterprise folks, that's why the pro drivers run really good on the professional workloads or whatever they're called, seeing that they're selling vega cards for datacenter too, this seems like the biggest priority, so they delayed gaming stuff a bit more so they can finish the drivers

>>61238106
The raw horsepower is clearly there, but can AMD transform it into gaymen performance?

>>61239369
Trust in King Pajeet.

Just Believe.

>>61239369
someone just has to write the drivers properly, if they managed to do it with the pro ones, why not the gaming ones too, given enough time?

>>61239369
That's just compute performance. Gaming performance needs more than that, since it needs to actually, you know, rasterize pixels to the screen. And they've always sucked at that - they always had great compute performance that translated into average game performance.

On the other hand, Polaris with 32 ROPs can perform exactly as fast as Hawaii with 64 ROPs. If Vega can pull that off, it'll be great.

Of course if it COULD pull that off, then Vega FE wouldn't run worse than Fiji in games.

>>61239845
Thing is you can't increase your compute perf by a 40% average per CLOCK while regressing in gaming without a good reason, you simply can't unless you literally remove the ROPs

>>61231078
did they give an estimate on the price ? i betting 1070 price if it's really bad or 500 $ if it's 1080 level

>>61239845
The rasterizers aren't the problem. Even with Tonga Nvidia's discard was like 1.7x more effective. It's clear that AMD cards do a lot of unnecessary work and it's most evident in tessellation, where meshes are subdivided into millions of tiny triangles that wouldn't even fill a pixel on their own, just unnecessary data being processed More over their tri fillrate is historically bad (though it's related in part to the poor discard). So on that front AMD is doubly bad, just in raw throughput and discard

>>61240066
AMD is the first one aware of this, Vega was (originally) designed to remove these obvious geometry bottlnecks, and bandwidth ones(doubly more important for raven ridge)

If AMD literally did nothing concerning this in 2+ years, what the fuck have they spend 40% more transistors on?

>poor volta they said

Also this shit will draw a lot of power

>>61240187
On-die street shitting neural network

>>61235195
Except people will literally buy AMD no matter how mediocre it is
Doesn't matter how slow the GPU is, you will still buy it no matter what

File: AMDramspeed.png (1MB, 1266x669px) Image search: [Google]

1MB, 1266x669px

Reminder
Ryzen is faster at gaming than the 7700k if you use fast ram on the since the AGESA update

>>61240282
We're going to need a new scale for this level of butthurt. >>61240332

>>61240282
>thermi outsold evergreen
Gaymurs are brainwashed zombies that will buy Ncuckya no matter what, then complain that their GPUs are slow in two years and run out to buy another one. Rinse and repeat, forever.

File: 1486204416309.jpg (130KB, 728x546px) Image search: [Google]

130KB, 728x546px

FUCKING WHEEEEEN

>>61240336
Not really

>1080 performance in AMD shilled games
>Slightly behind in Nvidia shilled games
>Still 30% to go to 1080 Ti which still costs less than Vega.

It's literally like AdoredTV said so many months ago.

File: uvnn.jpg (133KB, 853x480px) Image search: [Google]

133KB, 853x480px

Remember when they had cool names for their video cards?

>just get a new desktop
>excited to play games on it
>Ryzen based
>fire it up, download library
>use my laptop later to buy some games
>hardware survey comes up
>exit out because I want to take it later on the desktop I now use for games
>play some games later
>no survey
>get on laptop again
>survey comes back up
Nice unbiased survey Steam.

>>61240383
nvidia has drivers poojet

>>61242717
Explosive ones-

>>61242698
>anekdotal evidence

>>61242880
>anecdotal evidence
>Steam survey has AMD CPUs tracking down while every other source has them tracking up
Sure, it's just anecdotal.

>>61242916
>anecdotal evidence

>>61238993
>anti-mining
When your shit locks up because Witcher needs CUDA for HairWorks, I doubt that you'd hold that sentiment.

>>61242698
>own i7 4790K
>steam survey appears constantly
>anecdotal evidence

Dude, why are you using 3DM11 as a benchmark for a card built from the ground up for more than just that.

Are you comparing processors used?

>>61244251
Because it's all we got right now. Leaked benchmarks.

>>61240383
Why does one autist keep screaming the same thing in every thread? Dide we get it, AMD released a good CPU years ago. That doesn't change the fact Vega launch is a fucking disaster.

>>61240383
>Rinse and repeat, forever.
The only difference being soon there won't be a choice to even get AMD.

>>61238091
Doubtful. People were will to forgive AMD's sandbagging of Ryzen because the price to performance was so good and consumers and reviewers were starting turn on Intel already with Kaby Lake. Nvidia has more consumer good will than Intel and AMD's marketing of Vega is nonexistant, Ryzen marketing was everywhere. Also AMD won't win by much in price to performance due to HBM2 costs.

>>61234447
It doesn't even beat an oc'd 1080 let alone a ti

Stupid Anime poster

>>61240336
Nice cherrypicked benchmark of one single game faggit

File: 1434401708963.png (228KB, 858x725px) Image search: [Google]

228KB, 858x725px

>>61236690
>580 is priced at 300$
>Vega has a huge ass fucking die, 50% larger than a 1080's
>also expensive HBM2
>"If they'll price it at 350"

Why are people this fucking delusional

>>61245045
Look at who he was replying to.

>>61235487
The refresh will be another die shrink from 16nm to 12nm+

>>61245287
I know, I was only pointing it out because the same guy always posts the same exact phrase in every single GPU thread over the past week or so, it's just weird that someone could be this butthurt over the HD 5000 series.

>>61235174
>AMD may fix it, but they still fucked up.
They don't have to outperform the 1080ti you faggot

As if I'm going to splurge money on such a card. The worrying part is the power consumption.

>>61245269
Motivated reasoning to create false hope in the face of evidence that AMD screwed up with Vega + the disgusting pricing of all GPUs thanks to miners.

Now is an absolutely awful time to be in the PC gaming market. Consoles are honestly a better buy at this point.

>>61245269
What makes you think cost of product has any effect on price?

>>61245435
Oh boy here we go, the levels to which AMDrones will go to indulge in their fantasies

File: untitled-21.png (57KB, 512x743px) Image search: [Google]

57KB, 512x743px

>>61231078
~27900 is almost exactly what a Fury X at 1630Mhz would perform like.

17997*1630/1050 = 27938

The 14% spread between the highest and lowest score is a typical spread for a GPU overclock

bottom line: it's still behaving like Fiji in this benchmark

>>61246992

this is what i dont fucking get.

its a new arch.
new memory,
2017.

it cannot be the exact same chip with just "better clocks" and better memory.

well 24 days to release.

>>61234881
>50% OC
>25% improvement
>200% power consumption
S-so this is the poower of Taj Mahal

>>61247314
new memory means that they get better yields since there's only 2 stacks of HBM. And the cards are now 8GB instead of 4.

I'm guessing the rest of the improvements is just about making it capable of reaching higher clocks.

>>61247579
Fury X is 275 watts. We don't know what RX Vega is going to run at but doubt it's 300 watts.

Your entire shitpost makes no sense.

File: HNI_0046.jpg (14KB, 431x431px) Image search: [Google]

14KB, 431x431px

>>61245435

>>61247314
>new arch.
AMD lie, Vega is Fiji 1.5 with HBMeme2

>>61248929
Clock the Vega FE at the same Fiji clocks for a decent estimate, it really isn't just Fiji at a lower process node. That or we could just fling shit at each other, maybe someone will take over.

>>61238106
>>61238091

Can't wait until this thing comes out, runs like ass as all the benchmarks are showing and then years down the road when AMD learns how to make drivers for it - which should have already happened - the AMD shills will then go and "SEE, AMD AGES LIKE FINEWINE(tm) EXDEE".

>>61249305
Even if it takes that long for them to write proper drivers (not likely), by then all the 1080's of the world will have been relegated to the obsolete bin of GPU history. Meanwhile, Vega will still be trucking.

>>61249422

Yeah, you just have to endure years upon years of bottom of the barrel tier support and problems out the ass to the point that actually using it becomes a massive chore before it happens. Totally worth it yo!

>>61249526

Uh huh sure, because the r9 200 series always had amazing drivers, amirite?

>Keep riding that upgrade train, maybe someday Nvidya will make a GPU that won't be trash in two years.

I will enjoy actually having things work instead of troubleshooting them. Meanwhile, do keep waiting for that sweet sweet 2015 tier performance with double the power usage and heat generation for your 140th replay of Ashes of Singularity, pajeet.

File: Ball.gif (4MB, 344x203px) Image search: [Google]

4MB, 344x203px

Post yfw the RX is released and it performs exactly the same as the FE version or a little slower, being a cutdown, after being told MUH DRIVERS/MUH OPTIMIZATIONS for weeks on end

>>61242732
>Conveniently forgetting the crimson drivers that burned cards to a crisp by locking the fan speeds

>>61249590
t. Jen Hsun Huang

File: 1498792925698.jpg (54KB, 453x439px) Image search: [Google]

54KB, 453x439px

>>61249590
It's literally going to happen though which is the sad part
No amount of shitposting will save it
Ryzen shitposting worked because Ryzen is actually good

File: 1494781312617 (1).gif (2MB, 460x259px) Image search: [Google]

2MB, 460x259px

>>61244081
>use steam on amd for 11 yeras
>0 surveys

>>61249620
Will this be the new Bulldozer?

>>61249632
>blocks steam survey on the router
>0 surveys

Really makes you think

>>61249422
Typically people buy GPU's to play games they want to play now. Telling people "You'll get shitty performance today, but in 2 years it will totally be worth it! Even though in 2 years your GPU will be cheaper and a new generation will be out with better performance and new features!" is an awful marketing strategy. You want to know why the 2500K is the typical example of a "futureproof" CPU? because it was great on launch and had good performance even after later gens came out. You don't buy something shit on launch and hope it improves over time, that's what happened to people who got cucked by Bulldozer.

>>61249590
Why would I rejoice over AMD's failure? I refuse to buy another Nvidia card after I bought the 970 and I sure as hell am not going to buy anything AMD right now given that Polaris is not a huge upgrade and sold out everywhere because of miners. Might as well wait for Navi and two die cards.

>>61249445
You post this as if the Nvidia forums and subreddit isn't chock full of weird fucking driver and performance issues of their own.

Buying a high end graphics card, even in 2017, is like buying an exotic car. If you aren't prepared to do a little bit of work, it's going to annoy you at some point.

>>61249673
nVidia drivers are shit since ~mid 2014.

File: Screenshot_3.png (11KB, 509x300px) Image search: [Google]

11KB, 509x300px

>>61249639
Nah bulldozer had no redeeming qualities, even it's core count couldn't save it
Vega is at least slightly competent but the yearly delays and constant hypeposting was it's downfall, it made people expect a good graphics card

I have an entire folder of "Titan XP performance for 500 dollars"

>>61249673
You post this as if the AMD forums and subreddit isn't chock full of weird fucking driver and performance issues of their own.

>>61249689
Couldn't you just use the archives instead of wasting time like that?

File: i9.jpg (279KB, 800x1011px) Image search: [Google]

279KB, 800x1011px

>>61249702
It's my collection
Archives fuck off too much

I have an X299 folder on my laptop

>>61249700
Where did I imply AMD doesn't have problems too? I'm just saying that pretending like Nvidia graphics cards = years of trouble free ownership until they are naturally put out to pasture due to obsolescence is a fucking joke.

>>61249737
Sounds like fun.

>>61249689
>Nah bulldozer had no redeeming qualities, even it's core count couldn't save it

It was better at video encoding.

>>61249768
The 8350 actually took a second longer than the 3770k and had a higher handbrake fps of .05

It was incredibly worthless

>>61249819
I'm using these specific processors as an example because the 6100 etc were even worse

>>61249768
Maybe better than an i5, compared to an i7 2600K the 8150 was slightly worse, maybe matching it in a few programs. The 2600K was still a better value because of the horrid single core performance of FX though.

File: CEO_Jensen_Huang_GTC_keynote.jpg (214KB, 1000x667px) Image search: [Google]

214KB, 1000x667px

>>61249650
I'm not happy about the situation either. It's an abysmal state that lets Leatherman continually rub salt in the wound that miners have gouged out. However, I am sick of all of the excuses and rationalizations that deluded users have made for RTG ever since the first benches were put out by FE owners. Maybe when RX reviews are finally past NDA they will face the harsh reality and start discussing what went wrong with this launch and what Navi should do to fix problem (if it's even correctable in one cycle). I lay part of the blame on HBM which has to be one the biggest bills of goods that AMD was sold in a long time and they should have known better with their experiences with the Fury cards. Obviously there is a large transistor count that they committed to professional applications as SPECview alludes which means fuck all for game rendering.

>>61249872
>Obviously there is a large transistor count that they committed to professional applications as SPECview alludes which means fuck all for game rendering.
?
Anon are you retarded?
This smells /v/.

>>61249689
Why do you collect shitposts?

I'm gonna get a 1080, should I wait™ for rx vega?

>>61249902
Get a 1080. A certain man needs more leather jackets.

>>61249902
1080 prices seem to have come down on Newegg, they had spiked for a little while because of miners. Now's probably the best time to get them.

>>61249891
>shitposts
>this level of damage control
Because AMDumbfucks do it every GPU release
"X PERFORMANCE FOR Y DOLLARS"
>release day
"L-Literally nobody said it would be X performance we all knew it would be W"

>>61250713
Hello Jensen. How's GV100 yields?

>>61250734
Hello Raja. How's the toilet factory yields?

>>61250766
That's a weird way to deny another 1.7% yields on V100 hot lot, Jensen.

>>61250734
>J-Jen hsun

>>61250809
Yes, it was definitely his decision to make a fucking 815mm^2 GPU.
No, even POWER or Itanium chips were never THAT big.

>>61250833
Got no idea what you're talking about bud, think you should get yourself checked out

>>61250777
That's a weird way to deny 1.7% accuracy when using the toilet.

If it's as good as a 1080 at a decent price I'd think about it because I'll get freesync instead of goysync, and crossfire scales much better than SLI with no microstutter crap since fury era, so two of those might be good

>>61250847
Wait, the yields are even lower than 1.7%?

Its slightly above OC'd aftermarket 1070 on stock.
Its slightly below OC'd aftermarket 1080 on OC.

So Its basically a 1080 competitor.

>>61250849
You're not fooling anyone
You'll buy it no matter what

>>61250896
No one's going to buy RX Vega anyway. Just like no one bough previous Radeon flagships.

>titan x performance at the 1070 price the amdrones said

>>61252025
>ignoring everyone calling out their shitposting by just mentioning the big die size

>>61252025
Nobody said this, shill.

>>61231078
What if it costs $299 /g/? Cheaper than the 1080, and you can crossfire for $600, getting better performance than the 1080ti

That would finish Nvidia

>>61252274
Not with that die size and HBM.

>>61252274
>you can crossfire for $600
A gimmick that requires a lot of space and isn't even supported by a bunch of games? Same goes for SLI.

>>61242389
>Remember when they had cool names for their video cards?
The fuck you on about? AMD's Fiji was branded as Fury X, not edgy enough for you? Also Vega is cool as fuck as well, better than a bunch of numbers.

>>61249689
Holy shit that's pathetic. Collecting nearly a gigabyte of shitposts just so that you can compile them into "le epic xD AMDrone BTFO" images. Why can't I have as much free time as you.

>>61252274
It'll cost $500 minimum.

>>61252362
production cost is only about $50 higher than the RX580

>>61252175
You must me new

>>61252175
A lot people on /g/ said that...

>>61252478
Sauce?

>>61250713
so, people who said not to get gtx1080 on release and wait for gtx1080ti as it would be better price/perf were actually AMDumbfucks?

>>61253225
The 1080Ti really is better price/perf though? 1080 was a terrible buy until the price drop.

>>61253225
Waitfags are retarded no matter who they were. 1080Tifags said not to buy the 1080, and had to wait 6 fucking months for the 1080Ti only to get cucked by the Titan Xp like a month later. I've never seen a group of people so BTFO as 1080Ti waitfags, at least until vega waitfags, who are going to get screwed over even harder..

Oh wow fuck that. If they can't beat a 1080 then what's the point of vega? Why did we wait 2+ years for vega if it can't even compete with the highend cards?

>>61253309
No one cares about the Titan Xp. You're paying twice the price for 10% at most over the 1080 Ti. The people that got cucked are the Titan X Pascal since they thought they were getting the final GP102 chip.

>>61253309
The Titan Xp was better than Titan X (pascal,2 August) thought

>>61253335
APU's, AMD is skipping Polaris and putting Vega graphics in Raven Ridge. It's kind of hilarious, after Piledriver, a lot of reviewers though AMD would drop out of the high end CPU market and focus on mobile CPU's and APU's instead. Now people are thinking the opposite, AMD is focusing on high end desktop and server CPU's while the graphics division is focused on cheap and power efficient mobile GPU's while giving up on competing with Nvidia in the high end.

>>61253380
That's my point, the 1080Ti was the "best" GPU for only a few weeks, then Nvidia one-upped themselves. People waited 6 months to buy a 2nd place GPU.

>>61249305
Happened with Ryzen.

>>61253393
The people that care about 10% to have the "best" would have bought a Titan X before the Xp anyway. You wait for the 1080 Ti because it's a huge jump over the 1080 and was a predictable release.

>>61248240
>>61247314
>>61246992
the improvements seem to not be working in the driver yet, we know they have a new render pipeline but that shit is falling back to the old method rather then the new one.

current vega was for compute/pro apps on a driver that is in fallback mode when it comes to gaming.

>>61253428
>and was a predictable release.
You're still missing the point. Everyone on /g/ kept saying to avoid the 1080 since the 1080Ti would be out in one to two months. Everyone KNEW the 1080Ti would render the 1080 obsolete soon. I kept telling everyone that Nvidia had no reason to release the 1080Ti so soon since AMD had no high end GPU's, but nobody here listened to me. They kept saying to wait. Guess how long they waited? I was actually wrong, it wasn't 6 months, it was 10 months. The 1080 was released in May, while the 1080Ti was released in March. People were saying to wait for the 1080Ti for almost a year, while 1080fags were enjoying their high end GPU that whole time. My point is waitfags on /g/ are fucking retarded. Unless a release is officially announced and you know when it's coming, waiting is pointless.

>>61253550
That's too much text anon. The 1080 was a stupidly retarded buy back then and unless you had something less than a 970, then there was no incentive to upgrade. Picking up a 480 or 1070 was a significantly better choice if you wanted midrange or entry level high range. The 1080Ti got beaten, so what? Buying any Titan is retarded. We won't see a 1080TiTi since AMD really blundered with the Vega release.

>>61253550
>Everyone on /g/ kept saying to avoid the 1080 since the 1080Ti would be out in one to two months.

When? Because if they said that early in the year they'd be right.

1080 Ti release had a predictable time frame too. It was 8 months from 980 to 980 Ti release which isn't that far off from the 9 months from 1080 to 1080 Ti release.

>1080fags were enjoying their high end GPU that whole time.

You could have just bought a 1070 in the meantime and not get completely cucked with the price drop and hell you'd even be way better off since the 1070 can be used for mining and is way overpriced now. Sorry but the 1080 was a stupid buy.

>>61253621
>>61253632
If anything the 1070 was the pointless GPU in the Pascal lineup. It's overkill for 1080p, not quite good enough to max out 1440p/60 in newer games and cost more than the GTX 970 did. The only good thing about the 1070 was 8GB of VRAM. The 1080 allows you to comfortably hit 1440/60 and even 1440p/120.

Vega's new work distributor is fucked up. Neither geometry nor pixel throughput (or geometry discard) have been a problem in ages except for games that intentionally abuse tessellation to kill performance, which represents a handful of games mostly from 5 or so years ago, nothing being benchmarked now.

>>61253675
That's bullshit and I have a 1080. The 1070 is fine for 1440/60 and the performance jump from 1070 to 1080 isn't going to give you anything better than that. The 1080 was the worthless one since it couldn't get you 4k and it for sure can't get you 1440/120.

Why doesn't AMD just make shitload of GPUs then if miners buy them all up?
It's not like nobody will buy them, I only see a win x win situation for them by doing that.

I'm even pretty fucking sure that they do inhouse testings regarding mining at AMD where they know perfectly well how good their cards are at it.

Might as well just make a dedicated mining card, since it's market that's expanding that also seems to stay.

>>61243089
funny thing about hairworks, that was obviously pushed in by nvidia
it looks worse than what cdpr made

>>61254098

Nobody knows when the mining crash will occur and thus nobody (be it AMD, Nvidia or AIBs) want to saturate their supply chain with potentially millions of shekels worth of stock that won't sell.

>>61254200
IF a mining crash will ever occur, since alt-coins are popping left and right.

>>61234447
No one expected it to beat an overclocked 1080Ti, especially not in 1080p.
In 4k compared to a stock 1080Ti, maybe.

>>61236403
Hell I'd be happy with $350-$425 for a roughly nano sized card, but dual fans this time, to not have to deal with retarded Nvidia drivers and software and to have something that works correctly with more than 4 cores, even if it's just 1070 performance.

Why the fuck would that not be okay? 1070s are like $380-$500. The only thing it really has going for it is power efficiency. If it's beaten on size, drivers/software, etc, then a similar priced card that's better in those ways is worth it.

>>61249604
Except they throttled and didn't actually burn because AMD PCBs are overbuilt as fuck on reference cards, and even on AIBs that are worse than reference they're still not as bad as Nvidia reference cards.

Nvidia cards, on the other hand, literally fried and were destroyed by a similar bug that did the same thing. lmao.

>>61231078
posts clocks while there are no clocks to be found

>>61235174
>AMD may fix it, but they still fucked up.
the card works wonders on prof benchmarks
literally the market it launched in
random /g/ shill
>amd fucked up

If it's more powerful than the 1080 I think it will be good.

File: _MG_8191.jpg (105KB, 641x907px) Image search: [Google]

105KB, 641x907px

>>61254498
>https://www.youtube.com/watch?v=yudueaG5_rE
>Gayming mode is a placebo switch
>No performance difference between Pro Mode and Gayming Mode
>AMDrone fanboys everywhere continue the chant "b-b-but it's not designed for games!!"
AMD have taken a page out of the Nvidia/Intel guide, and shafted their red army followers!

File: mel-gibson1.jpg (107KB, 500x740px) Image search: [Google]

107KB, 500x740px

>>61252274
>and you can crossfire for $600
more like 'you can crossfire for 600 watts BAHAHAHAHAHAHAHAHAHAHAHAHAHAHSHAHAHAHAHAHAHAHAHAHAHAHAHA

File: newbarrier-mesh.jpg (359KB, 620x800px) Image search: [Google]

359KB, 620x800px

>>61253784
>except for games that intentionally abuse tessellation to kill performance, which represents a handful of games mostly from 5 or so years ago, nothing being benchmarked now.
Never forget.
https://techreport.com/review/21404/crysis-2-tessellation-too-much-of-a-good-thing/2

>>61245454

He is actually correct. Cost is a function of supply and demand. If you sell below costs, you simply leave the market after a while or you have to finance that stuff.

Let's say the rx vega costs amd 400$ per card and they want to sell it at 650$.

Of the market buys them, great. If not, prices go down and if the card is shit, they may go down to 300$ and it would be a smarter move to sell at 100$ loss per card instead of not selling any card at all.

It's basic economics

>>61254850
suace

>>61258074
Yes to cover overheads, but the price of the silicon/memory will decrease along the life cycle.

Did AMD ever hey crossfire to work in window mode? If Vega is cheap enough I wouldn't mind gracing two of them.

>>61245435
What did he mean by this