>780ti worse than 7850
Nvidiots will defend this
>>55560712
Thats a benchmark using SMAA (no async compute). A benchmark using TSSAA would show MORE improvement on AMD side.
Just making sure people know it so they don't think this is the true potential of Vulkan.
>>55560712
rx480 is still worse than the 980 though.
Higher max framerate with same minimum? Disgusting.
>>55560894
>rx480 is still worse than the 980 though.
>worse
Only a die hard shill could be this retarded
>>55560712
>nvidia still at the top of the chart
POO IN THE GPU
>>55560921
Enjoy your stuttering when your framerate drops from 110 to 90 all of a sudden :')
>>55560712
>TFW you bought a 780ti
REEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE
THIS IS THE LAST TIME NVIDIA
>>55560950
19% difference between Fury X and GTX 1080.
With no async compute.
Imagine the difference WITH async compute, if they did the test with TSSAA.
Protip, the situation would be reverse.
>>55560996
Wait for drivers you fucking retard.
>AMD developed API works better on AMD graphics cards
>>55561067
There aren't going to be drivers. Kepler is legacy.
>>55561067
>GTX 7XX
>Drivers
LOL NO
>>55560921
>rx480 is still worse than the 980 though.
105>110???
>>55560894
>>55561092
Stupid weaboo
>>55561067
I thought that line was only for AMD fanboys?
>>55560712
Kepler was abandoned by Nvidia for Maxwell, and soon you will see Maxwell abandoned for Pascal
>>55560712
does this mean I can wait a little longer?
7870 btw
>>55560712
>>55560996
Can confirm massive issues with Vulkan and my 780 Ti, I tried going to Argent Tower near the beginning I was seeing this for framerates:
OpenGL - 66-86 FPS
Vulkan - 50-64 FPS
And this is with the latest 368.69 driver.
>>55561601
Some Nvidia users are reporting stuttering or even straight up CTD under vulkan.
>>55561224
Stupid Nvidiot. There haven't been any improvements to the 700 series since The Witcher 3 came out, when their hand was forced due to it running like absolute shit and a shitstorm about gimping starting. They've done fuck all since then and never will again.
t. GTX 780 owner
>>55560712
>excellent open source AMD drivers on Linux with Vulkan support
Feels good man. Nvidia either has drivers where your kernel breaks everytime you update or boot animations don't work or your card burns.
>>55561621
Didn't play long enough with Vulkan to attest to that but Doom has been pretty damn crashy and buggy for me. 30 hours of play got my 5 CTDs, I had better luck with FO4. And sometimes when the game loads (new level or even just going back to a checkpoint) it would start playing at 6-10 FPS for no reason until I quit and restarted.
I have had quite a bit of fun playing overall but stability has been shit.
>>55561656
>he still thinks Linux will ever be relevant in gaming
lol
Devs don't even make OSX games anymore, and OSX is much bigger than Linux.
>>55561645
Nice try AMD shill, but your post just screams pajeet.
>>55561656
I find it amusing AMD fags think this is true :^)
>>55561699
I don't really care what's relevant or not relevant. Linux does my work and does it well. ZFS/Btrfs, LXC, KVM VMs, package manager, VNC, excellent network management with bonds, VLANs and OpenVPN/PPTP inbuilt is critical for me. If it means I can play games as well it's a huge bonus. I don't really care about a few FPS difference. If it works well, may as well spend more for a better card. Also you seem to discount the potential Steam OS and Steam inhouse streaming has.
>>55561788
But it is. I have actually tried running a GTX 650 on Linux. I gotta admit Nvidia worked well in my HTPC running OpenELEC.
https://www.phoronix.com/scan.php?page=news_item&px=Linux-4.8-DRM-Next-AMDGPU-1
>>55561656
I like where they're going with AMDGPU, but they're still behind in performance. I can confirm on kernel upgrades with proprietary drivers. The update from 4.5 to 4.6 forced me to reinstall drivers. I hope Vega gets the Polaris treatment in Linux drivers so I can switch whenever 4.8 with freesync comes out.
>>55562004
>but they're still behind in performance
True. Hopefully should improve. Power management, HDMI audio, audio passthrough, video encoding with GPU, video decoding with GPU, updating any component on the system all work without hitches for me. Right now making sure things work and latest OpenGL/Vulkan standards compliance is much more important then performance for AMD on Linux.
a-sync drivers will fix it
>>55561767
Aww. How cute. Yet you can't prove that Kepler didn't get gimped so you start spewing insults.
>>55561080
>N/v/idia ignores low level APIs despite their promise on consoles
>AMD tries to push it and ultimately ends up making their own API and offering N/v/idia a chance to take part in it's development (which they naturally decline)
>Somehow surprising that the company that spent years and shitloads of money on low end API development gets their foot in between the door really well with low level APIs
>>55561621
Does this mean vulkan wont see mass adoption?
>>55561080
>Vulkan
>AMD developed
It's funny because even though Mantle is the foundation for Vulkan, the Khronos CEO couldn't distance himself farther from AMD if he tried in all of the shit he's ever said to the tech press. He'd always downplay the connection so they wouldn't alienate Nvidia. It's actually quite hilarious how terrible Nvidia's arch is for the future. It's like Nvidia is actually surprised Vulkan became a working thing.
My 970 still runs games on ultra/60fps/2560x1080 but no performance gain from DX12 and Vulcan means it's on its last dying breath before I'll have to start sacrificing details while something like 480 should be fine for years. What's weird to me is that 480rx doesn't get such a massive boost when compared to older amd cards which really makes me think it's literally another, 3rd now, refurbish of 7xxx cards.
>>55562116
Do you use AMDGPU Pro? This one has extra goodies like better openGL and Vulkan.
>>55562162
780ti user.
I can confirm gimping.
>>55562298
We believe you :^)
>>55562313
I do, anon.
>>55562313
>>55562279
>which really makes me think it's literally another, 3rd now, refurbish of 7xxx cards
That makes zero sense. You said yourself it isn't boosted as much (relatively speaking) by Vulkan compared to older GCN GPUs. It behave differently than older GCN because older GCN was that much worse for DX11 while still being damn powerful when utilized properly.
>>55562257
I imagine it will be available as an alternative, but there's nothing Nvidia can do.
>>55562279
Nigga Polaris is new. Tahiti has been GOAT, I wish I never bought that 670 and waited for the 7950 to come back to stock.
>>55562289
I don't have a Polaris card, just a GCN 1.0 card, which should be getting AMDGPU PRO support SOON(tm).
>>55561067
Remember when Nvidia said to wait for asnyc compute drivers for Maxwell beginning of this year? Nothing yet. Nvidiots will keep waiting
>>55562343
100% believe you
>>55562348
How it doesn't make sense? They've upgraded and improved each alteration while dropping price. Differences between 290/390/480 are minimal in Vulcan because hardware wise those cards are very similar.
>>55562373
>>55562408
Nice photoshop kid, you're a real pro
>>55562373
sadly i don't think reversing a speccy will get you accurate results
>>55562410
>>55560712
>he bought a 780 TI
Jesus that thing was so expensive, what a waste.
>>55560712
>970 still below 290
Never buying Nvidia again.
Also where's the Vulkan graph showing the Fury X at 160+?
>>55562424
>those fonts
turn on subpixel hinting in "adjust the appearance and performance of Windows", smooth edges of screen fonts
>>55562393
>hardware wise those cards are very similar
The 290 and 390 are literally the same GPU, so yeah they are similar as you are well aware. The 480 just looks like a GPU that was made by the same company using some of the same principals. The 480's Polaris uses almost half the power of Hawaii to get similar results.
>>55560996
>tfw original titans in sli
it's like being born with a huge cock only to become impotent by your 13th birthday.
The rumor about 1060 3GB will sold for $149 started to sound believable now.
Who knows, maybe AMD will drop the price of 470/460 cards.
>>55562199
>offering N/v/idia a chance to take part in it's development
Source?
The only thing I find on this matter is Nvidia not wanting to support it, nothing about any sort of offer to be part of development.
>>55562550
>Original Titans
I'll trade you a 7970 for them, m8. :)
>>55562370
what drivers is he talking about? i thought nvidia didn't support async or does that preemption thing count as some sort of async?
>>55562614
http://www.tweaktown.com/news/52996/nvidia-culls-geforce-gtx-1060-3gb-arrive-1050-december/index.html
if they're turning the 1060 straight on into a 1050, just with less vram, that means that AMD's line got them by surprise.
I don't know the 1060 gonna perform, but I'm already anticipating the greatest shillfest of our time.
>>55562616
http://wccftech.com/amd-mantle-api-require-gcn-work-nvidia-graphic-cards/
http://www.dsogaming.com/news/nvidia-finally-officially-speaks-about-amds-mantle-will-not-support-it-no-real-benefit-using-it/
>>55562675
going to avoid /g/ on 1060 launch week
so with vulkan (let alone with async), a 480 really is 980 performance
so a 470 will be 970?
>tfw nvidia 970 3.5gb memers get beat by a $150 single fan qt card
>tfw a 960 gets beat by a $100 passively cooled PCI powered card
I remember /g/ said that Directx 12 will kill Mantle back then.
Turn out we got played like a fiddle by AMD.
>>55562772
Mam, the RX460 will be cute as fuck. I don't think it'll be passively cooled, but it won't have PCIe
>>55562820
>make Mantle
>new APIs take from it
>everyone's faces when AMD doesn't even have to write drivers now
>>55562132
Soon™ ;^)
>>55562441
This is why I'm scared to buy a 1070. I want to be fucking set for 1080p gaming for a good while dammit why can't they just not gimp their goddamn cards. Is there any chance that they won't be gimping Pascal?
>>55563282
They don't gimp though, that's just a silly /g/ meme.
>>55562850
>Vulkan and DX12 start to take off in 2016
>AMD only has a new midrange card out with their only flagships being left over 28nm housefires limited to 4GB HBM1
Man, Vega can't come soon enough. Talk about JUST.
>>55563282
>>55563292
They do. r9 290x was made to compete with 780ti, now the r9 290x is competing with 980.
Kepler cards were gimped by tessellation in The Witcher 3, Maxwell ran the game fine. Meanwhile AMD r9 200 and 7970 faggots only had to turn down tessellation to match Maxwell perf.
>>55563317
do I even have to say it
>>55563336
AMD drivers are shit, that's why they "age better", with nvidia you're getting pretty close to max performance from your hardware on day 1. That's why they don't age as well.
>>55563362
This is true. It depends on what you want when making a purchase. If you value upside, you'll go with AMD. If you just want the best right now, then you'll go Nvidia.
>>55563362
Finally, someone who's not a retard made a post ITT
>>55563362
max performance in day 1 doesn't warrant making the card outdated when the next arch rolls in
>>55563282
I wouldn't touch the 1070. The Pascal architecture doesn't seem to be at all worthwhile for Vulkan or DX12, so I have a feeling these nice DX11 figures from the 1070 are going to be short lived. It might be good for a couple years at best, but I don't think the 1070 will last at all. Also a 256-bit bus is just pathetic for the price imo.
>>55563441
They don't though
>>55563454
Yeah it's all about bus size and not memory bandwidth at all.
4096 bit fury x really destroyed the 384 bit 980 ti so so so hard!
So what the fuck do I buy, a 1070 or a 480?
>>55563454
rx 480 h-here I come
Honestly I can't fucking wait to upgrade my shitty 7870, anyone got dates on the aftermarket editions?
>>55563456
They do, RIP Titan cards when compared to 980ti
>>55563454
Pascal feels like a stop gap while they work on Volta.
Why would anyone wasted their money 1070 or yet alone a 1080 for god sake? Its still using GDDR.
Unless you are one of that retarded people who thinks "I need to max my FPS on my 4k MASTURD RACE PC GAMING monitor" then go for it.
I even doubt it will hold for less than a year.
>>55563498
Please do some research before buying and don't blindly trust anonymous people on 4chan
>>55563488
you want to play games now, feel like a cuck later? 1070
only ref 480s are out, so if you have to buy immediately, buy a 1070, aftermarket 480 is probably going to be end of month
>>55563512
It's like you haven't read anything in this thread before choosing a random post and replying
>>55563516
Do you not understand the nervous >h-here
stutter there? Too much of an autist?
>>55563515
Waiting for HBM2 is suffering.
>>55560921
>>55563557
>780ti above a 970
what the fuck is this supposed to be
>>55563576
Are you autistic?
>>55563557
only retards use passmark
>>55560712
>280X better than 380X
780Tis are reall shitty nowadays, such power left unused.
>MFW I actually use a 780Ti right now
>Cannot really play some newer games like Witcher because they're gimped
I wonder how the retards that bought titans are feeling right now
>>55563631
are you retarded?
>>55563482
It's beginning to age like fine wine, especially with Vulkan. I don't really see your point lad.
>>55563695
Does anyone in /g/ actually owned a Titan?
I don't remember anyone ever posted one.
>>55563671
Maybe if you tried updating drivers, or not lying on the internet?
I have a gtx680 and it still runs everything fine, just as I'd expect from a 4 year old GPU.
>>55563695
Friend bought one that he later upgraded from. If you have money and expensive computer parts are all you really save for then buying shit like that is easy. He gave it to be when he was done with it. Still a beast, although it was still a shit deal at the time. Man can't complain about free though.
>>55563848
How was your Vulkan test?
It really doesn't matter when there is another API already.
>>55563844
guts threads
>>55563557
Bruh passmark is fucking worthless, they've had the same 3 shitty dx9 tests since like 2003.
You should only use it to compare your own system scores to your own new hardware, the CPU and disk transfer tests are pretty much the only things that give legitimate results.
>>55563482
more like it was 320GB/s vs 512GB/s
the bus size is only factor, HBM doesn't need as high of clockspeeds. That GDDR5 is running at ~3500MHz while the HBM is at 1000MHz.
>>55560712
The fuck? My 780 pulls solid 60
>>55564220
In Vulkan on Ultra or with OpenGL?
I lost 25% of my performance on my 780 Ti when I switched over to Vulkan.
>>55563707
the 780ti is better than a 970....just because nivida cucks you guys with gimping so hard doens´t make it less true
>>55560894
$200 card competes with a $560 card
>>worse
hey look.
another GOu thread where every benchmark is different
sweeeet
why does't 4chan make a videcard board?
>>55563695
I'm confused by what you're implying.
Do you think anyone believes Titans are a magic card that will never be replaced by future gens?
The only people who actually have Titans are people who got Titan Z's as cheap Quadro alternatives.
>>55563557
Passmark is only good for CPU benchmarks. Everyone uses http://gpu.userbenchmark.com/ if they're serious.
>>55560748
>>55561006
I don't understand. Why does this matter? Would an anon kindly explain, please.
>>55562279
the 280x is the last time a 7970 was rebranded, the reason the 480 gets less is it had better initial performance and also its a mid range gpu, it is new, but its not made to be powerful but to be damn good at its price, and looking at the bench above, mission accomplished.
I would hold out for the 490 though, as the 480 could potentially drop in price from 240 to 200 for the 8gb and leave aib with a 300$ for cooler and binning to play with.
>>55562772
on dx12 and vulcan, probably otherwise the 480 on 11 is just above a 970 and a 470 by specs should be just below if not equal to a 970
Does any of these DX12/Vulkan games got Nvidia Physx in it?
I'm just curious how big the performance dropped in both cards between the DX11 version and DX12/Vulkan version.
>>55564304
Since when did the price increase to retail price in 2014? Fucking knobhead.
>>55563482
Bigger bus size is equal to higher bandwidth fucktard
Fiji is ROP and vRAM limited
>>55564918
Higher memory frequency also leads to higher bandwidth.
Also, how about delta compression?
256 bit on a 1070 is enough to put it ahead of the Titan x.
>>55560712
This bench seems pretty fucked up.
The 960 has pretty much the same performance as the 760, depending on the game the 760 can even have a slight advantage thanks to the bigger bus.
But here we see a sinlge 960 not only beating the 780, 780Ti and Titan, but also a 780Ti setup.
COnsidering that the 780Ti is actually quite similar in performance to the 970 this makes it look even more strange.
I call fake on this one and that before I even saw the 750Ti beating the 760.
>>55561264
When that happens I'll switch to vega if it turns out to be any good AMD cards seem to retain more performance over the long run.
Upgrading to a new flagship every 2 years is a massive burden, might as well just stick to midrange.
>>55564932
Vulkan is just that fucked on Kepler.
Here is the test with OpenGL.
>>55564967
Forgot the link: http://gamegpu.com/action-/-fps-/-tps/doom-api-vulkan-test-gpu
>>55564967
Nice to see my old 680 can still kick some ass.
I still think we should wait for driver updates before bitching about gimping though.
>>55564932
>I call fake on this one and that before I even saw the 750Ti beating the 760.
750ti is maxwell, you know
760 is kepler
look it up
>>55564990
Can we expect driver update for the 7xx lineup? I can't imagine they're particularly vulkan/dx12 friendly cards, so I'm not sure drivers would fix it, or even if they're coming (did nvidia say so?).
>>55565029
Drivers should help somewhat, but don't expect anything spectacular.
My guess is that keppler is just missing hardware to fully utilize the advantages of a modern low level API, and using software to fix it is resulting in large performance drops
Here's another one for you
>>55565137
Yeah, gonna need some sauce on that.
>>55565029
They aren't
http://www.nvidia.com/page/legacy.html
>>55565152
Source
https://nl.hardware.info/reviews/6843/10/nvidia-geforce-gtx-1080-sli-review-ultiem-gamen-benchmarksndoom-2016
>>55563515
>Memory bandwidth is a bottleneck
When will this meme die? I'm thinking of saudering ddr3 on my 1070 just so people will shut the fuck up
>>55565137
DELETE THIS RIGHT NOW
>>55561325
R9 270x here
>tfw were beating a 780ti right now
Still unplayable frames but hey, it at least means these low level api's have tons of potential. Hopefully more devs take advantage of dx12/vulkan sooner than later. Cant wait to see how bad ubisoft manages to fuck it up regardless.
>>55565137
Wow looks like AMD is a much better deal I'm gonna buy some AMD tommorrow and some stock too it's going up up up
>>55565269
and im sure you'll have great success doing so seeing how you can't even spell soldering. fucking retard
>>55565137
Rather interesting results, especially considering the date. That was released on the 9th before the Vulkan patch was out.
Test rig looks similar to gamegpu but has a lower clock on the processor and half as much RAM running at a slower speed but that shouldn't be enough to account for what we are seeing.
I would love to know what is going on here.
>>55565169
Lol
>>55565333
You are several months late.
>tfw poor student
>tfw not investing all of your savings on AMD stock at 1.6 per share because you would not be able to live
>>55562270
Mantle was completely open source. ANYONE could use it and build upon it if they wanted to, just like freesync is right now. Nvidia made the choice to ignore it. Microsoft made the choice to basically build dx12 out of it.
>>55565365
>supported implies changes were made to improve performance on every single card
>>55565374
Mantle is basically a massive FUCK YOU to Nvidia's architecture, so it's no surprise it's something that Nvidia wanted to ignore. Low level API's are Nvidia's kryptonite.
>>55565383
But they were, and they will continue to be
>>55565497
Do you not understand the meaning of LEGACY. It means hotfixes and bugfixes only. No optimization will be done for Kepler or older GPUs. If you don't like it, buy AMD.
>>55565527
Right, because AMD still provides driver updates for the HD5000 and HD6000 series cards?
What's that? No driver updates at all?
Yeah I'll stick with Nvidia who still provide game ready drivers for my gtx 670
>>55564561
That benchmark is a vulkan benchmark without async compute. The devs said async compute is enabled with TSSSA and without any AA.
If the benchmark used TSSAA the relative performance from AMD GPU is would be lot faster.
>>55560712
I will always admire the 780ti and the butt-rekking it has given to AMD for many years. Now it has been thrown under the bus and will not get the respect it deserves.
Sincerely,
AMDfag
>>55562733
Fucken nvidia fuck sake why do I always end up a cuck
>>55565564
>Game-ready
>LEGACY
That's an oxymoron
also, 7970 is still supported. If you're using anything older than that, you won't be getting decent gameplay anyway.
No Vulkan support for Fermi
>B-But its OK if Nvidia does it.
>>55565578
Embarrassing.
>>55565597
Funny, I gave my HD5870 to my younger brother and it still manages gta v 1080p on medium settings. Gtx 570 is still getting game ready drivers, hd 6870 isn't getting any new drivers
>280x better than 7970
>Literally absolutely the same card.
Did AMD adopt Nvidia's driver cucking techniques?
>>55565568
Thanks for the response, but shouldn't that be universal? Why are certain AA modes not async, while others are?
>>55565661
280x is GHz edition.
>>55565676
Dev said they will work on other AA and make them use async later on.
They're basically releasing them as they are done coding/testing/optimizing. They could make all the AA do async and release them together in 1 or two years later, but I think you'd see the benefits of releasing things as fast as possible.
>>55565768
Oh, oh, oooooh, so even something like MSAA needs to be explicitly coded for in vulkan/dx12. Somehow I expected that to be a innately available, since it's been such a long lived standard. I didn't even think the API provided it. I think I've underestimated just how low level these APIs are. Thanks.
Devs worth a shit will pick Vulkan up.
I'm worried over this.
No I missed >>55555555
>>55565805
Basically it's paralleling the GPU to do shaders/AA/compute/render and managing them properly. So each take have to managed properly.
If they can do that, then the async computer works as intended.
You think it's worthwhile to get an RX 480 for testing purposes?
I'm pretty much married to nVidia from the amount of times I've been burned by AMD.....
>Bulldozer
>Steamroller
>Mouse corruption
>Shitty CPU temp readings
But I like to color myself as a skeptic, and considering how cheap they are, I figured I might throw one into an APU build for shits and giggles.
Thoughts?
>>55565944
>throw one into an APU build
Don't do this. You'll get horrible performance from driver overhead
>>55565965
Define 'driver overhead'.
I've never heard this term before.
>>55565979
Gameworks
>>55565944
I owned 65 units of AMD GPU since seven years ago and still going.
I don't know anything about anything about the CPU but I can tell you the mouse corruption has been stop occurring for like 4 or 5 years now, I barelly remember it anymore until you just mention it now.
Oh and, APU is shit. better bought a cheap intel CPU and then pair it up with a better AMD GPU if you want a better cheap budget gaming PC.
>>55565979
The truth is that AMD GPUs put a bit more burden on the CPU relative to Nvidia GPUs due to the way AMD's drivers interact with the OS. Windows 10 improved this a great deal for AMD but Nvidia still has an edge when it comes to squeezing performance out of poorfag CPUs. Still, there aren't a ton of games that will fuck you up with a CPU bottleneck before the GPU will. They're out there and some of them are AAA titles like GTA V but I'd guess that in 75% of situations, poorfag builds with AMD CPUs are still going to be GPU bottlenecked, especially at 1080p and higher.
>>55565661
reference 280x is higher clocked than reference 7970
>>55566032
>>55565965
I guess I can always through it in my other ITX build. I was going to match it with an AMD CPU for OCD sake, but if you guys insist on this overhead.
>>55565979
It's a real issue unfortunately, AMD drivers only use a single core, which is why they're pushing dx12 and vulkan so hard since it overcomes this bottleneck
>>55566202
>AMD drivers only use a single core
Well, not exactly. DX11 doesn't know how to effectively use anything beyond the first core and AMD's driver is just weightier than Nvidia's, so the GPU ends up waiting for the CPU at an earlier point relative to the comparable Nvidia card when DX11 is involved.
DX12/Vulkan allow the developers to freely use the CPU cores as they see fit, so modern CPUs will have a way higher ceiling before being a bottleneck issue. Even if AMD's driver is still less efficient, you're not going to see that kind of bottleneck issue in DX12.
what should I buy I have a GTX 760 my budget is 1500PLN
>>55566522
RX 480
>>55566603
will non reference rx 480's have fixed problems occuring with reference cards?
>>55566816
If you live in the states the 480 is the best choice if you're on a budget and on a 1080P monitor. Get the 4Gb version
>>55566816
Drivers fixed the rx480 issues
>>55565657
Because of GCN. Older tech they had in cards before GCN came in to the market got shoved out pretty much
>>55566905
Fermi is a different arch than keppler and maxwell.
>>55566922
because AMD i guess.
>>55566864
There are no 4gb RX 480s. They're all bios locked 8gb. That's why they keep selling out. Read more about it here.
http://www.tweaktown.com/news/52891/amd-radeon-rx-480-4gb-models-8gb-ram-unlocked-bios-update/index.html
>mfw dx12
>295x2 beaten by 290x
wat
>>55566938
That's fucking terrible. What exactly would cause such a stutter?
>>55566938
Are those TW:Warhammer benchmarks a joke or do I miss something? I play in 2560x1080 with 2x MSAA and everything else set to Ultra and I get a stable 60 fps. Is 4x MSAA really taking away 20 frames?
>>55566936
The first 4gb models off the line were actually 8gbbut theyre making 4gb ones now
>>55566949
Each GPU in a 295x2 ran slower than a 290X so if you have a case where Crossfire is broken this is what you see.
Notice how all the SLI setups are equal to one card as well.
I'll buy AMD when they fix their energy usage, their heat output, and their drivers. Until then I'll be using the objectively better brand
stay mad, AMDumbass
>>55567058
I didn't realize that the 295x2's cards run slower individually compared to a 290x. Why would someone bother with getting a 295x2 anyway? The price was $1500, so it surely couldn't have been cheaper. Was it just for the aio water cooler?
>>55564647
Fucking kek retard, still a massive drop
$250 for RX 480 8GB
$400 for GTX 480
>>55567095
Mini itx
>>55566963
Quality coding.
Look at Hitman. It has texture streaming issues with DX12, and ends up looking like shit.
>>55567102
GTX 980, sorry, I'm pretty drunk
>>55563300
The rx480 is the flagship. Flagship does not have to be the most powerful, they wanted the card they put their name behind to be the mid range bang for the buck card.
>>55566864
I live in poland my post >>55566522
>>55567095
That monster had a 500 watt TDP as it was, full 290X performance would have put it closer to 600W.
As >>55567103 said it was basically for cases where you wanted to do Crossfire but couldn't fit two cards in your case, although you need need to be able to fit a long ass card in your case. Two cards was actually a good bit cheaper than the 295x2 if I remember right so you paid for the cooling and the form factor.
Say that I want to make an online game with no textures, only shaders, and I find OpenGL extremely vague and strange to work with.
Should I still stick with OpenGL, or go with Vulkan?
Also, what do you compile Vulkan with anyway?
>>55567160
Drinking your AMD buyer's remorse away, eh?
>>55562733
since we are finding retarded nvidia quotes, can anyone find how their rep said how freesync wouldn't ever work outside of laptop(when it was first demo'ed) and YOU MUST HAVE OUR $200 PLACEBO chip to run synchronization of any kind.
>>55567183
Why does it have to be like this? I don't want it to be like that.
Is it because of the users anal porn?
>>55567179
Why the hell you stick with openGL when the company who develop openGL already moved to Vulkan?
>>55567172
buy from Germany you dumb mother fucker.
>>55567202
Apparently Vulkan is harder.
>>55567179
Vulkan is mostly C++ you can compiler it with any cpp compiler that is up to date.
>>55567123
I remember when I had 770 always complained about streaming, now with 390 it seems to be gone for no reason in games that had this issue.
>>55567207
why?
>>55567210
this argument again, if company wants to stay relevant and expands market for itself it will learn no matter what
>>55567242
faster memory and better compression, would be my guess.
>>55567252
tax is lower? compare taxes
>>55567252
<Cheaper price
>Free delivery
No wonder Poland is in the shitter, they can't into economy even on basic level.
>>55567261
you don't say
>>55567283
Fuck off mate, you invited mister obvious and he came.
>>55567274
>>55567272
sorry never bought from germany
>>55563557
I just got back from work. Let me load up passmark and relax. Wow ! The numbers are so high.. I'm definitely relaxed now. Screw game benchmarks right..
lol the price in euros of rx 480 is like 30PLN less after conversion
Where is that little Nvidia shill that posts stuff like 'the 1060 is better than the 480 in every way', 'the 1060 destroys a 480'.'the 1060 easily beats a 480', 'the 480 is dangerous and damages your motherboard', ''there are hundreds of pages proving the 480 damages motherboards. Just Google it' etc etc ad nauseam.
He makes me giggle.
>>55560712
>gtx 760 worse than gtx 750ti
>>55567401
not on the clock yet, he works on EU hours
>>55567416
It's fake
>>55567416
>Kepler worse than Maxwell
Who would have thought?
>>55566938
These issues are fixed in the latest drivers. Guru3D did some testing anyhow (using Fraps because Fcap would not work). But still said it's not an issue.
http://www.guru3d.com/articles-pages/amd-radeon-r9-rx-480-8gb-review,27.html
>>55567517
Techreport used the same drivers guru3d used in your link, and still ran into the stutter (it's in the picture).
>>55566938
Damn those stutters are insane! Glad nvidia doesnt have them! Its insane!
>>55566938
Defend this.
What should I upgrade first from a budget 600 dollar build 2 years back?
My fx 8320 or my gtx 960?
>>55567551
so by this we can conclude that there is no reliable info.
>>55567612
nothing.
wait for GTX1060 and aftermarket rx480 reviews you can buy one of them and good i5 for 500bux and that would leave you with 100bux for mobo and memory.
>>55567615
The proof is in the gameplay. If you notice it then there's a problem. If not then fuck the reports for spreading fud.
>>55567612
cpu fof sure
>>55567573
The 390 stutters too in that you retard.
>>55567582
Instable cpu oc, you can tell because it only happens at 1080p
>>55561067
wait a second, isn't this something that AMD users get memed on for?
hahahahah eat shit nvidiot
thinking of upgrading from the 7950 core
would the 480 be a good upgrade?
seeing a lot of good things about it
>780 SLI at the same performance as a single 780
Wew Lad!
>>55560712
why this bench has 20 less fps on fury x? isnt running async ?
>>55560894
980 105
480 110
>worse
>>55567700
>>55562663
it was a shit tier tweet that started a war between them and oxide....
just like when 1080 launched and they discovered that the card wasnt rendering some shaders on aots resulting in higher perfomance
>>55565493
Shouldn't GPU manufacturers gravitate towards low-level APIs by default? I mean the closer you get to the metal the better performance you can get out of it. Why DOESN't Nvidia go low level?
>>55567178
>That monster had a 500 watt TDP as it was, full 290X performance would have put it closer to 600W.
Fun fact: each hawaii chip in the 295x2 is clocked higher than a reference 290x (1018mhz vs 1000mhz).
>>55563362
amd didnt had sub pixel culling hardware
>amd drivers are shit
how much stupid someone can be really
you moron giving perfomance away is something that happens on both companies nvidia just dont give too much when the new cards are about to come..
>>55567851
>there are people who actually believe this nonsense
>>55562675
smells like 2008 again kek
gtx260 gtx 280
4850 4870
history really do repeat itself
>>55567781
>implying avg fps is the only one that matters
Stupid AMDrone.
>>55567827
Primarily because their software team vastly outstrips AMD's. The closer you get to the hardware the less control Nvidia have of whats going on in their driver (which is absolutely enormous).
Its why Nvidia have seen so many gains for power effeciancy - ripping bits out of the gpu and moving it to software and fine tuning to the nuances that brings.
>>55567869
now damage control found another battleground
kek
in reality nothing really matters what really DO matter tho is the frame pacing to avoid the stutterfest nvidia has
>No crossfire support for vr
JUST
>>55567260
I'm not a huge company though, I just want to make simple games that would be fun.
>>55568915
Or pascal depending on your connections.
>>55565137
That cards fucked then as my 1080 on ultra in doom does a lot more fps than that!
>>55569105
My Fury X can't even break 100fps, have no idea how they got to almost 140...
CPU is an i5 4590k too so I doubt it's a bottleneck
>>55563362
nvidia is the opposite of AMD, performance starts decreasing over time
max performance on day 1, 50% of the performance after 3 years
>>55567700
>Instable cpu oc, you can tell because it only happens at 1080p
Wait, is this a thing?
I've been having some shitty frame times ever since I went up to 4.6ghz on my 2500k.
>>55569249
>>55560712
700 series is now considered legacy. It doesn't get driver updates any more.
>>55569316
Did you do proper stress testing? A chip can run at xyz ghz but not be 100% stable with some cores failing under stress testing. Prime95 is a good one for catching that as it monitors each core individually.
Other options include vrm thermal throttling or vdroop causing clocks to drop (or straight up crash).
>>55569320
The difference in resolution doesn't help when trying to do a 1:1 comparison. Plus from that picture alone we don't know if the test bed or test suite has changed.
