Nvida's drivers are fine, guys, really.

>>59715441

>TITAN XP LOSES TO A FURY X?
TITAN XP LOSES TO A FURY X!

>>59715441
From the youtube video, the TitanXP is sitting at some 50% usage, at full load it's a good 50-60% faster than a Fury X.

Driver issue, a real big one.

Great show Nvidia.
I can't believe Nvidia can't even get their drivers working for 720p CPU testing

>>59715602
/thread

At least it didn't take 150 posts for somebody to realize.

This bait thread will still get 150 replies though because /g/ has always been buttblasted when it comes to AMD and they're doing pretty bad in this "year of AMD".

>>59715441

Could this be related with all the botnet telemetry that you get with Nvidia Drivers? I switched to AMD when Nvidia forced us to sign in with facebook to get driver updates, maybe some telemetry config command hanged and is eating CPU cycles idk

>>59715650
The GPU shouldn't matter much when you're doing CPU testing at low resolution, but in this case unfortunately it does, when a generally 50% faster GPU (Titan XP) is doing worse than a Fury X,

>>59715694
you are
~retarded~

>>59715602
Seriously. I remember watching a Nvidia presentation on how they've achieved zero driver overhead. AMD drivers must be running on black magic.

>>59715441
What the fuck's going on here? I've been watching the GPU/CPU space for a good 13 years at least, and I don't remember ever a time when Nvidia was doing WORSE at low resolution.
In fact low resolution was pretty much AMD's Achilles heel

>>59715441
no it's pretty well known how horrid nvidia's drivers have gotten in the last 5 years, and especially the last 3 have been utter fucking shit. I've never had so many driver fails and driver related bsod than I did in march 2015.

>>59715872
>Nvidia drivers get slightly worse
>AMD's drivers are literally unfixable and constantly crashing

kys

File: 1489941816875.gif (2MB, 256x144px) Image search: [Google]

2MB, 256x144px

>>59715895
>slightly worse
>slightly

>>59715895
>it's ok because the competition is also shit
no anon, YOU kys. God forbid we have standards

File: 1491163475070.png (953KB, 1122x622px) Image search: [Google]

953KB, 1122x622px

>>59715895
>m-muh muh novideo...
>such perfect

>>59715939
>dual GPU card has more utilization than a single GPU card

Rofl

>>59715988
and not any gpu, it is the fucking titan xp.

>>59715939
Jesus Christ what the fuck is Nvidia doing?

>>59716011
sucking cock

>>59716011
Too busy getting clockspeeds up for Pascal they forgot about drivers.

>>59715939
This is pretty outrageous, if Nvidia doesn't fix their driver utilization on Ryzen then they're gonna be competing with Vega with half the GPU not doing its job.
I assume it's again one of the CCX issues, the driver probably migrates threads a lot and hits a lot latency when changing the core complex.
And for something ran in kernel space like drivers that need as low latency as possible, it's suicide.

>>59715650
humor of whole situation is
>muh 720p CPU bottleneck

>>59715895
https://www.youtube.com/watch?v=nIoZB-cnjc0&t

>>59716051
Nvidia isn't in the business of losing sales, they'll fix this soon, they owe Intel nothing.
Actually JHH hates Intel with a burning passion.

>>59715441
In order to make their cards appear more energy efficient, Nvidia didn't include a proper hardware scheduler to any microarchitecture succeeding Fermi.

Now, with Vulkan and DX12 rising more and more in prominence with each coming year, this decision has come to bite Nvidia in the ass. It will be very interesting to see how things pan out with Vega, since it's going to actually have a hardware scheduler from the get go.

All those jokes about Nvidia having shitty DX12 support are turning out to not be jokes at all. The truth is coming to light. Hopefully they learn their lesson with Volta.

>>59716080
Half the games in the OP aren't DX12 titles.
Also >>59715939

>>59715939
>a fucking Titan XP at only 50% usage
And this is the benchmark most reviewers so far have used to shit on Ryzen. Question is how is that none of these "hardware specialists" realized that nvidia's APIs might have had something to do with the CPU's underwhelming performance until literally yesterday?

>>59715988

which just shows that Amd CROSSFIRE works better than nvidias LMAO!

L

M

A

O

faggot

>>59716088
Xfire is great when it works, but it works in like 3 out 10 games without hiccup, not really reliable.

The issue here is Nvidia's GPU being completely underutilized.

>>59716080
and vega also have Tile based rasterization. nvidia will lose its advantage.

>>59716086
I hope you realized after posting that you only pretended to be retarded.

>>59716086
I breached the issue some 3 weeks ago on /g/, but just as speculation on possible issues, never did I think Nvidia was this incompetent.

>>59716088
>>59716092
technically this is not xfire, but dx12's mgpu. I really wonder what would happen if someone uses nividia+amd setup in this game with ryzen as cpu.

File: 1457536032060.png (49KB, 808x805px) Image search: [Google]

49KB, 808x805px

>>59716118

>>59716092

well sure, but with vulkan/dx12, we might actually see crossfire viable again.

which is a winwin for everyone.

>>59716086

its funny, now look at the hordes of incoming videos on the subject from all major reviewers.

also on that note, fuck all the major review sites. they are absolute trash.

i trust the smaller guys more and more at this point.

>>59715441
Man, it's real nice that someone actually bothered looking into the weird shit that's been going on for the last month!

>>59716142
Honestly, I would rather developers spend time on single GPu optimizations instead of wasting it on overly complicated dx12/vulcan mGPU even though it brings 100% scaling.

>>59716156
Intel hates when you do that, please do not do that!

Is everything now just that much more complicated or has 1st party hard/soft compatibility become WORSE and WORSE over time?

>>59716085
Whoops, I didn't notice that Nvidia is even suffering in DX11 titles. LOL! They're getting a good beating on home turf!

In that case, it looks like Nvidia just got embarrassingly lazy in optimization since AMD currently has no enthusiast gpus at the moment.

Since the Fury X (somewhat weaker than a GTX 1070) is going toe to toe with the Titan X Pascal, the performance of the Vega flagship should be incredible, given the available information.

As of right now, it seems AMD's driver game is way better than Nvidia's. Good on you AMD.

>>59716199
That's why this will get fixed, Nvidia won't lose money over something like a load balancing/CPU topology fix in their 10+ million SLOC driver!

File: gulp.jpg (10KB, 164x173px) Image search: [Google]

10KB, 164x173px

>>59716138
Nvidia has hardly anything to do with Ryzen being another massive disaster. It's not their fault if AMD isn't competent enough to optimize their new architecture for already existing hardware.

>>59716160

well the thing with crossfire/sli is that you can stuff two cheap cards in and performance doubles.

two 470 cards should be close to a 1080 in performance.

the benefits of this is that you dont have to replace your card after a year, you just buy another one of the same card.

this can also lead to more use of multiple gpus on a single card which should in theory bring costs down.

>>59716206
They don't need to, if everyone tests Vega with a 7700k the issue won't even be apparent, therefore won't exist.

>>59716209

up to date benchmarks are showing it beating even the 6950x in some tests. and beating the 7700k in gaming.

so the fuck are you on about being a disaster?

please KYS yourself.

>>59716209
Oh it's you, Ramesh. Nice to see you back.

>>59716221
Check the image md5, he's done nothing but shitpost with that image.

I wish he'd get a trip like that other faggot that got annihilated a week ago, so I don't have to read his junk

>>59716209
Software accommodates hardware, not the other way around.
Else anyone in the datacenter wouldn't be wasting massive money on coding to tiptoe around Intel's enormous errata list.

>>59715939
>100%, 49 C
>52%, 65 C
>JUST WAIT TILL IT ACTUALLY RUNS AT 100% GUYS

>>59716275
well, 295x2 is watercooled. it would be absurd if it got too hot.

>>59716275
>>59716297
Yeah, this is a cheap shot. Don't compare a water-cooled card with an air-cooled one.

>>59716275
by your pleb logic temp would double, of course
inbred simpleton

>>59716297
>>59716327
>>59716331
Read the thread, the topic is not the temps or the frames, but the GPU usage.

>>59716357
I totally get the gpu usage, but we were addressing the temps that were brought up for apparently no reason.

>>59716357
then the temp guy should have stayed on the topic.
>>59716377
exactly.

Lets brake this shit down for everyone.

Vega will be on par with a 1080.

But in dx12 coupled with an overclocked ryzen, it might (and im just asuming shit here) it might beat a fucking 1080ti.

Also to note if we are to believe adoredtv, crossfire works so well in tomb rider, that we might see dual vega cards oblitirating the competition.

i cannot wait to retire my 290x and my 4670k (rip in piece)

>>59716434
>Vega will be on par with a 1080
vega will be on par with a 1080ti. there is no reason for AMD to bring a card a year later if that card can not beat a 1080.

>>59716434
That seems to be the case.

If Fury X = Titan X, then, logically, Vega Flagship > 1080 Ti. Of course, this is subject due to change if Nvidia fixes their drivers anytime soon, or if AMD suddenly drops the ball.

This make's AMD's claim of "Poor Volta" appear more plausible than previously assumed.

I don't think Vega will outright destroy Volta per se, but it will trade blows at the very least.

>>59716491
Oops.

*subject to change

>>59716453

but tests already show it close to 1080. like amd's own videos when they showcased it.

>>59716542
>using an underclocked engineering sample
>with drivers suitable for fury x

>>59716542
>>59716556
underclocked engineering sample of a version intended for servers

>>59716542
That's true, but there are many variables to consider:

#1 It was kept poorly ventilated on purpose. The vents were taped up after all. It could have been temp throttling.

#2 It was using modified Fiji drivers at the time. Performance gains with proper drivers can be substantial.

#3 The tests were using an engineering sample. Final clocks have not been established yet. Ryzen had the same thing going on, with the clocks for the initial samples appearing disappointing (e.g. only 3.0 Ghz). Expect much higher clocks on release.

>>59716556
>>59716564
>>59716602

Alright i get the point. im hoping you guys are right, because my plan is still to get Vega. i just cannot use nvidia, i hate their drivers and their business ethics. so if it competes with the 1080ti then im all for that.

Im just hoping we get a Vega Nano for my itx system.

As a 1070 owner I can confirm that drivers have been real shit lately.

>>59716669
You dont have a 1070

>>59716857
Ur right but I have a big mule

Nvidia gimping AMD is wholly welcome and very excused, AMD did the same thing to Nvidia hundreds of times.

>720p

Who is going to use a card like that for fucking 720p? This thread is shit desu.

>>59719496
Intel users, duh.

File: specs __.jpg (351KB, 778x852px) Image search: [Google]

351KB, 778x852px

I truly want the fascist big tech giants to crumble into dust. That said, I still went ahead and bought an i7-7700k because I've had some personal bad experiences with AMD, and I put it inside a Windows 10 Pro 64-bit box because it just works and I no longer have time to burn on configuring the perfect GNU/Linux install. I combined that cpu with an EVGA GTX 1080 Superclocked running with v378.92 64-bit drivers. I am getting an average of 130 fps playing Fallout 4 at 1080p on full Ultra settings and I haven't spent five minutes overclocking this rig yet. I can't complain.

https://pcpartpicker.com/list/BfLrhq

>>59715441
it's funny how AMD put a ton of R&D into their drivers over the past few years and now they're actually pretty damn good and Nvidia is stagnating with bullshit like this.

>>59715895
amd user here, no crashes on any games. love the new drivers they've been great for the last year or two now.

>>59716011
"dude we know you want 100000X tessellated godrays in your next competitive meme shooter game so we went ahead and added 64X quadrocitropic meme filtering for best performance gimping possible."