AMD Vega vs Nvidia Quadro P6000

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9TL0YvNjkzMDg3L29yaWdpbmFsLzAwLVBvd2VyLURyYXctT3ZlcnZpZXcucG5n.png (25KB, 711x533px) Image search: [Google]

25KB, 711x533px

>>61437231
Pretty interesting power draw running pro shit and gaming.

So it's 99% of the performance of a $5,000 card for 1/5 the cost?

And that's bad because...?

>>61437309
It's not.
But this should put to rest the use of this card, it's close to a P6000 and actually has double rate FP16.

so, again, it is the drivers.
if it is sandbagging, it is the biggest sandbagging I've ever seen.

bigger than ryzen.

>>61437231
So what's the WX one gonna be like?

https://videocardz.com/70526/amd-preparing-radeon-pro-wx-9100-with-vega-gpu

>>61437231
> P6000
According to earlier benchmarks, Vega has never won against P6000, only against P4000.

>>61437498
Earlier benchmarks meaning SPECview? That's one benchmark.
There's a number of them tested by tom's hardware(german version)

>>61437463
It's not sandbagging, Vega was more than likely designed to be a pro card first with gaming being an afterthought.

>>61437512
> meaning SPECview?
Maybe not. Something, I didn't pay attention to the tool. Seen them on /g/ as well.

>>61437557
Those were the early preliminary workstation benchmarks, and it put it close to a P5000.

But a P6000 wasn't all that much faster than a P5000 in those tests so it wasn't that far off.

Anyway, tested here are AudoCAD, 3ds max, creo and solidworks, which make around 80% of the market so it's fine.

AMD could have easily priced this at $3000 and still be a bargain

>>61437551
That's what Nvidia does too.

>>61437582
AMD was stirring shit up with intel, now they're going at nvidia too? jeez

>>61437582
>AMD could have easily priced this at $3000 and still be a bargain

Oi vey

/g/ - AMD

>>61437607
Nvidia makes 2 different dies for gaming and pro use.

>>61437612
Just saying, it has more than enough performance, the P6000 is like $5000 and is Nvidia's strongest single GPU workstation card(using the full GP102 die)

>>61437636
I don't know about you, but I would even give them $6000 and it would *still* be a great deal.

>>61437711
Lets not go crazy here, Vega FE isn't certified yet.
The WX card will probably cost over $3000 >>61437472

>>61437231
How can there be such a huge fucking gap between gaming and pro performance? This makes no sense.
It performs like a full GP102 in pro stuff, but like a CUT DOWN GP104 in gaming?


'??????

>>61437610
Looks like it Ryzen all over again.

>>61437463
Vega FE was meant as a workstation card. The gaming driver was purely a base number given minimal thought.

The Vega RX should see better performance in gaming due to its higher clocks and more optimized gaming hw.

Prediction:

Low amount of RX Vegas generally.
Most dies will be sold as workstation parts

>>61437970
Drivers.

>>61437625
>GP102 is different for gaymen and pro
?

>>61437231
>better at geometry than fucking nVidia
The fuck?

>>61438329
>The Vega RX should see better performance in gaming due to its higher clocks

But the power draw?

>P6000 performance at similar power draw in pro
>1080 performance at Titan XP(overclocked) power draw in gaming

??!?!?!?!?!?

Like, what exactly is AMD fucking doing? How is this making any sense

>>61438341
I don't think yields are the problem for 484mm^2 die on refined process.
>>61438387
Lower, since power management of any kind (AVFS and power gating) is disabled in FE.
It pumps 1.2v at any given clock.

>>61438407
I dont believe you. Wanna see my 1080TI RECEIPT?? I BET YOU WANNA BITCH MY SHINY 1080 RECEIPT LOOK AT IT LOOOOK!!!!!!11

File: Quadro_GP100_Top (1).jpg (1MB, 4000x2746px) Image search: [Google]

1MB, 4000x2746px

The Quadro GP100 is even better?

>>61438439
No, it's actually slower in these things than P6000.
P100 is only good for FP64 and FP16(until Vega came at least)

>>61438439
No. It's less ALUs (3584).
What GP100 offers is 2:1 FP16 and 1:2 FP64.

>>61438424
did you quote the wrong post or are you having a seizure?

>>61438407
gamersnexus shaved off 85 watts from FE with undervolting it to 1090mv.

>>61438472
You forgot to say it was hitting 1600mhz with that voltage.

>>61438481
I thought it would be understood.

Boy, pajeet really really wants to pull anyther R300.
And the expectations can't even get any lower.

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wLzEvNjkzMzYxL29yaWdpbmFsL0Rvb20tRlBTLTM4NDB4MjE2MC1NYXgucG5n.png (36KB, 709x531px) Image search: [Google]

36KB, 709x531px

>hurr durr quadros are bad on gayming, thats why vega fe suck

>>61438560
Not this shit again please.

>>61438560
So explain why the Vega FE performs like a P6000 in everything besides in games?

>>61438584
Not in everything.
Just in SPEC workloads AFAIK.

>>61438560
It also costs 6000$.

>>61438584
D R I V E R S
R
I
V
E
R
S

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wLzAvNjkzMzYwL29yaWdpbmFsL0RpdmlzaW9uLUZQUy0zODQweDIxNjAtTWF4LnBuZw==.png (36KB, 709x531px) Image search: [Google]

36KB, 709x531px

>>61438575
Its from the op link tho

>>61438584
AMD can't into games because they can't into drivers.

Their solution to their driver writing incompetence was to attempt to get rid of drivers altogether and to push stuff like Vulka and Directx 12 where most of the work would have to be done by the game devs.

AMD are lazy, incompetent poo in loos. At least when it comes to software.

Until that changes they will never beat Nvidia, even if they do have the more powerful hardware on paper.

>>61438592
Who are you trolling?

Solidworks - 10% slower than P6000
Creo - 15% faster
AudoCAD - 5% slower
3ds MAX - it's all the same shit

In fact SpecView is the worst situation for Vega FE hilariously enough, where it's some 20-30% slower

>>61438619
>low-level APIs have anything to do with features like TBR
?

>>61438619
>getting close to metal and squeeze every bit of performance just like consoles is bad
enjoy your dx11.

>>61438449
GP100 beats it in 8K rendering but loses in to P6000 in 1080p

>>61438671
Where did you see GP100 compared to a Vega FE?

>>61438671
>GP100 beats it in 8K rendering
No shit, it's 4 stack HBM2.

>>61437254
that kind of less than 10% away from 1080 power draw
the fuck is everyone complaining about?

>>61438619
>AMD are lazy, incompetent poo in loos. At least when it comes to software.
Their CPU team was in Boston iirc, mostly white. RTG is in China. No pooloos

>>61438682
People are bitching it uses 300W in gaming while performing close to a 1080.

Which is true.
but that's all turned on its head when it uses P6000 like power at P6000 performance.

>>61438682
that one probably was throttling to 1400 mhz. if you hit 1600 mhz with 1.2 volts, it goes about 400 watts.

>>61438704
You don't need 1.2v to hit 1600mhz.

>>61438710
I know. tell that to AMD.

>>61438704
it doesn't though? one thing for sure tom's got power testing right

>>61438716
it says typical consumption. so it was throttling. for it not to throttle, you have to do some adjustments.

>>61438700
I can already imagine the some driver line saying if it detects a game engine it disables all the features.

>>61438609
The p6000 also has more cores than the 1080 ti. So yes, it would perform better from brute force alone.

huh, idle finally at under 20w
that's one huge achievement for AMD

>>61437970
Because GPU design is incredibly unbalanced with bunch of shortcomings.

AMD should just come out and say Vega was made for compute and workstations.

That will be easier on everyone.

>>61438806
>AMD should just come out and say Vega was made for compute and workstations.
No it wasn't.

File: 1474164806244.jpg (58KB, 1494x176px) Image search: [Google]

58KB, 1494x176px

>>61438806
can assume three things from pic related
it's either cheapest ~$300 model that performs as 1080 in BF1
two AMD is fucked
three it's between 1080 and 1080ti but costs $100 less

>>61438950
Or 4: 3584 version is priced as 1080 yet shits on it.

>>61438998
I'd buy that. It needs to be 250w though, higher than 390 power draw kills it for me.

>>61439057
It's probably going to be 225w.

>http://www.cnbc.com/2017/07/18/barclays-predicts-shares-of-amd-will-plunge-35-percent.html
The jew strikes again.

>>61439194
now that's some stupidity. who are these "analysts"? how much money are they making for repeating intel slides?

>>61439268
Probably a lot of shekels (fro, Intel) since even Goldman Kikes stay silent.

>>61439288
do they really think people are this much stupid?
this is plain wrong information. who would trust these fuckers?

>>61437231
do they cost the same?

>>61439324
>do they really think people are this much stupid?
Probably.
>who would trust these fuckers?
Roughly no one.

>>61438664
>Still lose to Nvidia even in Vulkan and Directx12 games

Bravo, AMD. *slow clap*

>>61439352
(You) tried.

>>61439352
>giving the chance to see smooth gameplay and high resulotion even with mediocre cards, whatever it may be the brand.

thanks AMD for pushing Mantle

>>61439194
What if they're trying to engineer a drop in stock prices so they have an opportunity to buy?

>>61439444
That's exactly what they are doing.
Goldman Sachs did the same.

>>61439329
p6000 is $5000
vega is $1000

>>61437231
That's all irrelevant when it can't game at all

>>61437231
Can I do anything with old quadros? Mine meme coins?

>>61439835

Boinc like a motherfucker due to that delicious FP64.

>>61439835
>old quadros
How old?

>>61439908
Not sure. They are large one's my company was throwing away. The cooling looks pathetic though with only one fan on a card larger than a titan x.

>>61439927

They tend to run at low clocks compared to the GTX versions so cooling isn't such an issue. Still free quadros can do some pretty cool things if you can afford to feed them juice to have them crunch data 24/7.

>>61439194
they are all buying nvidia.

http://www.cnbc.com/2017/07/12/nvidia-shares-rocket-toward-record-close-after-analyst-who-missed-the-hot-trade-buys-in.html

>P6000 comparison.

Titan XP is the same but 1000 usd.

Amd btfo

>>61437625
Since when?
They used to just put resistors on the boards, flash the BIOS, or hardcode/burn something inside the chip.

>>61439194
Curtis cited how Intel's server chips have "faster max clock speeds," which outperform AMD's offering in a wider array of applications.
>In addition, Intel's server chips have quicker connections to the computer's memory, according to the analyst.
he seems like an expert

>>61440602
>Titan XP
it is crippled for double precision.
Nvidia btfo

>>61438439
i don't have any preference for nvidia cards but I really wanna rub my dick on that

>>61438693
Even if they were, long as they do the job, who cares outside of forced memes and un-American fagmos

>>61440602
it's ~40% worse at pro appliocations

>>61439643
i just want $1500 performance for $300

>>61441934
Oh fuck off

>>61441980
well i guess i should take the 10% performance loss. $1350 performance for $300

Nothing causes gaymurs to explode in nuclear buttrage like a product that isn't aimed directly at them. Prosumer GPUs, multicore CPUs with lower clockspeeds, anything that isn't for MUH VIDYA is shit and should be shit on relentlessly regardless of it's merit in other regards.

tl;dr ban all /v/ermin.

>>61437344

The problem isn't the card. It's how amd Fucked up again. Vega is now considered a shitty card by normies because of all the 'reviews'. Amd should have waited and released it along with RX.

I'm guessing people who saw the first Vega reviews just ended up buying a 1080.

>>61442145
>The problem isn't the card. It's how amd Fucked up again. Vega is now considered a shitty card by normies because of all the 'reviews'.
They don't care.
People buy nvidia because it's nvidia.
AMD *can't* sell better gaymur product.

>>61442145
The faggots who are easily influenced aren't the market for this card.
AMD has said several times that this is a professional product.

MULTIPLE
TIMES

>>61442350
professionals live a fast live. usually the first best thing will fit.

>>61442231

no they do care. people where actually waiting for Vega to be a 1080ti rival.

so when fags and reviewers say it lands somewhere between a 1070 and 1080, guess what they do? wait or just go for a 1080.

nigga im not making shit up. after the Vega reviews. komplett.no (which is the largest website for selling pc parts) saw an increase in 1080, 1080ti and 1070 sales by close to 25%

>>61442350

they have but they also show the game mode video and touted it as being the best card blah blah blah.

you would expect vega to be faster than the gaming version, much like titan is faster than the ti.

people got confused. again AMD fucked up.

>>61442740
>people where actually waiting for Vega to be a 1080ti rival.
Loud minory on /g/ or leddit?
Come the fuck on, everyone knows Radeon is dead to normies since like fucking g80 times.

>>61438439
This card is so sexy desu

>>61443414
>>61443414
>>61443414

>>61443439
>Hey guys, pay attention to my retarded thread with invalid benchmarks on a workstation GPU. AMDEAD AMIRITE? XDDDDDDDD