>5120 cuda cores
>7.4 TFLOPs
This is a beast, how will AMD ever recover

Attached: Quadro_GV100-front_678x452.jpg (678x402, 28K)

Other urls found in this thread:



Oh boy, I can't wait to do some architectural rendering!

fp64? Or why is that impressive?

What if I unironically want to do architectural rendering?

>7.4 TFLOPs
vega is ~25TFLOPs

Wait for NAVI

> $9000
> 7.4 TFLOPs

meanwhile at AMD:
> Radeon Pro WX 9100
> 12.29 TFLOPs
> $1600

jeez... how will AMD ever recover?

> fp64?

This and soon more energy efficient with 32GB of HBM built on 7nm (Vega20)

Over 2020 release new Arch non-GCN

that's cool enjoy your GPU

what the fuck did you smoke op? That's HBM2 memory.
> 7.4 TFLOPs
don't forget that's fp64 performance, and the WX 9100 (basically a "FirePro" Vega64) already gets 6.14 TFLOPs on fp64 for less than 1/5 the price.

Attached: wx 9100.jpg (1920x1080, 650K)

Is it any good for gaming?

>WX 9100 (basically a "FirePro" Vega64) already gets 6.14 TFLOPs on fp64

>12.29 TFLOPs
What's it's FP64 performance?

It'll probably cost close to a iphone as be just as useless.

I'd rather pay one third the price and have a fully foss graphics stack.
Not to mention a company compliant with the wayland and vulkan APIs.

I guess I just have different priorities than you.

>probably cost close to a iphone
More like 10x the cost of an iphone.

Workstation GPUs are not optimized for gaming. A Geforce will perform better for that purpose.

2 + 2 = 4.1

Wait for the linus tech tips video

The Quadro V100 is 32GB of HBM2 ( ECC )

under .8 TFLOPs (ayy lmao)

wait for Volta

Attached: 1464270793313.jpg (610x288, 52K)

Architects are satans gift to humanity
Making stupid building with purely design in mind and getting mad when the engineers have to change the majority of it because none of it is structurally sound.

>what the fuck did you smoke op? That's HBM2 memory

Fuck architects in general.

Fuck architects, not to mention the ridiculous waste of materials on stupid shit for "art"

There's this building close to my home that has its entire exterior lit by rgb like a fucking gaymen pc Jesus Christ I swear fucking hell

And some buildings have shut jutting out like balconies just that nobody can get on them like dude what the fuck are those for you shit head

why do nvidiots act like they enjoy getting assraped by jensen on "pro" card prices?

who uses fp64? why does that matter?

My company informed me today that I have 2 ordered to replace my Titan V. I could care less about he AMD/Nvidia circlejerk going on. It is pretty nice to not have to pay for them.

It's not hard to understand. The more you buy, the more you save.

>4096 Stream Processors
>24,6 TFlops
How will Nvidia ever recover?

Attached: 0bda2e06f0f8f90c2f9cb27e7a1f2e0c_L.jpg (620x465, 132K)

Attached: Avatard.jpg (769x733, 120K)

So what actually drives up the price for Nvidia cards vs and counterparts of similar spec?

I know the Jews hand at hand in this but is there any actually reason they're pricier?

being this retarded

Double precision
Vega64: 0.79TFLOPs
V100: 7.4TFLOPs

Stop lying you dumb son of a bitch

WX9100 is only 1/16 FP64, it's not 6TFLOPS you stupid ignorant moron

Leather jacket man went full retard with his HUEG dies, which have lower yields.

t. colapsed "artistic" bridge designers

By actually selling some cards.

You cant really do proper simulations without FP64. The part of the market that the card partly targets.

>7.4 TFLOPs
2015 called it wants its tflop rating back

w-who c-cares mu-h-h sales


Attached: 1520980633175.png (171x278, 93K)


The fuck do i need ECC for?

for memes


For correcting meme errors

>t. collapsed "artistic" bridge designers

Attached: bridge 4.png (892x1288, 1.05M)

poo in the loo amdfags

>fully foss graphics stack
Do you know what quadro drivers are, you ignorant desktop ricer?

That's 7.4TFLOPs of FP64, if comparing FP16 like Vega it's 29.6TFLOPs

Due to the presence of radiation/cosmic rays, memory devices can sometimes flip bits for apparently no reason, and ECC works to prevent that. In a typical desktop system running programs for limited periods of time, you rarely experience these errors because it's likely not affecting anything important to the stability of the system.
However, if you have tons of RAM, and it's actively being occupied for days at a time, the rate at which errors pile up increases dramatically. If you were doing something like molecular simulation, where the results are based upon countless iterations of data, even one flipped bit at some level can invalidate the result. Even worse than simply crashing the program is it running successfully but producing the wrong data.

Quadro GV100
7.4 TFLOPS double-precision (for science)
14.8 TFLOPS single-precision ( game power)
118.5 TFLOPS tensor cores (AI deep learning BS)

Not sure about that number.


Vega 64 beats 1080ti and older quadro by twice the amount of fp64 scientific/financial analysis.


By being able to make x86 CPUs.

Attached: .png (459x538, 14K)

Without benchmark that number is meaningless. At least show how it compares to the older nvidia quadro.

Attached: hurr.png (966x478, 54K)

Yes Jensen, also PLASTER, you did teach me two things

Good look real nice in a chink warehouse mining crypto

That's not FP64 performance though.

Make up your mind.

Attached: 1519201410197.png (2160x1400, 1.35M)

Attached: durr.png (2000x1162, 1013K)