How well would Quadro's and Tesla's do in AAA games?

[Jakers038]

Yes, I am aware that these cards are meant for CAD renders, maybe video rendering and some very specific tasks like that.

 

However, if you were to compare game performance of those cards to mainstream cards, how would a Quadro 4, 5 or 6 thousand compare to GTX cards?

 

For example;

Quadro p5000 = Gtx ____

 

I've seen some video about cloud gaming, where you rent out a high end computer to do the work for your computer and many of these cloud computers use not GTX cards, but Quadro's and Tesla's, so I'm just wondering what would be the speed of them if they were compared with mainstream gaming cards. (and yes, I know they are extremely expensive, but that's not the topic, just pure performance)

[vanished]

Linus has checked this before.  A Quadro performs roughly the same as a similarly specced GeForce card in gaming.  There's no reason for it not to.  Same story with a Xeon vs i7.  As for Teslas, these are pure compute cards with no display output so that makes things a little messy.

 

 

[Droidbot]

P600 = 1030
P620 = Lower than 1050 performance
P1000 = 1050 Non-Ti
P2000 = better than 1050Ti but worse than 1060
P3000 = 1060
P4000 = 1070
P5000 = 1080
P6000 = 1080Ti

[Wockinine]

Quadros are basically reskined GTX (and soon RTX) cards with better cooling and added extras. There is usually a HUGE gap in price between let's say a GTX 1070 and a Quadros of the same spec. Tesla are confusing when it comes to displaying the graphical content but same sort of story.

[aisle9]

More or less equivalent to the GTX/RTX equivalent of the GPU they're based on. Quadro drivers aren't optimized for gaming, obviously, but they are more stable out of the gate. It's incredibly rare to hear about a major Quadro driver glitch because massive enterprises are depending on these cards to perform exactly as advertised. Meanwhile, in GeForce land, how many times has your display gone black since installing the new driver?

[Droidbot]

Just now, Wockinine said:

Quadros are basically reskined GTX (and soon RTX) cards with better cooling and added extras. There is usually a HUGE gap in price between let's say a GTX 1070 and a Quadros of the same spec. Tesla are confusing when it comes to displaying the graphical content but same sort of story.

Close but no cigar.

 

Quadros are the same core sillicon as the GTX and RTX cards that us consumers have access to, but with similar cooling (when compared to ref/founders), lower TDPs, and the most important part - professional drivers. These allow much higher performance in "double precision" operations allowing professional programs to run accurately (no errors in final product) on these GPUs. As well as a fuckton of VRAM, because everybody needs a fuckton of VRAM.

Since these cards are the only thing with this performance and double-precision capability (which is available on consumer cards but runs at a MUCH slower rate), Nvidia can charge significantly higher prices for these cards when compared to their consumer counterparts.

Teslas are similar - but built primarily for the datacenter (hence the lack of video outputs). These GPUs can be assigned to VMs and remotely accessed, or used as part of a render farm (or machine learning farm), which is their primary purpose. These cards have a much higher performance density when compared to their consumer and Quadro counterparts, as they are usually inside well-ventilated chassis with a large amount of airflow.

 

Sorry about the wall of text.

 

[manikyath]

2 minutes ago, Droidbot said:

Close but no cigar.

 

Quadros are the same core sillicon as the GTX and RTX cards that us consumers have access to, but with similar cooling (when compared to ref/founders), lower TDPs, and the most important part - professional drivers. These allow much higher performance in "double precision" operations allowing professional programs to run accurately (no errors in final product) on these GPUs. As well as a fuckton of VRAM, because everybody needs a fuckton of VRAM.

Since these cards are the only thing with this performance and double-precision capability (which is available on consumer cards but runs at a MUCH slower rate), Nvidia can charge significantly higher prices for these cards when compared to their consumer counterparts.

Teslas are similar - but built primarily for the datacenter (hence the lack of video outputs). These GPUs can be assigned to VMs and remotely accessed, or used as part of a render farm (or machine learning farm), which is their primary purpose. These cards have a much higher performance density when compared to their consumer and Quadro counterparts, as they are usually inside well-ventilated chassis with a large amount of airflow.

 

Sorry about the wall of text.

 

forgot a small but very important detail about tesla's:

they're usually designed to fit in a GPU server chassis, which means a cooler with front-to-back airflow, and a power connector on the "short" side of the card, instead of the "long" side as it is on most workstation cards.

[Droidbot]

3 minutes ago, manikyath said:

forgot a small but very important detail about tesla's:

they're usually designed to fit in a GPU server chassis, which means a cooler with front-to-back airflow, and a power connector on the "short" side of the card, instead of the "long" side as it is on most workstation cards.

Not always. Cards like the P4 are meant for other chassis such as blade servers and 1U servers, which aren't exactly GPU chassis.

[ZeouLs]

1 hour ago, aisle9 said:

More or less equivalent to the GTX/RTX equivalent of the GPU they're based on. Quadro drivers aren't optimized for gaming, obviously, but they are more stable out of the gate. It's incredibly rare to hear about a major Quadro driver glitch because massive enterprises are depending on these cards to perform exactly as advertised. Meanwhile, in GeForce land, how many times has your display gone black since installing the new driver?

like 3 times and 2 game crashes

[Jakers038]

So basically if a P6000 is a GTX 1080 Ti, would it get more or less equal FPS and get better results in VRAM hungry games? I can't think of a single game that uses nearly as much VRAM, but maybe on really big resolutions, multiple screens or whatever.

[aisle9]

6 hours ago, Jakers038 said:

So basically if a P6000 is a GTX 1080 Ti, would it get more or less equal FPS and get better results in VRAM hungry games? I can't think of a single game that uses nearly as much VRAM, but maybe on really big resolutions, multiple screens or whatever.

Quadro cards usually won't quite match GTX because Quadro drivers aren't optimized for gaming. They will be pretty damn close, though.