[IQ Test Card?] What's the purpose of Quadro P600? Why not 1050?

[spacehunter0413]

Today I found a weird card while our IT guy was playing with some old PCs. Quadro P600 has same architecture and less CUDA cores than 1050.

 

[CommanderAlex]

Is there a question you would like to be answered?

[TVwazhere]

-Topic  locked-

 

This is best suited as a status update. 

[spacehunter0413]

So, our IT guy pulled out a weird card from a dusty PC, actually a workstation, (called ThinkStation P320, probably a pre-build PC). And I found this Quadro P600, I know that we often use Quadro cards for some sort of professional usage, graphic processing. I compare the card to gtx1050 to see the specs difference. And to my suprise, the card has same architecture, same RAM, and bits, and less CUDA cores, but costs more. 

 

In addition, this card released a year later than gtx1050. So why the heck nvidia wants to do that, and why anyone would like to buy it? What's the target market of this card?

My assumption is the good chip yield went wrong in that batch and they have to block some CUDA cores to make a low tdp product.. But why does it cost more? Does the card really has something special about itself? I did not find any dedicated accelerator in the specs. And mining fever is not a thing back in 2017. 

 

The PC is inside a computer lab BTW. Our floor are doing ML related stuff so I doubt it tensorflow or pytorch has relatively better performance on the card.

 

Here is where I got the specs https://technical.city/zh/video/GeForce-GTX-1050-vs-Quadro-P600

Edited August 16, 2022 by spacehunter0413
Added topics

[SorryBella]

1 minute ago, spacehunter0413 said:

why anyone would like to buy it?

ECC, Quadro unwavering support bias for engineering and content creation optimization, low TDP, and "the enterprise general 0 fucks about value for money when it gives them reliability and support" cliche.

[mariushm]

Certified/Validated drivers for use with various applications (Autocad, for example,3d rendering applications ... the old "oh no this bridge design will break because you didn't see the issue during render due to a floating point error in the video card drivers when rendering on screen" cliche , other stuff gamers don't care about.... 

No extra power requirement, which can mean packing more of these in a system or a rackable server and using them in parallel (low thickness and maybe low height video cards)

 

 

 

 

[minibois]

Usually Quadros are preferred when it comes to professional 2D/3D modeling programs, as they are specifically supported in these programs. Often the performance on Quadro cards is better than their GeForce counterparts in these programs, but it really depends on the situation.

This video is a couple years old, but should explain shed some light on the situation:

17 minutes ago, spacehunter0413 said:

And mining fever is not a thing back in 2017. 

Actually there was. Off the top of my head, the worst of it was from May to about December of 2017 and availability of graphics cards only became better in 2018.

The RX 480/580 was miners favorite cards, shortly followed by the GTX 1060. But all the rest of the cards - especially RX 470/570 and GTX 1070 - were bought too, as people scrambled to get cards.

Here is a video about mining in that time from LTT:

 

They even started out the video talking about the lack of availability.

[PDifolco]

Let's say Quadros are "professional" version GPU, much more expensive but more reliable, more efficient, better support

A P600 don't make much sense now, but look at an Quadro 4000, for $1,200 you have a much smaller and efficient card equivalent to a 3060

 

[8tg]

Here’s a brief explanation as the good details have already been explained above by others.

 

Nvidia GTX/RTX:

 

Nvidia Quadro/Tesla:

 

Just a van all the same, you can cram 6 kids and a dog in a sprinter van, and you can also put a pallet or two of freight into a Chrysler town and country.

But these are different vans designed for different purposes, even if they’re really about the same in general concept.

 

Thats kinda what Quadro cards are, they’re gpus that instead of being tailored to gaming performance are instead tailored towards professional applications. The bios is different, the drivers are different, it’s clock speeds and such are different compared to its consumer equivalent.

Ever notice how something like the Quadro RTX A2000 manages to outperform the RTX 3050? Even though it’s a half height card and draws half the power? That’s down to silicon lottery a little bit but it’s also that a use case or goal is designed and in order to operate a specific way a card is built around that. It’s not “how powerful can we make this gpu we fab a million of”. Instead it’s “what’s the most optimal solution we can come up with for this product requirement”.

Beyond that as others have stated, Quadro drivers are verified for absolute compatibility and stability, a lot of professional software is verified to work 100% with Quadros, etc. For a lot of businesses big and small you’re paying more for a guarantee against having any problems. There’s no guessing if your RTX 3050 will shit the bed trying to handle some gpu accelerated rendering in Autodesk Maya, switch to the A2000 and it’s just going to work and nvidia promises it’ll work through extensive testing.

[Skiiwee29]

@spacehunter0413 I have merged your threads together. Please stick to a single thread for the same topic.