As far as I understand about NVidia Quadro graphics cards, they are used for CAD and other graphics stuff while GeForce cards are better suited for video games.
Can somebody go in depth and explain how NVidia make one type of card work better for graphics and the other work better for gaming? What is the difference between the two series of cards?
Google it you lazy bastard....
One thing I've noticed is that the Nvidia software for Quadro has more features to make desktop space more usable to create virtual monitors on one screen for easy placement of windows on one screen. Quadro is more expensive because (((professional))). I doubt there is much performance difference between gaming card and quadro besides driver and software tweaks to create two different experiences.
there is no difference ,people since have been flashing their 8800GTs to quadro card a decade ago,im sure its the same now unless there is a hardware lock down on it
The silicon is still identical, AFAIK. It's a lot like how Core i5 and i7 used to be identical ICs with performance throttled through microcode. Since firmware and microcode is the only way you get to do anything with the card at all, they end up supporting different sorts of functionality.
A crude sort of analogy would be the way C++ libraries implement stacks and queues in the form of wrappers on deques, except that you can't access the underlying deque to use the hardware to its maximum potential.
They use higher quality materials so they can last 24/7 without problems.
It's a lie you fucking shill
-t. worked there
More monitors supported monitors on a single card, More RAM, ECC RAM, usually higher memory bandwidth, better compute performance. More stable drivers. Better cooling. What it comes down to is gaming cards are playback devices, optimized to give you the next frame as quick as it can, errors be damned. Workstation cards are optimized to get your frame right. A lot of DCC, scientific and engineering applications can chew through graphics memory, for some workloads it is impossible to get work done on a 1080ti. It just doesn't have enough ram.
>corporation bent on ripping customers off uses higher quality materials on a specific model when marketing jargon sells a product
I doubt that. Gaymen may not have their systems on 24-7 but they are way more abusive to thier tech and have big mouths. If Nvidia used shitty hardware for gaming cards and good hardware for (((quadro))) it would be well known that they are screwing gaymers and gaymers would be endlessly shrieking about it.
Quadro uses different drivers that makes the compute more eficient on expence of graphical performance. It uses less energy and runs cooler too. A simple bios mod and driver change would turn it into a geforce equivalent.
Question is — how exactly does it work?
API calls for the Quadro would use the driver and underlying framework to make the chip be able to do something (say perform AA operations on two parallel pipelines).
On non-Quadro GPUs, such an operation would not have the driver call necessary for the increased performance, or the firmware call that would back the driver call. This may cause that call to fail or fall back gracefully on a less performant pathway, depending.
There are no different drivers for Quadro cards, unless that's a Windows-only thing.
Quadro cards typically have more memory than GeForce cards (the card I use at work has 16 GB; some cards have 24 GB), and most Quadro cards have more display outputs than GeForce cards, and some other fancy features for using a large number of displays across multiple cards. Some Quadro cards have better 64-bit float performance than GeForce cards and some support NVLink. Some have ECC RAM. As far as Pascal cards go, only the Quadro GP100 has the same CUDA compute capability as the Tesla P100 (6.0), while other cards are 6.1, which allows for more CUDA kernel concurrency, which may be better for some applications. That card also has HBM2 memory. None of the GeForce cards have these.
Quadros and Teslas also support RDMA, something GeForce cards doesn't.
quadro cards had 6 times the memory of a normal gpu, last time i checked
My Dad worked at Nvidia and he said you're a lying faggot.
drivers. you can mod a 680 to a quadro 6000 and a 690 to a 680 + q6000 or sli q6000
But you still need to trick the onboard bootloader to load the correct onboard "OS", which is harder to do.
nope, you can just swap some resistors and then the device ID changes so you can install the alternate drivers iirc
No, that will trick the host drivers for sure (as proven when you do deviceQuery and it shows up as a quadro), but the onboard OS will still be off and you're not able to do stuff like GPUDirect RDMA.
no, you can, it literally shows up as a quadro to the OS.
>shows up as a quadro to the OS.
You are not reading what I am writing. The host OS (as in Linux/Windows/whatever), yes. But these GPUs are so complex they also run their own internal OSes (in order to schedule threads on the SMs etc), and the onboard/internal bootloader does not rely on the device ID which is a PCI thing.
If you don't believe me, just try to run anything RDMA on a modified GeForce (with only the resistor trick). This will simply not work.
The quadro 6000 doesn't support RDMA anyway iirc, it's not mentioned on the nVidia website
Well the Kepler and Maxwell variants do, but not for Fermi and older that's true.
I know you can do the same to get the kxxxx quadros, I'm assuming those are the ones without RDMA
I haven't personally the K6000, but uses the GK110 core IIRC, which is the same architecture as the Tesla k20 (which I have tried), and that definitively supports RDMA. So I would assume that anything using the GK110 core would in theory support RDMA if loading the correct firmware.
*haven't personally tried
*but it uses
Sorry for my broken English.
Don't be homophobic
in the past only pro cards like Quadro was supporting 10-bit color but now this is not case anymore
Its basically just trying to get extra sheckles from "enterprise customers".
Quadro cards have minor technical differences compared to Gforce but its mostley marketing for "professional usage". The Nvidia utilities have more features with a Quadro driver like exporting screen setting files.
Some CAD programs even refuse to start when you have a Gforce instead of a Quadro installed.
Pretty sure the CAD studios get a piece of the cake when they make their software locked Quadro.
I don't know if there is some backhand palm greasing going on, but it some applications I've used, not CAD but 3d modeling, some tools are highly unstable with a Geforce series card, but are rock solid with a Quadro
they have more memory therefore stuff like rendering and other memory intensive tasks may perform better, otherwise they are almost exactly identical to their geforce counterparts. Tasks that do not require lots of gpu memory will run same on quadro and and geforce cards