Slashdot Mirror
NVIDIA GeForce To Quadro Software Mod
babyshiori writes "The NVIDIA Quadro family of professional graphics cards are very, very expensive. But many people know that Quadro and GeForce graphics cards are virtually identical in hardware. Obviously, you cannot just use Quadro drivers with your GeForce graphics cards. However, there is an easy way to soft-mod an NVIDIA GeForce desktop graphics card into an NVIDIA Quadro professional graphics card. Tech ARP shows us just how to do it. 'It all revolves around the driver support for professional 3D applications like 3ds Max or Maya. Quadro drivers allow the Quadro to be used to accelerate the rendering operations of such professional 3D applications while GeForce drivers do not. This is the basis for the premium prices NVIDIA (and ATI) charge for their professional-grade graphics cards.'"
That's because the Quadro drivers are optimised for accuracy, since you are using them to do real calculations you will rely on, rather than small-ish floating point which is all the regular gforce allow.
There are some other things, optimised anti-aliasing for lines, interface layering over the top of render windows, etc.
For a quick and dirty explanation, see NV docs here (warning, pdf file), page 2 onward is where it gets interesting.
...
There are mobile Quadros, so you just need to make sure your notebook is using the same die.
It isn't necessarily so evil. You interpret the value to only be in the circuitry, but the cost of the drivers to be able to use that circuitry is a different matter.
Game users is a very broad base, develop game-optimized drivers and you can develop very cheaply, per person.
The users of engineering software is a very tiny user base, and the cost of maintaining drivers for software that may have several thousand users instead of several million needs to by paid for by those that need to use the engineering software. The rendering for engineering software is optimized for accuracy, game drivers are optimized for speed. There is quite a disparity between the different user bases in size and what they need, so I don't have a problem with charging different prices.
I've had a look at the forum thread linked at the very end of the article. Softmodding only works up to the Geforce 6x00 series. It seems that after that NVidia put in some more checks than only the PCI ID. As reported in the thread, there's no performance increase in professional 3D apps, and OpenGL is broken.
My understanding is that the difference between the two lines is primarily the drivers. It's not that they are disabling functionality on the chip, it's that they only provide drivers for gaming applications with the consumer cards. If a professional modelling app uses OpenGL and GLSL then it will use these cards just fine. With the pro cards, they also provide optimised drivers for more specialist APIs. These may cost the same amount to develop as the OpenGL and DirectX drivers, but this cost is spread around a lot fewer people (the market for 3DS Max is orders of magnitude smaller than the market for whatever the latest FPS game is) and so these drivers cost a lot more per person.
If you are using the pro drivers with a consumer card, then you are using the drivers unlicensed, which is no different from using any other piece of software unlicensed. If you are doing this to run a pirated application better, then I doubt this will concern you, but if you are a business then it ought to.
If someone else wants to write drivers for all of these bespoke applications then nVidia couldn't complain, but I think they'd have a tough job recouping their investment.
I am TheRaven on Soylent News
Because the Quadro's extra features are not beneficial to gaming. I have a Compaq nw9440 with QuadroFX 1500 graphics (256MB, PCIEx16.) The additional features are that you can have the card render to a buffer (GPU-accelerated rendering) and you can use 10 bits per channel (r,g,b) color. Whee! Neither is useful for gaming.
The additional color depth could be neat, if it's even used when your source textures only have 8 bits per channel. I don't know the answer to that. But let's face it, 24 bit color is probably enough for gaming and frankly, I never minded so much when I had to use 16 bit color back in the day because my computers were weak. this is pretty much the only cutting-edge system I've ever had and it was only cutting-edge for a month :)
There WAS a SoftQuadro hack for some of the older geforce cards which had corresponding quadros. Quadros were offered with a lot more memory too, which is not something you can fix with a driver... But the mobile quadros certainly don't have more memory, so there's nothing lost there...
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
There's quite obviously no Quadro to Go card, as nobody but Alienware (in one model, iirc) ever cared to implement swappable mobile graphics modules with the intent of actually selling different adapters.
There, however, are several notebooks equipped with mobile quadro chips, most notably Lenovo's ThinkPad Tp Mobile Workstations. There's even some T series ThinkPads (without the p) equipped with QuadroFX chips.
Also, note that the discussed hack identifies a GeForce series card as it's equivalent Quadro version. If there weren't any mobile Quadros, there'd be no PCI-ID to mimick and the driver would consequently not use it as a Quadro.
Nope. What happens with the Quadro drivers more than anything else is that they do combining and optimization of immediate mode rendering
requests, namely doing "write this poly, now write this one, now write this one, etc..." which is the easier way to do CAD software rendering. It can
only be "sped up" by a little bit, taking advantage of the fragment shader path in a minimal manner. Games, on the other hand, present a command list to
the engine and then say "Go render this pool of commands, come back to me when you're done" to it. Major speed boosts are obtained because the stuff's
running in the host's memory, running in parallel on the host (in immediate mode you ask the chip to render and come back to you before issuing the next
command, meaning the stream processor's WAITING most of the time for you...). If you combine all the immediate mode rendering requests that occur between
a glBegin and a glEnd call, remove redundant operations, and then reorder operations so that the card will not waste steps, you end up with a fairly
massive speed boost with things like CAD software when using the special drivers.
What you're paying for is effectively a "dynarec" core driver for CAD operations to speed up the rendering on select GPUs.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
IIRC someone who used to work for Nvidia posted that the difference between the Quadro and the regular cards was the QA that went into output testing. Basically nobody cared if the consumer card didn't give you a pixel perfect representation of the data being sent it as long as it ran fast, smooth, and looked "close enough". Whereas with the Quadro they went through a lot more extra QA testing to make sure that it rendered the data it was sent accurately.But that is what I heard anyway,so I could be wrong.
ACs don't waste your time replying, your posts are never seen by me.
yes that would be interesting, but you'd still need a dongle like the nuvision 60gx NSR box to get the signal out of the dvi/vga connectors because it doesn't have a 3 pin mini din connection.
http://www.nvidia.com/object/quadro_geforce.html
The whitepaper says that Quadros have got support for window clipping, hardware accelerated clip planes, antialiased wireframe rendering, more memory, etc. Although it doesn't say if the hardware accelerated features do exist in the GeForce family but are disabled by software.