NVIDIA's Quadro FX 5800 with 4GB graphics memory is 'the most powerful graphics card in history'
That's some serious boasting by NVIDIA, but this is some serious graphics horsepower. The Quadro FX 5800, already seen in NVIDIA's Quadro Plex D data cruncher, replaces the 5600 at the top of the NVIDIA heap with 240 CUDA-programmable parallel cores and the industry's first card with 4GB of graphics memory. MSRP? Just $3,499 for you big spender -- pennies for the companies who can harness the power for the purposes of oil and gas exploration, 4D modeling, and graphics design.



















Reader Comments (Page 1 of 2)
pang @ Nov 10th 2008 6:48AM
Holy GPU power Batman
Jon @ Nov 10th 2008 9:03AM
Correct me if I'm wrong, but aren't these kinds of GPUs (the Quadro line or example) not any good with graphic-intensive games? But I mean still, holy GPU power Batman indeed!
JerkfacedFed @ Nov 10th 2008 9:15AM
Yes, you are correct actually. These things put all their processing power towards geometry and dont perform well with shader operations. They are meant for CAD work where there are no shader operations, just drawing models as fast as possible is the most important thing.
Kurian @ Nov 10th 2008 10:15AM
Its not powerful at all. All these new 'Quadro' scams coming out now are simply GTX280s with slightly optimized boards and MOAR vRAM.
The Adobe CS4 acceleration is also a scam. There is nothing these can do that a GTX280 cannot, at maybe 95% its speed. A KO Edition GTX280 will rip the balls off this thing.
john @ Nov 10th 2008 12:36PM
Nah, I think the Mayan might have something more powerful from the alien that allowed them to design their pyramids and study the sky.
codelogic @ Nov 10th 2008 1:16PM
@JerkfacedFed
That's not accurate. The primary difference between the two lines is the QA and features that go into the drivers. In terms of hardware they are very much alike.
I LOVE THE CAPS LOCK KEY @ Nov 12th 2008 12:28AM
In the past with some Quadro cards, you could simply opt to run standard GeForce Drivers (read: second hardware profile or dual boot) and get decent 3D gaming performance out of them. However, I am not up to date and do not know if this still works with more contemporary nVIDIA Quadro hardware.
The Same used to be true with ATi's FireGL line (Pre-AMD), but I don't know you can still run Catalyst drivers on more modern stuff.
Dopefish @ Nov 10th 2008 6:55AM
I dont know, but wouldn't it be cheaper to get a rendering farm stuffed with loads of slightly older cards...
AlekZander @ Nov 10th 2008 7:33AM
Not in terms of electricity cost over the long run. That's what these big number-crunching data centers care about more.
broli @ Nov 10th 2008 9:37AM
I think any research facility will think hard and long on whether they really need the extra RAM. With that same price you can build a super computer of 9xGTX 280's. As long as you're using CUDA you don't need to change anything.
vfxAdmin @ Nov 10th 2008 3:38PM
Keep in mind that there is not a lot of places that render in GPU.
GPU is still only being used at the front end while making a model.
Unless you are rendering realtime, for the price dual quadcores are the way to go (for now)
blueangel00100 @ Nov 10th 2008 6:56AM
Most powerful....until next week and the next model comes out.
RogerMcDodger @ Nov 10th 2008 10:28AM
Being that it has been 19 months since the FX 5600 came out, that is probably unlikely.
Chris @ Nov 10th 2008 6:57AM
I can't wait for Intel to enter this market. Nvidia need a kick up the arse.
Chuckles McGee @ Nov 10th 2008 7:47AM
Competition drives innovation!
Erb @ Nov 10th 2008 6:59AM
"4D modeling"
Woah, It models TIME? AMAZING!
Bobs @ Nov 10th 2008 7:17AM
Or thousands of hyper cubes
kal326 @ Nov 10th 2008 9:10AM
It's so powerful that Nvidia had to utilize a new dimension just to show its true potential.
thedesolate1 @ Nov 10th 2008 4:58PM
Kind of like how the PS3 can push 4D graphics.
Remember that one?
phil @ Nov 10th 2008 5:54PM
Actually, the fourth dimension, time, is used to predict fluid and gas migration in oil and gas reservoirs over a specific period of time.
So yes, it is used.
nuclearjello @ Nov 10th 2008 7:04AM
in b4 can it play crysis
Ray @ Nov 10th 2008 10:51AM
I wonder is crytek likes being at the end of this joke?
Rav @ Nov 10th 2008 7:04AM
Does it play Doom?
Rowan @ Nov 10th 2008 7:06AM
You spent harness wrong j.q.
Thi mam(kris120890) @ Nov 10th 2008 7:17AM
You spelt spelt wrong dumbass.
Algorhythm @ Nov 10th 2008 8:10AM
You both spelled spelled wrong.
http://www.google.com/search?sourceid=navclient&ie=UTF-8&rlz=1T4ADBF_enUS299US299&q=define%3a+spelt
Thi mam(kris120890) @ Nov 10th 2008 8:31AM
No we haven't you idiot It's part of the English language. Might not be the usa because you don't know how to spell.
Thi mam(kris120890) @ Nov 10th 2008 8:37AM
http://www.askoxford.com/concise_oed/spelt?view=uk
Scotty Doo @ Nov 10th 2008 7:07AM
Splendid.
Samboini @ Nov 10th 2008 2:01PM
Thanks.
Michael @ Nov 10th 2008 7:18AM
If it is the most powerful it wont be for long,
ATI is planning to get serious in the workstation market and take that away from Nvidia ;)
Cole Mitguard @ Nov 10th 2008 7:18AM
Looking back, its really amazing how fast these graphics cards or improving. Reminds me of the old processor speed races back in the late 90's early 00's when every month processors would increase by some "at the time" huge leap of several hundred MHz, haha. I like what I'm seeing though, don't play many games, but its always better for the video editing... and I cant wait to see what OS's are going to start looking like when these super powerful graphics chips become common place. 4GB!!! Damn.
ChrisAR @ Nov 10th 2008 7:28AM
But can it play Outside?
http://blog.codefront.net/wp-content/uploads/2007/08/outside_mmorpg.jpg
Harry Wagstaff @ Nov 10th 2008 9:02AM
Please tell me a poster of that exists.
Reader @ Nov 16th 2008 8:33PM
How does their PvP ruleset work? I hate it when guards come chasing me in town because I killed someone.
Lorre @ Nov 10th 2008 7:29AM
Amazing card, just don't use it with Vista Home or you'll have NO RAM left :-p
But NVidia probably won't have 32bit drivers anyway:)
Laughing Man @ Nov 10th 2008 3:25PM
You really have no idea what you are talking about do you.
RobertMfromLI @ Nov 11th 2008 3:10AM
Laughing_Man, Lorre is correct - as is everyone else who pointed out that you are wrong. The way it works in most 32bit OS's is that every piece of hardware gets mapped to a memory address - the Intel and AMD CPUs, in 32bit mode can access 4GB. Take that 4GB and remap a device's address space in the first available region. When complete, that 4GB is now less. You can look it up online. Thus, using this card in a computer running a 32bit OS would leave you with negative available memory. 4GB max system RAM/address space MINUS the PCI and other busses MINUS all other devices LEAVES something already less than 4GB - MINUS 4GB+ for this card leaves you a negative number.
Now... theoretically, on certain versions of Windows, and on supported hardware, you can enable PAE mode for the kernel, which will then do 36bit addressing and help alleviate that issue. Or you can run a 64bit OS.
Or, one final possibility... nVidia can write drivers that swap memory the old fashioned way (ie: no flat memory space accessing), though there may be a big speed penalty... because lets say they use 512MB blocks... it would take 8 swaps to handle the card's full 4GB - and since the data could not be acted upon in one chunk, it would mean those 8 swaps may have to occur numerous times to complete a transaction.
Laughing Man @ Nov 18th 2008 12:49PM
Thank you for joining the dumb ass brigade. Lets talk about computers. First off, by your logic, since my CPU has to address all memory in my computer, that would mean my hard drive takes from that as well. Seeing as how my hard drives total around 880 GB, I am pretty damn sure that is wrong.
The way it ACTUALLY WORKS is that limitation is imposed on your system RAM only. The GDDR RAM on a Graphics Card (unless integrated) is address and accessed by the GPU on the card only. You CPU cannot address that RAM because it does not have access to it. This is why graphics even have an address controller stat (in bit size) in the first place, because it addresses its memory. A dedicated graphics card is basically a small computer inside your big computer. It has its own memory, own processor core, etc. The size of the memory on you graphics card has no barring on your computer available memory (unless integrated).
Randy @ Nov 10th 2008 7:37AM
Sweeeeet, I bet you can get Crysis running at 18fps!
VeganFreak @ Nov 10th 2008 7:54AM
at 3840 x 2400 ? yeah it probably can =]
chrisk1590 @ Dec 21st 2008 2:15PM
Probably not, but it'll be the most detailed 1FPS you can get!
PeterPumpkin @ Nov 10th 2008 7:45AM
I'll take 2.
Can anyone tell me where i can sell both my kidneys?
Samboini @ Nov 10th 2008 2:03PM
Africa, South America and China are good places to start. Well they were when I sold mine at least.
Mr Handy @ Nov 10th 2008 7:49AM
4D modelling is a type of 3D seismogram synthesis migrated over time (primarily only displaying the first derivative of the horizons of interest) :P It differs from conventional 3D modelling which is normally static and is used very differently. Hence the name....
Shinigami @ Nov 10th 2008 9:10AM
Now I want 4D modeling in PC games! Like changing the seismogram data with my rocket launcher xD
Thanks for explanation!
loosely_coupled @ Nov 10th 2008 8:08AM
Seriously, what happens with a 32-bit OS with this card??? The 4GB of VRAM would use up all the 32-bit addressable memory space !
otomoton @ Nov 10th 2008 8:46AM
You don't use a 32-bit OS for this type of data processing. You have to graduate to a 64-bit OS to run with the big boys...
Laughing Man @ Nov 10th 2008 3:28PM
You are an idiot. That has nothing to do with your computer's address space. The VRAM on a graphics card is only accessable by the graphics card. You computer doesn't see that memory, so it can't address it.
thethirdmoose @ Nov 10th 2008 5:40PM
Actually, it is addressable by the computer and does count towards the 4GB memory limit.