Printable View
I have 7 of these cards pulled from working Dells;
They include both the vga and dvi dongles. These are low profile PCI-e x16 cards but have the full profile bracket. No software or drivers are included. The drivers can be downloaded from nVidia.
$20 plus shipping. (or make me an offer)
I have 7 of these.
I also have 2 of the 64mb versions (p283) for $10 + shipping.
Bumpitty...
What exactly are these good for?
I never have understood the point of a workstation graphics card and what the difference is. I might very well be interested if they could fold well or something similar as I have a second PC that doesn't need to game.
Let me know, and maybe I'll pick one up at Rod's and Mod's when I see you there.
Workstation video cards are useful not only for low profile machines that need multiple monitors, but they are certified to work with a lot of high-end CAD and such programs. They are designed for ultimate stability and accuracy of reproduction.
...and basically these are the low end cards really only good for multiple monitors...
That's all I'll need. If you want to bring one of the 64MB ones to Rods and Mods I'll bring $10 for you.
can you ship one card+ cables in NC?Quote:
Originally Posted by Airbozo
I have 7 of these cards pulled from working Dells;
They include both the vga and dvi dongles. These are low profile PCI-e x16 cards but have the full profile bracket. No software or drivers are included. The drivers can be downloaded from nVidia.
$20 plus shipping. (or make me an offer)
I have 7 of these.
I also have 2 of the 64mb versions (p283) for $10 + shipping.
GPU cards were stagnant over much of the last four years, same old ASICs on same old 28nm (all fabbed at TSMC, no less), just annual rebadges with endless minor (almost trivial) tweaks and twiddles. Only since 2014 have we seen some truly major GPU revisions come out, new architectures running at insane speeds, loaded with phat VRAM capacities, plenty of multi-monitor pixel crunching for everyone. Interestingly, the extreme end (NV Titan cards) bridge the performance and price gap between gaming and workstation cards ... too bad so many fools think the latest-greatest Titan X is going to have the staying power of the old Titan cards (it won't, and I think in a few years it will price out much as a 780Ti does today because it lacks the FPDP which makes the old Titans retain their appeal to prosumers with serious needs and budgets).
Plenty of pulled workstation cards are floating around craigslist these days, seemingly more each month. I suppose more of their "obsolescence" (in the eyes of enterprise users) comes from warranty expiration than from comparatively limited performance.
++This. So much this.Quote:
Originally Posted by Konrad
At work I support an organization that maintains one of (if not the) largest server fleets in the world. The second something is out of warranty I want it gone, out of my sight, and never to return. If I can't easily get parts to fix it I don't want it anywhere near me. Worst case, keep it around until it breaks, but honestly with old hardware it rarely makes financial sense to keep it around even if it does work just as well as it did when it was new.
Sadly, too often people and their poor decisions make my ideals in these regards unattainable. :( Doing better all th
Send me a PM the next time you're pulling a handful of "expired" Tesla GPU Compute Modules or somesuch, haha!
Er, so long as they can plug into desktop PCIe 3.0 slots, of course - I don't happen to have a rack-mounted server at home, so not a whole lot of use for things like 1U blades. (Although I confess sometimes drooling over some of those chassis towers. But it seems that while server folks will happily build anything to custom order you wanna pay for, they don't really bother to advertise particulars about exactly what they're selling to the public. And so I'm am forced to buy into "enthusiast" stuff with the understanding that nothing is perfect until I mod and customize it to my specific needs.)
Yeah, SuperMicro, Tyan, and ASRock are three of the few companies I know of that sell a lot of the same stuff retail as they do to big customers.
Haha, we'll probably be getting rid of some in a couple years, actually...but only if you want to buy 10-20 racks at a time... :whistler:Quote:
Originally Posted by Konrad
In the meantime, you can always rent one for $0.65/hour/GPU. ;)
Damn, $5700 per year for each GPU?
Aw, no thanx lol. Even a pair of $1300 Titan X cards is already pretty excessive for my needs, in price and in performance. I know people who spend less buying cars then I did on computers, but I still can't resist great old GPUs selling for ten bucks each, lol.
I am one of those people who spent more on my computer than my transportation (motorcycles). Kind of puts a crimp in the computer budget to be honest.
I have a couple of Quadro FX 4800's lying around now....
How often are you actually using those GPUs' power though? Let's say for the sake of argument that the Titan X offers the same performance as one of those instances (talking GPGPU or other stuff that can be done remotely; obviously gaming/etc is not yet a feasible application for these). I don't remember what model card they use, but it's likely to be in the same rough ballpark. If you use the card less than ~23% of the year (that's total time, not just time you're awake), then it would be cheaper to use the EC2 instance.Quote:
Originally Posted by Konrad
Think of it less as "this would cost me $[X] per year with it running the whole time" and more as "what would it cost if I only paid for this when I actually used it".
lol, I lust after a Titan X pair, but I have to make do with "only" a pair of 980 Classifieds. About $900 each (with backplates), but I think I'm getting great value out of them, and I expect to continue getting value out of them for at least a couple more years.
I do CAD crunch/render/sim stuff enough to want the best I can justify buying, lol. Hate those frustrating little pauses (seconds, minutes, even hours) every single time I execute even the most trivial little schema or syntax changes. I game a bit, not all that much, and fps isn't really my thing unless I can find games which have particularly engaging single-player stories/environments or unless I can find some not-immature multi-player opponents with roughly the same "skill" level I possess. For me, the most appealing feature of the Titan X is the 12GB onboard VRAM, mainly because that seems to be where my GPGPU bottlenecks occur the most. I'm an oddball since I game a little, I compute a lot, and I don't really need the DPFP or godlike support/warranty which define value for workstation GPUs. It's handy to "have it all" in one machine, but I'm just as content to work (or game) one one while another crunches away at some other work task.
