Where can I buy a GPU (Nvidia Tesla, AMD Firestream)?

D

dreammaker2012-02-23 03:04:04

NVIDIA

dreammaker, 2012-02-23 03:04:04

I want to make such a personal supercomputer for myself, most likely I’ll do it in the summer, now I decided to look at the filling for it and saw the light - it seems that the GPU is practically not on sale. AMD Firestream is more interesting in the Ukrainian expanses, but even in the Russian Federation something is a little muffled with this.

The problem is that requests like "AMD Firestream buy", etc. display mostly news. Some of the Khabrovites were generally buying such devices, and where can they be found?

ps I know that you can use “regular” professional cards, but somehow I don’t want to drive fans without the need for graphical output, and it’s not convenient to poke into the server.

Reply

Answer the question

In order to leave comments, you need to log in

6 answer(s)

K

Konstantin, 2012-02-23
@Norraxx

And AMD 7XXX will not work?

A

Akson87, 2012-02-23
@Akson87

So you can't order from amazon? Here I remember recently that HP was proud of servers for this business just, you can have them right away. It seems like there are integrators in Ukraine who work with them.
Can I ask a small question, what are you planning to count? What kind of cards does your software support and on which does it work faster? And the choice may depend on this.

C

Chii, 2012-02-23
@Chii

Order either from system integrators (they will take care of customs and everything, but overpay them 20-30%) or contact the local nvidia office (I don’t know like amd) and arrange delivery. Their phone number in Moscow is +7 495 981 03 00, they do not work very regularly, they deliver Teslas for a long time (there is almost no demand for them in Moscow for some reason), it can take up to a month to go.

A

AxisPod, 2012-02-24
@AxisPod

Are you sure that you need a Tesla or something else, do you want to calculate the structure of proteins at home?))) It’s not easier to take ordinary top videos if the money is not critical, yes, the memory is perhaps smaller, but otherwise not worse.

V

vimvim, 2012-02-24
@vimvim

Have you considered the option of using a ready-made server? For example:
Cluster Compute Instance on Amazon Web Services
( 2xNVIDIA Tesla M2050 "Fermi" GPUs., 2x quad-core Intel "Nehalem" X5570, 22 GB of RAM )
Hourly rate: $2.10 per hour.
Advantages - you can start using right now, scales as needed.
Drawbacks - Price (vs. own server)
aws.amazon.com/hpc-applications/