I am going to buy a workstation for starting my CUDA work. I wanted to find a mobo for X6 1090 with nvidia graphic card onboard, so that I can use that for display and a FERMI GTX for computing. I have been told that the MSI NF980i-G65 would be a good choice, because I can also mount 3 GPUs on it plus the onboard one. However, the stores here in HK have not this item anymore… so I am still searching.
Which AM3 mobo would you suggest with an onboard gpu which can also support at least 2 other GPUs? Furthermore, if the onboard gpu is an ATI… were there be problems in using a NVIDIA gpu for computing??
I am going to buy a workstation for starting my CUDA work. I wanted to find a mobo for X6 1090 with nvidia graphic card onboard, so that I can use that for display and a FERMI GTX for computing. I have been told that the MSI NF980i-G65 would be a good choice, because I can also mount 3 GPUs on it plus the onboard one. However, the stores here in HK have not this item anymore… so I am still searching.
Which AM3 mobo would you suggest with an onboard gpu which can also support at least 2 other GPUs? Furthermore, if the onboard gpu is an ATI… were there be problems in using a NVIDIA gpu for computing??
There are a couple others, but they tend to be mini ATX boards with only one GPU slot.
Really nothing touches the 980-G65… I’m almost tempted to stock up on a few just in case they start disappearing!
I don’t know anyone who has actually gotten CUDA to work with onboard ATI video. Though perhaps it’d be possible if you run Windows, you could use the Tesla Compute driver.
I know it may be hard to find in Hong Kong, but that 980i-G65 is so useful it’s likely worth paying even a pretty big shipping premium to get one.
You might even try eBay, where they’re easy to find, and make sure the seller will send to HK.
The US retailers of Amazon and NewEgg also carry it, and perhaps can do international shipping. (I’ve bought products from Amazon.co.jp and had purchases sent from Japan to the US.)
There are a couple others, but they tend to be mini ATX boards with only one GPU slot.
Really nothing touches the 980-G65… I’m almost tempted to stock up on a few just in case they start disappearing!
I don’t know anyone who has actually gotten CUDA to work with onboard ATI video. Though perhaps it’d be possible if you run Windows, you could use the Tesla Compute driver.
I know it may be hard to find in Hong Kong, but that 980i-G65 is so useful it’s likely worth paying even a pretty big shipping premium to get one.
You might even try eBay, where they’re easy to find, and make sure the seller will send to HK.
The US retailers of Amazon and NewEgg also carry it, and perhaps can do international shipping. (I’ve bought products from Amazon.co.jp and had purchases sent from Japan to the US.)
I just bought one week ago a Gigabyte 890GPA-UD3H with 8GB DDR3 and Phenom II x6 1090T BE. You can see a review here.
It has ATI onboard, and 2xPCIe 16x slots. If you use them both they are working as 8x. From my experience with CUDA this is not important, but for some particular applications can be a problem. This MB is relatively cheap, good quality, and, very important, the PCIe slots are three spaces apart. Considering that all fast NVIDIA GPUs are at least double slot, having space btw them is a must.
I manage to run even to debug locally with NSIGHT! Right now I only have one GTX 460 on this machine, but another can be easily installed.
I just bought one week ago a Gigabyte 890GPA-UD3H with 8GB DDR3 and Phenom II x6 1090T BE. You can see a review here.
It has ATI onboard, and 2xPCIe 16x slots. If you use them both they are working as 8x. From my experience with CUDA this is not important, but for some particular applications can be a problem. This MB is relatively cheap, good quality, and, very important, the PCIe slots are three spaces apart. Considering that all fast NVIDIA GPUs are at least double slot, having space btw them is a must.
I manage to run even to debug locally with NSIGHT! Right now I only have one GTX 460 on this machine, but another can be easily installed.
I bet you are working under Linux because as far as I know there are problems for CUDA applications under Windows when you have an ATI onboard and NVIDIA on PCIe. Or am I wrong?
I bet you are working under Linux because as far as I know there are problems for CUDA applications under Windows when you have an ATI onboard and NVIDIA on PCIe. Or am I wrong?