What is the minimum amount of system memory to use the GPU client in Windows 7?
The reason for the odd question is that I'm using an ESXi 5.5 host (FX8350 8-core, 32GB RAM) with a HD6990 to fold. I have another computer that has been folding, a Core-i3 550 with a GTX260/216. I'm retiring the Core-i3 and have been thinking about putting the GTX260 in the ESXi host. Due to the peculiarities of ESXi passthrough I had to create 2 VM's to use both GPUs of the HD6990 for folding. I've read that adding a Geforce GPU to a computer (real or virtual) folding with a Radeon GPU can cause problems because the workunits can get assigned to the wrong GPU and fail. That means to use the GTX260 I have to create yet another VM.
Normally I wouldn't care how many VM's I create (especially with 32GB) but passthrough of videocards in ESXi 5.5 requires RAM reservations if the VM is assigned more than 2GB. I have enough RAM reserved already that it will become an issue. What I want to know is will a VM with 2GB be able to feed data to a GTX260 efficiently enough to be worth creating, or would I be better off allocating the RAM and CPU to the other folding VMs? Does it make sense to fold on a GTX260 at all?
Thanks in advance!
The reason for the odd question is that I'm using an ESXi 5.5 host (FX8350 8-core, 32GB RAM) with a HD6990 to fold. I have another computer that has been folding, a Core-i3 550 with a GTX260/216. I'm retiring the Core-i3 and have been thinking about putting the GTX260 in the ESXi host. Due to the peculiarities of ESXi passthrough I had to create 2 VM's to use both GPUs of the HD6990 for folding. I've read that adding a Geforce GPU to a computer (real or virtual) folding with a Radeon GPU can cause problems because the workunits can get assigned to the wrong GPU and fail. That means to use the GTX260 I have to create yet another VM.
Normally I wouldn't care how many VM's I create (especially with 32GB) but passthrough of videocards in ESXi 5.5 requires RAM reservations if the VM is assigned more than 2GB. I have enough RAM reserved already that it will become an issue. What I want to know is will a VM with 2GB be able to feed data to a GTX260 efficiently enough to be worth creating, or would I be better off allocating the RAM and CPU to the other folding VMs? Does it make sense to fold on a GTX260 at all?
Thanks in advance!