ive just won a second titanX gpu on ebay.
so now i have 4 pcie cards to plug into 4 pcie3.0 slots.
i need to decide the best arrangement.
populating all 4 pcie3 slots i get the following
one at 16x
two at 8x
one at 4x
there are also two pcie2.0 1x slots.
i now have 2 Titanx for rendering
an older gtx670 for viewports
and a pcie 4x NVME adapter for a samsung 950 pro ssd.
my idea is to put the two titanx'es in the 8x slots
the ssd in the 4x slot, and the gptx670 in the 16x slot.
however something annoys me about putting the oldest board in the fastest slot!
also im not sure if the 16x slot is connected driectly to the cpu and the others are not.. ive read that the ssd benefits from being on a slot connected directly to the cpu.
im guessing that there is no point running one titanx on 16x and the other on 8x, since they both need to load the same data, and therefore the faster connection will just wait for the slower one to catch up. also i understand that the difference for rendering is minimal between 8x and 16x.
any thoughts?
does the 670 for viewports benefit from the 16x slot? im not gaming, just max viewports.. if i wanted to add another pcie ssd, id have to shift one card of the above down to pcie2.0 1x slot.. would that bottleneck the 670 for viewports noticeably? i imagine massively.
alternatively i could ditch the 670 and use one of the titanx'es for viewports too.. its not like i can work when rendering anyway. does this impact available gpu ram for rendering? i guess so.
too many combinations, its hurting my head!
cheers!
so now i have 4 pcie cards to plug into 4 pcie3.0 slots.
i need to decide the best arrangement.
populating all 4 pcie3 slots i get the following
one at 16x
two at 8x
one at 4x
there are also two pcie2.0 1x slots.
i now have 2 Titanx for rendering
an older gtx670 for viewports
and a pcie 4x NVME adapter for a samsung 950 pro ssd.
my idea is to put the two titanx'es in the 8x slots
the ssd in the 4x slot, and the gptx670 in the 16x slot.
however something annoys me about putting the oldest board in the fastest slot!
also im not sure if the 16x slot is connected driectly to the cpu and the others are not.. ive read that the ssd benefits from being on a slot connected directly to the cpu.
im guessing that there is no point running one titanx on 16x and the other on 8x, since they both need to load the same data, and therefore the faster connection will just wait for the slower one to catch up. also i understand that the difference for rendering is minimal between 8x and 16x.
any thoughts?
does the 670 for viewports benefit from the 16x slot? im not gaming, just max viewports.. if i wanted to add another pcie ssd, id have to shift one card of the above down to pcie2.0 1x slot.. would that bottleneck the 670 for viewports noticeably? i imagine massively.
alternatively i could ditch the 670 and use one of the titanx'es for viewports too.. its not like i can work when rendering anyway. does this impact available gpu ram for rendering? i guess so.
too many combinations, its hurting my head!
cheers!
Comment