Hello,
I am currently looking at using Blue Iris for 8-12 cameras with Deepstack offloaded to the GPU.
Is there a direct relationship between the number of cuda cores and performance, or does architecture also play a huge role for this specific use case?
For example, an old Tesla P4 card has 2560 cores and can be found on ebay for 300$, whereas a P400 has 256 cores and can be found for 100$.
Would the P4 offer ~10x the performance of a P400? Would that performance increase be worth it for such a system?
Thanks
I am currently looking at using Blue Iris for 8-12 cameras with Deepstack offloaded to the GPU.
Is there a direct relationship between the number of cuda cores and performance, or does architecture also play a huge role for this specific use case?
For example, an old Tesla P4 card has 2560 cores and can be found on ebay for 300$, whereas a P400 has 256 cores and can be found for 100$.
Would the P4 offer ~10x the performance of a P400? Would that performance increase be worth it for such a system?
Thanks