When I read A5000 manual, there is no information about the max performance of FP64.
I think A5000 utilizes tensorcores for calculating FP64.
Is there anyone who knows the max perf. of FP64 or how to calculate for A5000?
Best regards.
RTX A5000 is based on the GA102 chip architecture.
It does not. The Tensor core units in the GA102 chip architecture do not contain the FP64 path that is present in e.g. the Tensor core units in the A100 GPU. Therefore FP64 calculations (including Tensor core emulation) will be performed by the FP64 units in the SM. You can get some confirmation of this by reading the GA102 whitepaper, particularly the note under Figure 2.
Therefore I would direct you to information such as here which is based on the FP64 units in the SM(s). Note that is a peak theoretical number. Actual delivered performance will be lower; your mileage may vary.