[ofa-general] SRP/mlx4 interrupts throttling performance

[Cameron Harr]

Vladislav Bolkhovitin wrote:
>>> Were the IOPS, CS and IRQ rates results stable and consistent 
>>> between runs?
>>
>> I don't remember about the IOPs, but CS and IRQ rates were stable. 
>> However, they've also become stable in my later runs with 
>> scst_threads=[123].
>
> But with srpt_thread=0 - not, right?
I don't remember unfortunately. And during the running of my last tests, 
I had a lot of badness on the initiator that is making reporting and 
running things difficult right now (main drive has symptoms of being 
full and it doesn't show up in df and it shouldn't be close to full).