Re: [Valgrind-users] cpu cycles

[Joerg B. <jo...@we...>]

Nicholas Nethercote wrote:
> On Mon, 28 Jul 2003, Joerg Beyer wrote:
> 
> 
>>I am aware that valgrind (with the calltree skin) counts
>>the processor instructions, used to run some code. Is there
>>also a way to count the CPU cycles?
> 
> 
> No.
> 
would it be good to have such a cycle counter?

>>valgrind is a cool tool to do optimizations on the source
>>(you could even do challenges: "you finds the fastest variant
>>of this piece of code ?"). Now somebody tells me, that counting
>>the instructions might be wrong, because not all instructions
>>take the same amount of cycles. I assume, that this is right,
>>but irrelevant from a practical point of view.
> 
> 
> On the contrary, it's extremely relevant from a practical point of view.
> It's almost impossible to know how long any one instruction will take.
> For example, on my machine, a memory read takes 1 cycle if it hits the L1
> cache, 10 cycles in the worst case if it only hits the L2 cache, and 206
> cycles in the worst case if it misses both caches.  But out-of-order
> execution and all the other fancy modern CPU mumbo-jumbo means that these
> delays are rarely as bad as the worst case.  Then throw in branch
> mispredictions, and other pipeline stalls... it's a mess.
> 
> N

OK, so an implementation to do cycle counting needs to have
a table that lists for every instruction how many cycles it
need for the different cache-hit/miss situations.

Are these informations available (for all the processors)?

	Joerg