On Fri, 8 Jun 2012, Ali chaker wrote:
if I've understood, the cache miss latency is calculated during the
simulation.I've found the following formula for misslatemcy in
cache_impl.hh file:
*completion_time = tags->getHitLatency() +(transfer_offset ?
pkt->finishTime : pkt->firstWordTime);*
*missLatency[target->pkt->cmdToIndex()][target->pkt->req->masterId()]
+= completion_time - target->recvTime;*
I don't understand how the cache miss latency is calculated ? the latency
cache parameter is for cache hit latency or cache miss latency?
Why should the cache miss latency be a fixed constant? Suppose there is a
shared interconnect between two levels of caches. If multiple transactions
can go on concurrently, it may not be possible to assign a fixed latency
on how quickly the next level would respond.
The latency parameter associated with the cache, is the time required for
accessing the cache. It assumed to be the same whether or not the address
being accessed is in the cache.
Miss latency is being computed as the (time when request was
completed) - (time when the request was received).
--
Nilay
_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
