Date: | Fri, 18 Jun 2010 08:42:55 -0700 (PDT) |
---|---|
From: | Muhammad abid Mughal <mabidm_pieas@xxxxxxxxx> |
Subject: | [Gems-users] MLP-Memory level parallelism |
Hi guys, Hope you all doing good I ran couple of simulations with Opal+Ruby to measure MLP. But in each case Avg. MLP was so high (multiple of hundred and thousand). I am wondering the way Opal calculate MLP (in rubycache_t class) is not correct. Here is the description how Opal calculates MLP : for each miss it calculates the difference of cycles = current miss time - m_last miss time localsum = difference of cycles *
m_tot_miss_outstanding m_totalmlpsum += localsum variables start with m_ are data members of rubycache_t class Finally it divides m_totalmlpsum/total_local_cycles_core to calculate Avg. MLP. Any comments? Muhammad abid
|
[← Prev in Thread] | Current Thread | [Next in Thread→] |
---|---|---|
|
Previous by Date: | [Gems-users] Retire Not-Ready Reasons, mihai pricopi |
---|---|
Next by Date: | [Gems-users] Ruby crashing at a LDDA instruction in Sunfire target, Miguel Ojeda |
Previous by Thread: | [Gems-users] Lack of forward progress detected, Ying Zhang |
Next by Thread: | [Gems-users] prefetch in opal, Jerry Lin |
Indexes: | [Date] [Thread] |