[Gems-users] avoiding coherent read misses count for barrier variable

[Muhammad abid Mughal <mabidm_pieas@xxxxxxxxx>]

hi,

     I am performing simulations to count read requests(Load) that incur a coherence miss but are not a spin on a contended lock or barrier variable. I do know that Sparc uses Atomic instructions (e.g  CASA) to implement locks so i did not count Atomic inst.Thats way i avoided coherent read requests that were a spin on a contended lock. How can i avoid counting read requests that incur a coherence miss but are a spin on a barrier variable?

Thanks in advance.

Muhammad abid

---

  New Email addresses available on Yahoo!  
Get the Email name you've always wanted on the new @ymail and @rocketmail.
Hurry before someone else does!