| Date: | Mon, 23 Aug 2010 14:25:03 -0400 |
|---|---|
| From: | Edward Lee <edwl202@xxxxxxxxx> |
| Subject: | [Gems-users] coherence misses |
Hi, I am using the MOESI_SMP_directory protocol. In order to calculate the coherence misses I am looking at the ruby stats, namely the state transitions as suggested earlier. I am summing up all the FWD_GETX and FWD_GETS that cause state transitions, but I am not sure whether I should include transient states in this calculation. Are they referring to coherence misses? Such transient state transitions are: OM --> Fwd_GETX --> IM OI --> Fwd_GETX --> II MI --> Fwd_GETX --> II Regards, Ed |
| [← Prev in Thread] | Current Thread | [Next in Thread→] |
|---|---|---|
| ||
| Previous by Date: | [Gems-users] dynamic instruction of an exception, Miray Kas |
|---|---|
| Next by Date: | [Gems-users] Help with making checkpoint, George Bezerra |
| Previous by Thread: | Re: [Gems-users] Are there any design documents for Ruby?, Dan Gibson |
| Next by Thread: | [Gems-users] Average packet latency VS. Ruby cycles, fubinzhang |
| Indexes: | [Date] [Thread] |