----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.gem5.org/r/3099/#review7313 -----------------------------------------------------------
Hi Nilay, I could be wrong but it looks to me that this patch is more like a "buffering of requests in the sequencer" and not really coalescing of requests. The reason being this patch actually issues the aliased requests to the memory subsystem on a hit callback whereas I was expecting coalescing to reduce the number of requests to the memory subsystem by issuing only a single request to the cacheline. Did I miss something? src/mem/ruby/system/Sequencer.cc (line 525) <http://reviews.gem5.org/r/3099/#comment6193> From a readability perspective, would it be better if this one large condition is broken down into smaller conditions with some comments? - Sooraj Puthoor On Sept. 5, 2015, 12:20 a.m., Nilay Vaish wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.gem5.org/r/3099/ > ----------------------------------------------------------- > > (Updated Sept. 5, 2015, 12:20 a.m.) > > > Review request for Default. > > > Repository: gem5 > > > Description > ------- > > Changeset 11106:04307eea6681 > --------------------------- > ruby: sequencer: coalesce requests > > This patch coalesces requests to that are for an address which already > has a pending request. The requests coming in later are issued after the > original request has completed. > > > Diffs > ----- > > src/mem/ruby/system/Sequencer.hh 95eb22928d0f > src/mem/ruby/system/Sequencer.cc 95eb22928d0f > src/mem/ruby/system/Sequencer.py 95eb22928d0f > > Diff: http://reviews.gem5.org/r/3099/diff/ > > > Testing > ------- > > > Thanks, > > Nilay Vaish > > _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
