I tried running a bunch of applications (single-core/single process). For some reason, no memcopy op is detected in the decoder. Any idea what's going on? Are we really skipping mem to mem transfer of any kind while decoding each bbl?
After looking at instr for an array copy of different kinds, I see that it is always done from reg to mem (and vice versa). I was under the wrong impression that it will have a mem-to-mem transfer.
I tried running a bunch of applications (single-core/single process). For some reason, no memcopy op is detected in the decoder. Any idea what's going on? Are we really skipping mem to mem transfer of any kind while decoding each bbl?