Rework %cmpi/u, %cmp/u and %ix/get for speed
commit07ae300e0cb79b32fed4b7da51e2bd919346ea74
authorStephen Williams <steve@icarus.com>
Fri, 23 May 2008 01:19:40 +0000 (22 18:19 -0700)
committerStephen Williams <steve@icarus.com>
Fri, 23 May 2008 01:19:40 +0000 (22 18:19 -0700)
treebd8a85c55257d7cc730db2e6d6b837c3730fc0bd
parent4c5f24c7a732fff41dc41cf646efaa9fefb34ea7
Rework %cmpi/u, %cmp/u and %ix/get for speed

These instructions can take advantage of the much optimized
vector_to_array function to do their arithmetic work quickly and
punt on X very quickly if needed. This helps some benchmarks.
vvp/vthread.cc
vvp/vvp_net.cc
vvp/vvp_net.h