From 6514751d12ba6e1c01ba8f14862c99ecc1fff343 Mon Sep 17 00:00:00 2001 From: Richard Sandiford Date: Fri, 9 Aug 2013 09:25:57 +0000 Subject: [PATCH] [SystemZ] Update README git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188062 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Target/SystemZ/README.txt | 20 +++++--------------- 1 file changed, 5 insertions(+), 15 deletions(-) diff --git a/lib/Target/SystemZ/README.txt b/lib/Target/SystemZ/README.txt index 2782b63a78d..563513b5f42 100644 --- a/lib/Target/SystemZ/README.txt +++ b/lib/Target/SystemZ/README.txt @@ -47,7 +47,7 @@ There is no scheduling support. -- -We don't use the BRANCH ON COUNT or BRANCH ON INDEX families of instruction. +We don't use the BRANCH ON INDEX instructions. -- @@ -56,18 +56,7 @@ and conditional returns. -- -We don't use the condition code results of anything except comparisons. - -Implementing this may need something more finely grained than the z_cmp -and z_ucmp that we have now. It might (or might not) also be useful to -have a mask of "don't care" values in conditional branches. For example, -integer comparisons never set CC to 3, so the bottom bit of the CC mask -isn't particularly relevant. JNLH and JE are equally good for testing -equality after an integer comparison, etc. - --- - -We don't use the LOAD AND TEST or TEST DATA CLASS instructions. +We don't use the TEST DATA CLASS instructions. -- @@ -77,9 +66,10 @@ condition codes. For example, we could use LCDFR instead of LCDBR. -- -We don't optimize block memory operations. +We don't optimize block memory operations, except using single MVCs +for memcpy. -It's definitely worth using things like MVC, CLC, NC, XC and OC with +It's definitely worth using things like CLC, NC, XC and OC with constant lengths. MVCIN may be worthwhile too. We should probably implement things like memcpy using MVC with EXECUTE.