diff options
| author | Geoffrey Mainland <gmainlan@microsoft.com> | 2013-01-16 16:06:48 +0000 |
|---|---|---|
| committer | Geoffrey Mainland <gmainlan@microsoft.com> | 2013-02-01 22:00:24 +0000 |
| commit | b39e4de19a0aa3cbc42d72a02f83cec77f48f3a0 (patch) | |
| tree | d665f392023a04c2ffcaf8a609c77ec057141984 /compiler/llvmGen | |
| parent | cc56f34e71201e27e4e15af01a5a6258648d6a32 (diff) | |
| download | haskell-b39e4de19a0aa3cbc42d72a02f83cec77f48f3a0.tar.gz | |
Mimic OldCmm basic block ordering in the LLVM backend.
In OldCmm, the false case of a conditional was a fallthrough. In Cmm,
conditionals have both true and false successors. When we convert Cmm to LLVM,
we now first re-order Cmm blocks so that the false successor of a conditional
occurs next in the list of basic blocks, i.e., it is a fallthrough, just like it
(necessarily) did in OldCmm. Surprisingly, this can make a big performance
difference.
Diffstat (limited to 'compiler/llvmGen')
| -rw-r--r-- | compiler/llvmGen/LlvmCodeGen/CodeGen.hs | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/compiler/llvmGen/LlvmCodeGen/CodeGen.hs b/compiler/llvmGen/LlvmCodeGen/CodeGen.hs index 609be3d3b0..915981752e 100644 --- a/compiler/llvmGen/LlvmCodeGen/CodeGen.hs +++ b/compiler/llvmGen/LlvmCodeGen/CodeGen.hs @@ -40,7 +40,7 @@ type LlvmStatements = OrdList LlvmStatement -- genLlvmProc :: LlvmEnv -> RawCmmDecl -> UniqSM (LlvmEnv, [LlvmCmmDecl]) genLlvmProc env (CmmProc infos lbl live graph) = do - let blocks = toBlockListEntryFirst graph + let blocks = toBlockListEntryFirstFalseFallthrough graph (env', lmblocks, lmdata) <- basicBlocksCodeGen env live blocks ([], []) let info = mapLookup (g_entry graph) infos proc = CmmProc info lbl live (ListGraph lmblocks) |