[llvm-commits] [llvm] r155797 - in /llvm/trunk: CREDITS.TXT lib/Target/CellSPU/README.txt

[Kalle Raiskila]

Author: kraiskil
Date: Sun Apr 29 15:00:55 2012
New Revision: 155797

URL: http://llvm.org/viewvc/llvm-project?rev=155797&view=rev
Log:
Update the documentation of CellSPU, in case it gets removed in 3.1.

Modified:
    llvm/trunk/CREDITS.TXT
    llvm/trunk/lib/Target/CellSPU/README.txt

Modified: llvm/trunk/CREDITS.TXT
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/CREDITS.TXT?rev=155797&r1=155796&r2=155797&view=diff
==============================================================================
--- llvm/trunk/CREDITS.TXT (original)
+++ llvm/trunk/CREDITS.TXT Sun Apr 29 15:00:55 2012
@@ -342,6 +342,10 @@
 E: ghost at cs.msu.su
 D: Made inst_iterator behave like a proper iterator, LowerConstantExprs pass
 
+N: Kalle Raiskila
+E: kalle.rasikila at nokia.com
+D: Some bugfixes to CellSPU
+
 N: Xerxes Ranby
 E: xerxes at zafena.se
 D: Cmake dependency chain and various bug fixes

Modified: llvm/trunk/lib/Target/CellSPU/README.txt
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/CellSPU/README.txt?rev=155797&r1=155796&r2=155797&view=diff
==============================================================================
--- llvm/trunk/lib/Target/CellSPU/README.txt (original)
+++ llvm/trunk/lib/Target/CellSPU/README.txt Sun Apr 29 15:00:55 2012
@@ -37,6 +37,20 @@
 ---------------------------------------------------------------------------
 
 TODO:
+* In commit r142152 vector legalization was set to element promotion per
+  default. This breaks half vectors (e.g. v2i32) badly as they get element
+  promoted to much slower types (v2i64).
+
+* Many CellSPU specific codegen tests only grep & count the number of 
+  instructions, not checking their place with FileCheck. There have also
+  been some commits that change the CellSPU checks, some of which might
+  have not been thoroughly scrutinized w.r.t. to the changes they cause in SPU
+  assembly. (especially since about the time of r142152)  
+
+* Some of the i64 math have huge tablegen rules, which sometime cause
+  tablegen to run out of memory. See e.g. bug 8850. i64 arithmetics 
+  should probably be done with libraries.
+
 * Create a machine pass for performing dual-pipeline scheduling specifically
   for CellSPU, and insert branch prediction instructions as needed.