- Jan 19, 2013
-
-
Dmitri Gribenko authored
llvm-svn: 172917
-
Jakub Staszak authored
sufficient. llvm-svn: 172916
-
Chris Lattner authored
has past the point of making sense. Lets tidy things up: first step, moving a ton of big functions out of line. llvm-svn: 172904
-
Benjamin Kramer authored
This separates the check for "too few elements to run the vector loop" from the "memory overlap" check, giving a lot nicer code and allowing to skip the memory checks when we're not going to execute the vector code anyways. We still leave the decision of whether to emit the memory checks as branches or setccs, but it seems to be doing a good job. If ugly code pops up we may want to emit them as separate blocks too. Small speedup on MultiSource/Benchmarks/MallocBench/espresso. Most of this is legwork to allow multiple bypass blocks while updating PHIs, dominators and loop info. llvm-svn: 172902
-
Nadav Rotem authored
llvm-svn: 172894
-
Craig Topper authored
Use MVT instead of EVT when computing shuffle immediates since they can only be for legal types. Keeps compiler from generating unneeded checks and handling for extended types. llvm-svn: 172893
-
Chandler Carruth authored
includes. llvm-svn: 172891
-
Jack Carter authored
but I cannot reproduce the problem and have scrubed my sources and even tested with llvm-lit -v --vg. Formatting fixes. Mostly long lines and blank spaces at end of lines. Contributer: Jack Carter llvm-svn: 172882
-
Renato Golin authored
llvm-svn: 172877
-
Jakob Stoklund Olesen authored
llvm-svn: 172874
-
Andrew Kaylor authored
llvm-svn: 172872
-
David Greene authored
configure checks whether -Wno-missing-field-initializers is a valid compiler flag but it was never actually used in Makefile.rules. Enable it to avoid some ridiculous warnings from gcc. llvm-svn: 172870
-
Nadav Rotem authored
On Sandybridge loading unaligned 256bits using two XMM loads (vmovups and vinsertf128) is faster than using a single vmovups instruction. llvm-svn: 172868
-
Jordan Rose authored
llvm-svn: 172866
-
Michael Gottesman authored
llvm-svn: 172864
-
Michael Gottesman authored
llvm-svn: 172863
-
- Jan 18, 2013
-
-
Eric Christopher authored
llvm-svn: 172857
-
Bill Wendling authored
llvm-svn: 172854
-
Bill Wendling authored
Further encapsulation of the Attribute object. Don't allow direct access to the Attribute object as an aggregate. llvm-svn: 172853
-
Bill Wendling authored
Remove unused parameter. Also use the AttributeSet query methods instead of the Attribute query methods. llvm-svn: 172852
-
Jordan Rose authored
This is duplicated in a couple places in the codebase. Adopt this in APFloat. llvm-svn: 172851
-
Jack Carter authored
but I cannot reproduce the problem and have scrubed my sources and even tested with llvm-lit -v --vg. Support for Mips register information sections. Mips ELF object files have a section that is dedicated to register use info. Some of this information such as the assumed Global Pointer value is used by the linker in relocation resolution. The register info file is .reginfo in o32 and .MIPS.options in 64 and n32 abi files. This patch contains the changes needed to create the sections, but leaves the actual register accounting for a future patch. Contributer: Jack Carter llvm-svn: 172847
-
Tom Stellard authored
Some instructions like memory reads/writes are executed asynchronously, so we need to insert S_WAITCNT instructions to block before accessing their results. Previously we have just inserted S_WAITCNT instructions after each async instruction, this patch fixes this and adds a prober insertion pass. Patch by: Christian König Tested-by:
Michel Dänzer <michel.daenzer@amd.com> Reviewed-by:
Tom Stellard <thomas.stellard@amd.com> Signed-off-by:
Christian König <deathsimple@vodafone.de> llvm-svn: 172846
-
Tom Stellard authored
We shouldn't insert KILL optimization if we don't have a kill instruction at all. Patch by: Christian König Tested-by:
Michel Dänzer <michel.daenzer@amd.com> Reviewed-by:
Tom Stellard <thomas.stellard@amd.com> Signed-off-by:
Christian König <deathsimple@vodafone.de> llvm-svn: 172845
-
Bill Wendling authored
Because the Attribute class is going to stop representing a collection of attributes, limit the use of it as an aggregate in favor of using AttributeSet. This replaces some of the uses for querying the function attributes. llvm-svn: 172844
-
Jack Carter authored
but I cannot reproduce the problem and have scrubed my sources and even tested with llvm-lit -v --vg. Removal of redundant code and formatting fixes. Contributers: Jack Carter/Vladimir Medic llvm-svn: 172842
-
Daniel Dunbar authored
llvm-svn: 172841
-
Benjamin Kramer authored
llvm-svn: 172839
-
Daniel Dunbar authored
llvm-svn: 172838
-
Daniel Dunbar authored
- Also, fixup syntax errors in LangRef and missing newline in the MCAsmStreamer. llvm-svn: 172837
-
Dmitri Gribenko authored
We don't have DejaGNU tests now. llvm-svn: 172836
-
Eli Bendersky authored
grep is now only mentioned once in a sentence that explicitly says it's deprecated. For FileCheck, there's no reason to repeat part of the documentation that exists in CommandGuide/FileCheck. llvm-svn: 172835
-
Benjamin Kramer authored
llvm-svn: 172824
-
NAKAMURA Takumi authored
llvm/test/CodeGen/X86/Atomics-64.ll: Tweak for 2nd RUN not to overwrite %t. It sometimes causes spurious failure on lit win32. Feel free to prune or suppress each output. llvm-svn: 172823
-
Alexey Samsonov authored
llvm-svn: 172813
-
Will Dietz authored
llvm-svn: 172806
-
Manuel Klimek authored
Summary: This allows unit tests for components that use Support/Debug.h to print debug information from test runs by specifying -debug when running the test. CC: llvm-commits Differential Revision: http://llvm-reviews.chandlerc.com/D301 llvm-svn: 172801
-
Craig Topper authored
Calculate vector element size more directly for VINSERTF128/VEXTRACTF128 immediate handling. Also use MVT since this only called on legal types during pattern matching. llvm-svn: 172797
-
Craig Topper authored
llvm-svn: 172795
-
Dmitri Gribenko authored
llvm-svn: 172794
-