[LLVMdev] [PATCH] parallel loop awareness to the LoopVectorizer
tobias at grosser.es
Mon Jan 28 04:58:20 PST 2013
On 01/28/2013 12:58 PM, Pekka Jääskeläinen wrote:
> Attached is a patch which uses a simple "parallel_loop" metadata attached
> to the loop branch instruction in the loop latch for skipping
> memory dependency checking in the LoopVectorizer. This was briefly
> in the email thread "LoopVectorizer in OpenCL C work group
Can you provide a test case?
> It also converts the "min iteration count to vectorize" to a parameter so
> this can be controlled from the command line.
This seems to be a separate patch?
More information about the llvm-dev