projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 750160e) | patch
[GPGPU] Support PHI nodes used in GPU kernel
authorTobias Grosser <tobias@grosser.es>
Tue, 9 Aug 2016 15:35:06 +0000 (15:35 +0000)
committerTobias Grosser <tobias@grosser.es>
Tue, 9 Aug 2016 15:35:06 +0000 (15:35 +0000)
commitb06ff4574e22bf997aa741bc09fc143c265b1816
treea6ee4fa01aaf6cbc134455f2c75c87b565091f50tree | snapshot
parent750160e2605af4212727e254f863bbbbe0353174commit | diff
[GPGPU] Support PHI nodes used in GPU kernel

Ensure the right scalar allocations are used as the host location of data
transfers. For the device code, we clear the allocation cache before device
code generation to be able to generate new device-specific allocation and
we need to make sure to add back the old host allocations as soon as the
device code generation is finished.

llvm-svn: 278126
polly/lib/CodeGen/PPCGCodeGeneration.cpp diff | blob | history
polly/test/GPGPU/phi-nodes-in-kernel.ll [new file with mode: 0644] blob
Domain: System / Toolchain;