| commit | 0d58938ed798882063737dedfe897a9b484295ad | |
| author | Alexander Monakov <amonakov@ispras.ru> | |
| Mon, 11 Jan 2016 12:55:31 +0000 (11 15:55 +0300) | ||
| committer | Alexander Monakov <amonakov@gcc.gnu.org> | |
| Mon, 11 Jan 2016 12:55:31 +0000 (11 15:55 +0300) | ||
| tree | 9e5efed6bfd5c03b33fb3b973ab51fdeae168032 | treesnapshot (tar.gz zip) |
| parent | b34b1c343600f7587213dafee5e305a8e4000132 | commitdiff |
nvptx plugin: do not force JIT target SM version
When link_ptx runs, a CUDA device is already bound to current thread, so the
driver library knows the target architecture. There isn't any benefit from
forcing a specific target here; on the contrary, hardcoding sm_30 breaks
offloading on later (Maxwell, sm_5x) devices.
* plugin/plugin-nvptx.c (link_ptx): Do not set CU_JIT_TARGET.
From-SVN: r232227
When link_ptx runs, a CUDA device is already bound to current thread, so the
driver library knows the target architecture. There isn't any benefit from
forcing a specific target here; on the contrary, hardcoding sm_30 breaks
offloading on later (Maxwell, sm_5x) devices.
* plugin/plugin-nvptx.c (link_ptx): Do not set CU_JIT_TARGET.
From-SVN: r232227
| libgomp/ChangeLog | diffblobblamehistory | |
| libgomp/plugin/plugin-nvptx.c | diffblobblamehistory |