diff options
| author | Michel Dänzer <michel.daenzer@amd.com> | 2014-08-20 17:51:35 +0900 |
|---|---|---|
| committer | Michel Dänzer <michel@daenzer.net> | 2014-08-20 17:51:35 +0900 |
| commit | 7dab6b2fe9cf690b0dae1c127d0374322bd8278e (patch) | |
| tree | 0644f472b23a41ba7d44948a55146844218e838a | |
| parent | 1ff383360886c5cee1bb75abcc20675aca0336a6 (diff) | |
Revert "glamor: Set environment variable RADEON_THREAD=0"
This reverts commit 4b5060f357a3cb248c9359c92c1e9c42ef6434c8.
Further testing shows that disabling the thread is only a moderate win in
some cases, but a much bigger loss in some other cases.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
| -rw-r--r-- | src/radeon_glamor.c | 10 |
1 files changed, 0 insertions, 10 deletions
diff --git a/src/radeon_glamor.c b/src/radeon_glamor.c index 02fa6583..210ddcf2 100644 --- a/src/radeon_glamor.c +++ b/src/radeon_glamor.c @@ -132,16 +132,6 @@ radeon_glamor_pre_init(ScrnInfoPtr scrn) "Incompatible glamor version, required >= 0.3.0.\n"); return FALSE; } else { - /* This prevents the Gallium radeon winsys from spawning - * a separate thread for submitting command streams to - * the kernel. Improves scores of at least gtkperf and - * x11perf -putimage/-shmput/-getimage/-shmget, probably - * because glamor causes too many command stream flushes - * / too much synchronization with pending command - * streams for the separate thread to pay off. - */ - setenv("RADEON_THREAD", "0", 0); - if (glamor_egl_init(scrn, info->dri2.drm_fd)) { xf86DrvMsg(scrn->scrnIndex, X_INFO, "glamor detected, initialising EGL layer.\n"); |