diff options
| author | Samuel Pitoiset <samuel.pitoiset@gmail.com> | 2016-09-07 00:12:51 +0200 |
|---|---|---|
| committer | Samuel Pitoiset <samuel.pitoiset@gmail.com> | 2016-10-07 00:18:57 +0200 |
| commit | 11e75fffeb4afc5be0021477f11e5a18a6ff6abf (patch) | |
| tree | b48f069cb7e75320e72238436d89caa3985926ee /src/gallium/drivers/nouveau | |
| parent | 590734fa0d2c97035eae33278cce6065c98b880f (diff) | |
| download | external_mesa3d-11e75fffeb4afc5be0021477f11e5a18a6ff6abf.zip external_mesa3d-11e75fffeb4afc5be0021477f11e5a18a6ff6abf.tar.gz external_mesa3d-11e75fffeb4afc5be0021477f11e5a18a6ff6abf.tar.bz2 | |
nv50/ir: set number of threads/block for variable local size
When a variable local size is defined as specified by
ARB_compute_variable_group_size, the fixed local size is set to 0
and a SIGFPE occurs when we compute the maximum number of regs.
This allows to use 64 GPRs/thread.
v4: - use 512 threads on Fermi, 1024 on Kepler+
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Diffstat (limited to 'src/gallium/drivers/nouveau')
| -rw-r--r-- | src/gallium/drivers/nouveau/codegen/nv50_ir_target.h | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_target.h b/src/gallium/drivers/nouveau/codegen/nv50_ir_target.h index 4a701f7..eaf50cc 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_target.h +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_target.h @@ -175,6 +175,8 @@ public: virtual void parseDriverInfo(const struct nv50_ir_prog_info *info) { threads = info->prop.cp.numThreads; + if (threads == 0) + threads = info->target >= NVISA_GK104_CHIPSET ? 1024 : 512; } virtual bool runLegalizePass(Program *, CGStage stage) const = 0; |