diff options
| author | Tor Aamodt <[email protected]> | 2010-10-21 07:16:49 -0800 |
|---|---|---|
| committer | Tor Aamodt <[email protected]> | 2010-10-21 07:16:49 -0800 |
| commit | dc93f319051a9a9936a02cd9c1f7843a382a2da0 (patch) | |
| tree | 6c042ccab67be43b8fe442ab435ffbfd0f34e56e /configs | |
| parent | ee5ea34857e4ecc6c63d4971e549076c6a9888ba (diff) | |
1. rewriting memory access generation code (from scratch), why not...
passing CUDA 3.1 and ptxplus correlation, but correlation still bad (0.62)...
after debugging 1 to get it working with ptxplus, problem is very clear:
shared and constant cache accesses not occuring for operations that combine these with ALU operations.
TODO:
have a "read-operands" stage, which somehow combines operand collector
register reading with shared and const memory accesses...
[git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 7895]
Diffstat (limited to 'configs')
| -rw-r--r-- | configs/QuadroFX5800/gpgpusim.config | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/configs/QuadroFX5800/gpgpusim.config b/configs/QuadroFX5800/gpgpusim.config index 6f1c936..3d7d218 100644 --- a/configs/QuadroFX5800/gpgpusim.config +++ b/configs/QuadroFX5800/gpgpusim.config @@ -22,7 +22,7 @@ -gpgpu_const_cache:l1 64:64:2:L:R:f,A:2:32,4 -gpgpu_cache:dl2 64:32:8:L:R:m,A:16:4,4 --gpgpu_shmem_pipe_speedup 2 +-gpgpu_shmem_warp_parts 2 -gpgpu_shmem_port_per_bank 2 # interconnection |