diff options
author | Krunoslav Kovac <Krunoslav.Kovac@amd.com> | 2019-06-18 17:38:43 -0400 |
---|---|---|
committer | Alex Deucher <alexander.deucher@amd.com> | 2019-07-18 14:27:25 -0500 |
commit | e752058b8671c6c87e484cff144c5c6309a37253 (patch) | |
tree | aa7835b4ae24d597f31f33b5947324d109b1f4f5 /drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c | |
parent | c43f89f81cc0c54c7c460f1d6365445939069c83 (diff) | |
download | linux-e752058b8671c6c87e484cff144c5c6309a37253.tar.bz2 |
drm/amd/display: Optimize gamma calculations
[Why&How]
1. Stack usage is pretty high as fixed31_32 struct is 8 bytes and we
have functions with >30 vars on the stack.
2. Optimize gamma calculation by reducing number of calls to
dc_fixpt_pow Our X points are divided into 32 regions wth 16 pts each.
Each region is 2x the previous, meaning x[i] = 2*x[i-16] for i>=16.
Using (2x)^gamma = 2^gamma * x^gamma, we can recursively compute powers
of gamma, we just need first 16 pts to start it up. dc_fixpt_pow() is
expensive, it computes x^y by doing exp(y*logx) Exp is done by Taylor
series approximation, and log by Newton-like approximation that also
uses exp internally. In short, it's significantly heavier than
run-of-the-mill addition/subtraction/multiply.
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Diffstat (limited to 'drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c')
0 files changed, 0 insertions, 0 deletions