Mesa (main): nir/algebraic: optimize Cyberpunk 2077's open-coded bitfieldReverse()

[GitLab Mirror]

Module: Mesa
Branch: main
Commit: 12294026d5104e386b3d9156b580e73b94a50a7c
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=12294026d5104e386b3d9156b580e73b94a50a7c

Author: Rhys Perry <pendingchaos02 at gmail.com>
Date:   Thu Nov  4 10:32:13 2021 +0000

nir/algebraic: optimize Cyberpunk 2077's open-coded bitfieldReverse()

fossil-db (Sienna Cichlid):
Totals from 9 (0.01% of 128647) affected shaders:
CodeSize: 29900 -> 28640 (-4.21%)
Instrs: 5677 -> 5443 (-4.12%)
Latency: 96561 -> 95025 (-1.59%)
Copies: 571 -> 544 (-4.73%)

Signed-off-by: Rhys Perry <pendingchaos02 at gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof at gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13673>

---

 src/compiler/nir/nir_opt_algebraic.py | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index c690b72a4e0..6e6c1ed1356 100644
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -2195,7 +2195,7 @@ optimizations += [
 ]
 
 # Unreal Engine 4 demo applications open-codes bitfieldReverse()
-def bitfield_reverse(u):
+def bitfield_reverse_ue4(u):
     step1 = ('ior', ('ishl', u, 16), ('ushr', u, 16))
     step2 = ('ior', ('ishl', ('iand', step1, 0x00ff00ff), 8), ('ushr', ('iand', step1, 0xff00ff00), 8))
     step3 = ('ior', ('ishl', ('iand', step2, 0x0f0f0f0f), 4), ('ushr', ('iand', step2, 0xf0f0f0f0), 4))
@@ -2204,7 +2204,18 @@ def bitfield_reverse(u):
 
     return step5
 
-optimizations += [(bitfield_reverse('x at 32'), ('bitfield_reverse', 'x'), '!options->lower_bitfield_reverse')]
+# Cyberpunk 2077 open-codes bitfieldReverse()
+def bitfield_reverse_cp2077(u):
+    step1 = ('ior', ('ishl', u, 16), ('ushr', u, 16))
+    step2 = ('ior', ('iand', ('ishl', step1, 1), 0xaaaaaaaa), ('iand', ('ushr', step1, 1), 0x55555555))
+    step3 = ('ior', ('iand', ('ishl', step2, 2), 0xcccccccc), ('iand', ('ushr', step2, 2), 0x33333333))
+    step4 = ('ior', ('iand', ('ishl', step3, 4), 0xf0f0f0f0), ('iand', ('ushr', step3, 4), 0x0f0f0f0f))
+    step5 = ('ior(many-comm-expr)', ('iand', ('ishl', step4, 8), 0xff00ff00), ('iand', ('ushr', step4, 8), 0x00ff00ff))
+
+    return step5
+
+optimizations += [(bitfield_reverse_ue4('x at 32'), ('bitfield_reverse', 'x'), '!options->lower_bitfield_reverse')]
+optimizations += [(bitfield_reverse_cp2077('x at 32'), ('bitfield_reverse', 'x'), '!options->lower_bitfield_reverse')]
 
 # "all_equal(eq(a, b), vec(~0))" is the same as "all_equal(a, b)"
 # "any_nequal(neq(a, b), vec(0))" is the same as "any_nequal(a, b)"