Mesa (main): pan/va: Rename RSCALE to LDEXP
GitLab Mirror
gitlab-mirror at kemper.freedesktop.org
Tue Jun 21 22:55:23 UTC 2022
Module: Mesa
Branch: main
Commit: 21bedd2c97194d3cf8281f99cc9bd337e1510816
URL: http://cgit.freedesktop.org/mesa/mesa/commit/?id=21bedd2c97194d3cf8281f99cc9bd337e1510816
Author: Alyssa Rosenzweig <alyssa at collabora.com>
Date: Fri Jun 17 11:34:51 2022 -0400
pan/va: Rename RSCALE to LDEXP
This avoids needless variation from Bifrost. While at it, fix the opcode
definition: there are no abs/neg/swizzle modifiers on the signed integer source,
and there's no clamp. However, there are round and infinity modes, like on
Bifrost.
Signed-off-by: Alyssa Rosenzweig <alyssa at collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17101>
---
src/panfrost/bifrost/valhall/ISA.xml | 19 ++++++++++---------
1 file changed, 10 insertions(+), 9 deletions(-)
diff --git a/src/panfrost/bifrost/valhall/ISA.xml b/src/panfrost/bifrost/valhall/ISA.xml
index 71677fe6ce1..a5e35f9f572 100644
--- a/src/panfrost/bifrost/valhall/ISA.xml
+++ b/src/panfrost/bifrost/valhall/ISA.xml
@@ -1763,18 +1763,19 @@
<src absneg="true">B</src>
</group>
- <group name="FRSCALE" title="Floating-point rescaling" dests="1" opcode2="6" unit="FMA">
- <ins name="FRSCALE.f32" opcode="0xA4"/>
- <ins name="FRSCALE.v2f16" opcode="0xA5"/>
+ <group name="LDEXP" title="Floating-point rescaling" dests="1" opcode2="6" unit="FMA">
+ <ins name="LDEXP.f32" opcode="0xA4"/>
+ <ins name="LDEXP.v2f16" opcode="0xA5"/>
<desc>
Computes $A \cdot 2^B$ by adding B to the exponent of A. Used to calculate
various special functions, particularly base-2 exponents. Special case
handling differs from an actual floating-point multiply, so this should
not be used outside fixed instruction sequences.
</desc>
- <clamp/>
<src absneg="true" swizzle="true">A</src>
- <src absneg="true" swizzle="true">B</src>
+ <src/>
+ <roundmode/> <!-- Also has rtna -->
+ <!-- Also has infinity handling for arctan -->
</group>
<ins name="FEXP.f32" title="Floating-point exponent" dests="1" opcode="0xA4" opcode2="8" unit="SFU">
@@ -2648,7 +2649,7 @@
special transcendental function sequences. It should not be used for
general code as its special case handling differs from two back-to-back
`FMA.f32` operations. Equivalent to `FMA.f32` back-to-back with
- `RSCALE.f32`
+ `LDEXP.f32`
</desc>
<clamp/>
<src absneg="true">A</src>
@@ -2664,7 +2665,7 @@
ordinary multiply would return NaN. Used in special transcendental
function sequences. It should not be used for general code as its special
case handling differs from two back-to-back `FMA.f32` operations.
- Equivalent to `FMA.f32` back-to-back with `RSCALE.f32`
+ Equivalent to `FMA.f32` back-to-back with `LDEXP.f32`
</desc>
<clamp/>
<src absneg="true">A</src>
@@ -2680,7 +2681,7 @@
ordinary multiply would return NaN. Used in special transcendental
function sequences. It should not be used for general code as its special
case handling differs from two back-to-back `FMA.f32` operations.
- Equivalent to `FMA.f32` back-to-back with `RSCALE.f32`
+ Equivalent to `FMA.f32` back-to-back with `LDEXP.f32`
</desc>
<clamp/>
<src absneg="true">A</src>
@@ -2695,7 +2696,7 @@
interpreted as a 16-bit value. Used in special transcendental function
sequences. It should not be used for general code as its special case
handling differs from two back-to-back `FMA.f32` operations. Equivalent
- to `FMA.f32` back-to-back with `RSCALE.f32`
+ to `FMA.f32` back-to-back with `LDEXP.f32`
</desc>
<clamp/>
<src absneg="true">A</src>
More information about the mesa-commit
mailing list