nir: Add store_agx intrinsic

This works like store_global, but lets us optimize address arithmetic. Like
load_agx, it is formatted to match the hardware semantic. We don't make use of
any clever formats in this series, though.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20558>
This commit is contained in:
Alyssa Rosenzweig
2022-12-03 17:10:13 -05:00
committed by Marge Bot
parent f506dac18f
commit cc5ca8164d

View File

@@ -1616,7 +1616,7 @@ intrinsic("store_zs_agx", [1, 1, 1], indices=[BASE], flags=[])
intrinsic("block_image_store_agx", [1, 1], bit_sizes=[32, 16],
indices=[FORMAT, IMAGE_DIM], flags=[CAN_REORDER])
# Formatted loads. The format is the pipe_format in memory (see
# Formatted load/store. The format is the pipe_format in memory (see
# agx_internal_formats.h for the supported list). This accesses:
#
# address + extend(index) << (format shift + shift)
@@ -1624,13 +1624,16 @@ intrinsic("block_image_store_agx", [1, 1], bit_sizes=[32, 16],
# The nir_intrinsic_base() index encodes the shift. The sign_extend index
# determines whether sign- or zero-extension is used for the index.
#
# All loads on AGX uses these hardware instructions, so while these are
# logically load_global_agx (etc), the _global is omitted as it adds nothing.
# All loads and stores on AGX uses these hardware instructions, so while these are
# logically load_global_agx/load_global_constant_agx/store_global_agx, the
# _global is omitted as it adds nothing.
#
# src[] = { address, index }.
load("agx", [1, 1], [ACCESS, BASE, FORMAT, SIGN_EXTEND], [CAN_ELIMINATE])
load("constant_agx", [1, 1], [ACCESS, BASE, FORMAT, SIGN_EXTEND],
[CAN_ELIMINATE, CAN_REORDER])
# src[] = { value, address, index }.
store("agx", [1, 1], [ACCESS, BASE, FORMAT, SIGN_EXTEND])
# Logical complement of load_front_face, mapping to an AGX system value
system_value("back_face_agx", 1, bit_sizes=[1, 32])