caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
cf5d8def5cf66fbdfffa00a4845bd648ec58ed60
clang-p2996/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.ptr.tbuffer.load.d16.ll
Krzysztof Drewniak faa2c678aa [AMDGPU] Add buffer intrinsics that take resources as pointers 
In order to enable the LLVM frontend to better analyze buffer
operations (and to potentially enable more precise analyses on the
backend), define versions of the raw and structured buffer intrinsics
that use `ptr addrspace(8)` instead of `<4 x i32>` to represent their
rsrc arguments.

The new intrinsics are named by replacing `buffer.` with `buffer.ptr`.

One advantage to these intrinsic definitions is that, instead of
specifying that a buffer load/store will read/write some memory, we
can indicate that the memory read or written will be based on the
pointer argument. This means that, for example, a read from a
`noalias` buffer can be pulled out of a loop that is modifying a
distinct buffer.

In the future, we will define custom PseudoSourceValues that will
allow us to package up the (buffer, index, offset) triples that buffer
intrinsics contain and allow for more precise backend analysis.

This work also enables creating address space 7, which represents
manipulation of raw buffers using native LLVM load and store
instructions.

Where tests simply used a buffer intrinsic while testing some other
code path (such as the tests for VGPR spills), they have been updated
to use the new intrinsic form. Tests that are "about" buffer
intrinsics (for instance, those that ensure that they codegen as
expected) have been duplicated, either within existing files or into
new ones.

Depends on D145441

Reviewed By: arsenm, #amdgpu

Differential Revision: https://reviews.llvm.org/D147547
2023-06-05 16:59:07 +00:00

143 lines
8.0 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs -show-mc-encoding | FileCheck -enable-var-scope -check-prefixes=PREGFX10-UNPACKED %s
; RUN: llc < %s -march=amdgcn -mcpu=gfx810 -verify-machineinstrs | FileCheck -enable-var-scope -check-prefixes=PREGFX10-PACKED %s
; RUN: llc < %s -march=amdgcn -mcpu=gfx900 -verify-machineinstrs | FileCheck -enable-var-scope -check-prefixes=PREGFX10-PACKED %s
; RUN: llc < %s -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs | FileCheck -enable-var-scope -check-prefixes=GFX10-PACKED %s
; RUN: llc < %s -march=amdgcn -mcpu=gfx1100 -amdgpu-enable-delay-alu=0 -verify-machineinstrs | FileCheck -enable-var-scope -check-prefixes=GFX11-PACKED %s
define amdgpu_ps half @tbuffer_load_d16_x(ptr addrspace(8) inreg %rsrc) {
; PREGFX10-UNPACKED-LABEL: tbuffer_load_d16_x:
; PREGFX10-UNPACKED: ; %bb.0: ; %main_body
; PREGFX10-UNPACKED-NEXT: tbuffer_load_format_d16_x v0, off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] ; encoding: [0x00,0x00,0xb4,0xe8,0x00,0x00,0x00,0x80]
; PREGFX10-UNPACKED-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x0f,0x8c,0xbf]
; PREGFX10-UNPACKED-NEXT: ; return to shader part epilog
;
; PREGFX10-PACKED-LABEL: tbuffer_load_d16_x:
; PREGFX10-PACKED: ; %bb.0: ; %main_body
; PREGFX10-PACKED-NEXT: tbuffer_load_format_d16_x v0, off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM]
; PREGFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; PREGFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX10-PACKED-LABEL: tbuffer_load_d16_x:
; GFX10-PACKED: ; %bb.0: ; %main_body
; GFX10-PACKED-NEXT: tbuffer_load_format_d16_x v0, off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX11-PACKED-LABEL: tbuffer_load_d16_x:
; GFX11-PACKED: ; %bb.0: ; %main_body
; GFX11-PACKED-NEXT: tbuffer_load_d16_format_x v0, off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX11-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX11-PACKED-NEXT: ; return to shader part epilog
main_body:
%data = call half @llvm.amdgcn.raw.ptr.tbuffer.load.f16(ptr addrspace(8) %rsrc, i32 0, i32 0, i32 22, i32 0)
ret half %data
}
define amdgpu_ps half @tbuffer_load_d16_xy(ptr addrspace(8) inreg %rsrc) {
; PREGFX10-UNPACKED-LABEL: tbuffer_load_d16_xy:
; PREGFX10-UNPACKED: ; %bb.0: ; %main_body
; PREGFX10-UNPACKED-NEXT: tbuffer_load_format_d16_xy v[0:1], off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] ; encoding: [0x00,0x80,0xb4,0xe8,0x00,0x00,0x00,0x80]
; PREGFX10-UNPACKED-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x0f,0x8c,0xbf]
; PREGFX10-UNPACKED-NEXT: v_mov_b32_e32 v0, v1 ; encoding: [0x01,0x03,0x00,0x7e]
; PREGFX10-UNPACKED-NEXT: ; return to shader part epilog
;
; PREGFX10-PACKED-LABEL: tbuffer_load_d16_xy:
; PREGFX10-PACKED: ; %bb.0: ; %main_body
; PREGFX10-PACKED-NEXT: tbuffer_load_format_d16_xy v0, off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM]
; PREGFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; PREGFX10-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v0
; PREGFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX10-PACKED-LABEL: tbuffer_load_d16_xy:
; GFX10-PACKED: ; %bb.0: ; %main_body
; GFX10-PACKED-NEXT: tbuffer_load_format_d16_xy v0, off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX10-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v0
; GFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX11-PACKED-LABEL: tbuffer_load_d16_xy:
; GFX11-PACKED: ; %bb.0: ; %main_body
; GFX11-PACKED-NEXT: tbuffer_load_d16_format_xy v0, off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX11-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX11-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v0
; GFX11-PACKED-NEXT: ; return to shader part epilog
main_body:
%data = call <2 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v2f16(ptr addrspace(8) %rsrc, i32 0, i32 0, i32 22, i32 0)
%elt = extractelement <2 x half> %data, i32 1
ret half %elt
}
define amdgpu_ps half @tbuffer_load_d16_xyz(ptr addrspace(8) inreg %rsrc) {
; PREGFX10-UNPACKED-LABEL: tbuffer_load_d16_xyz:
; PREGFX10-UNPACKED: ; %bb.0: ; %main_body
; PREGFX10-UNPACKED-NEXT: tbuffer_load_format_d16_xyz v[0:2], off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] ; encoding: [0x00,0x00,0xb5,0xe8,0x00,0x00,0x00,0x80]
; PREGFX10-UNPACKED-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x0f,0x8c,0xbf]
; PREGFX10-UNPACKED-NEXT: v_mov_b32_e32 v0, v2 ; encoding: [0x02,0x03,0x00,0x7e]
; PREGFX10-UNPACKED-NEXT: ; return to shader part epilog
;
; PREGFX10-PACKED-LABEL: tbuffer_load_d16_xyz:
; PREGFX10-PACKED: ; %bb.0: ; %main_body
; PREGFX10-PACKED-NEXT: tbuffer_load_format_d16_xyz v[0:1], off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM]
; PREGFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; PREGFX10-PACKED-NEXT: v_mov_b32_e32 v0, v1
; PREGFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX10-PACKED-LABEL: tbuffer_load_d16_xyz:
; GFX10-PACKED: ; %bb.0: ; %main_body
; GFX10-PACKED-NEXT: tbuffer_load_format_d16_xyz v[0:1], off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX10-PACKED-NEXT: v_mov_b32_e32 v0, v1
; GFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX11-PACKED-LABEL: tbuffer_load_d16_xyz:
; GFX11-PACKED: ; %bb.0: ; %main_body
; GFX11-PACKED-NEXT: tbuffer_load_d16_format_xyz v[0:1], off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX11-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX11-PACKED-NEXT: v_mov_b32_e32 v0, v1
; GFX11-PACKED-NEXT: ; return to shader part epilog
main_body:
%data = call <3 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v3f16(ptr addrspace(8) %rsrc, i32 0, i32 0, i32 22, i32 0)
%elt = extractelement <3 x half> %data, i32 2
ret half %elt
}
define amdgpu_ps half @tbuffer_load_d16_xyzw(ptr addrspace(8) inreg %rsrc) {
; PREGFX10-UNPACKED-LABEL: tbuffer_load_d16_xyzw:
; PREGFX10-UNPACKED: ; %bb.0: ; %main_body
; PREGFX10-UNPACKED-NEXT: tbuffer_load_format_d16_xyzw v[0:3], off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] ; encoding: [0x00,0x80,0xb5,0xe8,0x00,0x00,0x00,0x80]
; PREGFX10-UNPACKED-NEXT: s_waitcnt vmcnt(0) ; encoding: [0x70,0x0f,0x8c,0xbf]
; PREGFX10-UNPACKED-NEXT: v_mov_b32_e32 v0, v3 ; encoding: [0x03,0x03,0x00,0x7e]
; PREGFX10-UNPACKED-NEXT: ; return to shader part epilog
;
; PREGFX10-PACKED-LABEL: tbuffer_load_d16_xyzw:
; PREGFX10-PACKED: ; %bb.0: ; %main_body
; PREGFX10-PACKED-NEXT: tbuffer_load_format_d16_xyzw v[0:1], off, s[0:3], 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM]
; PREGFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; PREGFX10-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v1
; PREGFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX10-PACKED-LABEL: tbuffer_load_d16_xyzw:
; GFX10-PACKED: ; %bb.0: ; %main_body
; GFX10-PACKED-NEXT: tbuffer_load_format_d16_xyzw v[0:1], off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX10-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX10-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v1
; GFX10-PACKED-NEXT: ; return to shader part epilog
;
; GFX11-PACKED-LABEL: tbuffer_load_d16_xyzw:
; GFX11-PACKED: ; %bb.0: ; %main_body
; GFX11-PACKED-NEXT: tbuffer_load_d16_format_xyzw v[0:1], off, s[0:3], 0 format:[BUF_FMT_32_FLOAT]
; GFX11-PACKED-NEXT: s_waitcnt vmcnt(0)
; GFX11-PACKED-NEXT: v_lshrrev_b32_e32 v0, 16, v1
; GFX11-PACKED-NEXT: ; return to shader part epilog
main_body:
%data = call <4 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v4f16(ptr addrspace(8) %rsrc, i32 0, i32 0, i32 22, i32 0)
%elt = extractelement <4 x half> %data, i32 3
ret half %elt
}
declare half @llvm.amdgcn.raw.ptr.tbuffer.load.f16(ptr addrspace(8), i32, i32, i32, i32)
declare <2 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v2f16(ptr addrspace(8), i32, i32, i32, i32)
declare <3 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v3f16(ptr addrspace(8), i32, i32, i32, i32)
declare <4 x half> @llvm.amdgcn.raw.ptr.tbuffer.load.v4f16(ptr addrspace(8), i32, i32, i32, i32)
Reference in New Issue View Git Blame Copy Permalink