AI/rocm-systems
2
0
Forka 0
Codice Problemi Pull Request Actions Pacchetti Progetti Rilasci Wiki Attività

SWDEV-541514 - Docs update 2025-09-15 (#993)

Sfoglia il codice sorgente 
Co-authored-by: Julia Jiang <56359287+jujiang-del@users.noreply.github.com>
This commit is contained in:
Istvan Kiss
2025-09-24 18:57:00 +02:00
committato da GitHub
parent 0575606e49
commit 83fb0c8c47
39 ha cambiato i file con 2973 aggiunte e 807 eliminazioni
Scarica il file Patch Scarica il file Diff Expand all files Collapse all files
projects/hip/docs/tutorial/saxpy.rst
+47 -47
Vedi File
@@ -348,89 +348,89 @@ find out what device binary flavors are embedded into the executable?
artifacts on disk. Add the ROCmCC installation folder to your PATH if you
want to use these utilities (the utilities expect them to be on the PATH).
You can list embedded program binaries using ``roc-obj-ls``.
You can list embedded program binaries using ``llvm-objdump`` with
``--offloading`` option.
.. code-block:: bash
roc-obj-ls ./saxpy
llvm-objdump --offloading ./saxpy
It should return something like:
.. code-block:: shell
1 host-x86_64-unknown-linux file://./saxpy#offset=12288&size=0
1 hipv4-amdgcn-amd-amdhsa--gfx803 file://./saxpy#offset=12288&size=9760
./saxpy: file format elf64-x86-64
Extracting offload bundle: ./saxpy.0.host-x86_64-unknown-linux-gnu-
Extracting offload bundle: ./saxpy.0.hipv4-amdgcn-amd-amdhsa--gfx942
The compiler embeds a version 4 code object (more on `code
object versions <https://www.llvm.org/docs/AMDGPUUsage.html#code-object-metadata>`_)
and used the LLVM target triple `amdgcn-amd-amdhsa--gfx803` (more on `target triples
and used the LLVM target triple ``amdgcn-amd-amdhsa--gfx942`` (more on `target triples
<https://www.llvm.org/docs/AMDGPUUsage.html#target-triples>`_). You can
extract that program object in a disassembled fashion for human consumption
via ``roc-obj``.
via ``llvm-objdump``.
.. code-block:: bash
roc-obj -t gfx803 -d ./saxpy
llvm-objdump --disassemble saxpy.0.hipv4-amdgcn-amd-amdhsa--gfx942 > saxpy.s
This creates two files on disk and ``.s`` extension is of most interest.
Opening this file or dumping it to the console using ``cat``
lets find the disassembled binary of the SAXPY compute kernel, something
similar to:
This creates a file on the disk called ``saxpy.s`` Opening this file or
dumping it to the console using ``cat`` lets find the disassembled binary of
the SAXPY compute kernel, something similar to:
.. code-block::
saxpy.0.hipv4-amdgcn-amd-amdhsa--gfx942: file format elf64-amdgpu
Disassembly of section .text:
<_Z12saxpy_kernelfPKfPfj>:
s_load_dword s0, s[4:5], 0x2c // 000000001000: C0020002 0000002C
s_load_dword s1, s[4:5], 0x18 // 000000001008: C0020042 00000018
s_waitcnt lgkmcnt(0) // 000000001010: BF8C007F
s_and_b32 s0, s0, 0xffff // 000000001014: 8600FF00 0000FFFF
s_mul_i32 s6, s6, s0 // 00000000101C: 92060006
v_add_u32_e32 v0, vcc, s6, v0 // 000000001020: 32000006
v_cmp_gt_u32_e32 vcc, s1, v0 // 000000001024: 7D980001
s_and_saveexec_b64 s[0:1], vcc // 000000001028: BE80206A
s_cbranch_execz 22 // 00000000102C: BF880016 <_Z12saxpy_kernelfPKfPfj+0x88>
s_load_dwordx4 s[0:3], s[4:5], 0x8 // 000000001030: C00A0002 00000008
v_mov_b32_e32 v1, 0 // 000000001038: 7E020280
v_lshlrev_b64 v[0:1], 2, v[0:1] // 00000000103C: D28F0000 00020082
s_waitcnt lgkmcnt(0) // 000000001044: BF8C007F
v_mov_b32_e32 v3, s1 // 000000001048: 7E060201
v_add_u32_e32 v2, vcc, s0, v0 // 00000000104C: 32040000
v_addc_u32_e32 v3, vcc, v3, v1, vcc // 000000001050: 38060303
flat_load_dword v2, v[2:3] // 000000001054: DC500000 02000002
v_mov_b32_e32 v3, s3 // 00000000105C: 7E060203
v_add_u32_e32 v0, vcc, s2, v0 // 000000001060: 32000002
v_addc_u32_e32 v1, vcc, v3, v1, vcc // 000000001064: 38020303
flat_load_dword v3, v[0:1] // 000000001068: DC500000 03000000
s_load_dword s0, s[4:5], 0x0 // 000000001070: C0020002 00000000
s_waitcnt vmcnt(0) lgkmcnt(0) // 000000001078: BF8C0070
v_mac_f32_e32 v3, s0, v2 // 00000000107C: 2C060400
flat_store_dword v[0:1], v3 // 000000001080: DC700000 00000300
s_endpgm // 000000001088: BF810000
0000000000001900 <_Z12saxpy_kernelfPKfPfj>:
s_load_dword s3, s[0:1], 0x2c // 000000001900: C00200C0 0000002C
s_load_dword s4, s[0:1], 0x18 // 000000001908: C0020100 00000018
s_waitcnt lgkmcnt(0) // 000000001910: BF8CC07F
s_and_b32 s3, s3, 0xffff // 000000001914: 8603FF03 0000FFFF
s_mul_i32 s2, s2, s3 // 00000000191C: 92020302
v_add_u32_e32 v0, s2, v0 // 000000001920: 68000002
v_cmp_gt_u32_e32 vcc, s4, v0 // 000000001924: 7D980004
s_and_saveexec_b64 s[2:3], vcc // 000000001928: BE82206A
s_cbranch_execz 20 // 00000000192C: BF880014 <_Z12saxpy_kernelfPKfPfj+0x80>
s_load_dwordx4 s[4:7], s[0:1], 0x8 // 000000001930: C00A0100 00000008
v_mov_b32_e32 v1, 0 // 000000001938: 7E020280
v_lshlrev_b64 v[0:1], 2, v[0:1] // 00000000193C: D28F0000 00020082
s_load_dword s0, s[0:1], 0x0 // 000000001944: C0020000 00000000
s_waitcnt lgkmcnt(0) // 00000000194C: BF8CC07F
v_lshl_add_u64 v[2:3], s[4:5], 0, v[0:1] // 000000001950: D2080002 04010004
v_lshl_add_u64 v[0:1], s[6:7], 0, v[0:1] // 000000001958: D2080000 04010006
global_load_dword v4, v[2:3], off // 000000001960: DC508000 047F0002
global_load_dword v5, v[0:1], off // 000000001968: DC508000 057F0000
s_waitcnt vmcnt(0) // 000000001970: BF8C0F70
v_fmac_f32_e32 v5, s0, v4 // 000000001974: 760A0800
global_store_dword v[0:1], v5, off // 000000001978: DC708000 007F0500
s_endpgm // 000000001980: BF810000
s_nop 0 // 000000001984: BF800000
Alternatively, call the compiler with ``--save-temps`` to dump all device
binary to disk in separate files.
.. code-block:: bash
amdclang++ ./HIP-Basic/saxpy/main.hip -o saxpy -I ./Common -lamdhip64 -L /opt/rocm/lib -O2 --save-temps
amdclang++ ./HIP-Basic/saxpy/main.hip -o saxpy -I ./Common -lamdhip64 -L /opt/rocm/lib -O2 --save-temps --offload-arch=gfx942
List all the temporaries created while compiling ``main.hip`` with:
.. code-block:: bash
ls main-hip-amdgcn-amd-amdhsa-*
main-hip-amdgcn-amd-amdhsa-gfx803.bc
main-hip-amdgcn-amd-amdhsa-gfx803.cui
main-hip-amdgcn-amd-amdhsa-gfx803.o
main-hip-amdgcn-amd-amdhsa-gfx803.out
main-hip-amdgcn-amd-amdhsa-gfx803.out.resolution.txt
main-hip-amdgcn-amd-amdhsa-gfx803.s
main-hip-amdgcn-amd-amdhsa-gfx942.bc
main-hip-amdgcn-amd-amdhsa-gfx942.o
main-hip-amdgcn-amd-amdhsa-gfx942.out.resolution.txt
main-hip-amdgcn-amd-amdhsa-gfx942.hipi
main-hip-amdgcn-amd-amdhsa-gfx942.out
main-hip-amdgcn-amd-amdhsa-gfx942.s
Files with the ``.s`` extension hold the disassembled contents of the binary.
The filename notes the graphics IPs used by the compiler. The contents of
this file are similar to what ``roc-obj`` printed to the console.
this file are similar to the `*.s` file created with ``llvm-objdump`` earlier.
.. tab-item:: Linux and NVIDIA
:sync: linux-nvidia
@@ -491,7 +491,7 @@ find out what device binary flavors are embedded into the executable?
We can see that the compiler embedded a version 4 code object (more on code
`object versions <https://www.llvm.org/docs/AMDGPUUsage.html#code-object-metadata>`_) and
used the LLVM target triple `amdgcn-amd-amdhsa--gfx906` (more on `target triples
used the LLVM target triple ``amdgcn-amd-amdhsa--gfx906`` (more on `target triples
<https://www.llvm.org/docs/AMDGPUUsage.html#target-triples>`_). Don't be
alarmed about linux showing up as a binary format, AMDGPU binaries uploaded to
the GPU for execution are proper linux ELF binaries in their format.