llvm.org GIT mirror llvm / cb5868c
AMDGPU/Docs: Fix unreadable characters git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@316171 91177308-0d34-0410-b5e6-96231b3b80d8 Konstantin Zhuravlyov 1 year, 10 months ago
1 changed file(s) with 6 addition(s) and 6 deletion(s). Raw diff Collapse all Expand all
522522
523523 The AMDGPU backend code object uses the following ELF note records in the
524524 ``.note`` section. The *Description* column specifies the layout of the note
525 records ``desc`` field. All fields are consecutive bytes. Note records with
525 record's ``desc`` field. All fields are consecutive bytes. Note records with
526526 variable size strings have a corresponding ``*_size`` field that specifies the
527527 number of bytes, including the terminating null character, in the string. The
528528 string(s) come immediately after the preceding fields.
672672
673673 **G**
674674 Represents the offset into the global offset table at which the relocation
675 entrys symbol will reside during execution.
675 entry's symbol will reside during execution.
676676
677677 **GOT**
678678 Represents the address of the global offset table.
12741274 for a memory region with the kernarg property for the kernel agent that will
12751275 execute the kernel. It must be at least 16 byte aligned.
12761276 4. Kernel argument values are assigned to the kernel argument memory
1277 allocation. The layout is defined in the *HSA Programmers Language Reference*
1277 allocation. The layout is defined in the *HSA Programmer's Language Reference*
12781278 [HSA]_. For AMDGPU the kernel execution directly accesses the kernel argument
12791279 memory in the same way constant memory is accessed. (Note that the HSA
12801280 specification allows an implementation to copy the kernel argument contents to
20052005
20062006 The second SGPR is 32 bit
20072007 byte size of a single
2008 work-items scratch memory
2008 work-item's scratch memory
20092009 usage. CP obtains this from
20102010 the runtime, and it is
20112011 always a multiple of DWORD.
21052105 then Work-Group Id Z 1 32 bit work-group id in Z
21062106 (enable_sgpr_workgroup_id dimension of grid for
21072107 _Z) wavefront.
2108 then Work-Group Info 1 {first_wave, 14b0000,
2108 then Work-Group Info 1 {first_wave, 14'b0000,
21092109 (enable_sgpr_workgroup ordered_append_term[10:0],
21102110 _info) threadgroup_size_in_waves[5:0]}
21112111 then Scratch Wave Offset 1 32 bit byte offset from base
40624062 .. [HSA] `Heterogeneous System Architecture (HSA) Foundation `__
40634063 .. [ELF] `Executable and Linkable Format (ELF) `__
40644064 .. [DWARF] `DWARF Debugging Information Format `__
4065 .. [YAML] `YAML Aint Markup Language (YAML™) Version 1.2 `__
4065 .. [YAML] `YAML Ain't Markup Language (YAML™) Version 1.2 `__
40664066 .. [OpenCL] `The OpenCL Specification Version 2.0 `__
40674067 .. [HRF] `Heterogeneous-race-free Memory Models `__
40684068 .. [AMD-AMDGPU-Compute-Application-Binary-Interface] `AMDGPU Compute Application Binary Interface `__