Artem-B

--- bin/res.clang	2021-10-05 16:31:10.824553505 -0700
+++ bin/res.clang11	2021-10-05 16:56:46.800860833 -0700
@@ -11,39 +11,39 @@
  Common:
   GLOBAL:402 CONSTANT[3]:2844
  Function _ZN13cupostprocess14calcVortDeviceIL10KernelType3EL12BoundaryType4EEEv24neibs_interaction_paramsIXT0_ENSt11conditionalIXcvbeqT0_LS2_2EE18sa_boundary_params5emptyIS5_EE4typeEEP6float3:
-  REG:47 STACK:112 SHARED:0 LOCAL:0 CONSTANT[0]:424 CONSTANT[2]:16 TEXTURE:0 SURFACE:0 SAMPLER:0
+  REG:46 STACK:0 SHARED:0 LOCAL:0 CONSTANT[0]:424 CONSTANT[2]:16 TEXTURE:0 SURFACE:0 SAMPLER:0
  Function _ZN13cupostprocess20calcTestpointsDeviceIL10KernelType3EL12BoundaryType4E12FullViscSpecIL12RheologyType0EL15TurbulenceModel1EL26ComputationalViscosityType0EL12ViscousModel0EL15AverageOperator0ELm513ELb0EEEEvNS_17testpoints_paramsIXT0_ET1_XeqsrSB_9turbmodelLS5_3EE24neibs_interaction_paramsIXT0_ENSt11conditionalIXcvbeqT0_LS2_2EE18sa_boundary_params5emptyISE_EE4typeEENSD_IXcvbeqsrSB_9turbmodelLS5_3EE15keps_tex_paramsSF_ISK_EE4typeENSD_IXcvbeqsrSB_9turb

Artem-B

--- bin/res.clang	2021-10-05 16:31:10.824553505 -0700
+++ bin/res.nvcc	2021-10-05 16:31:00.712472386 -0700
@@ -9,219 +9,232 @@
 
 Resource usage:
  Common:
-  GLOBAL:402 CONSTANT[3]:2844
+  GLOBAL:0
+
+Fatbin elf code:

Artem-B

namespace {

template <int N>
struct __Tag;
# 54 "__clang_cuda_texture_intrinsics.h" 3
template <class>
struct __FT;

template <>

Artem-B

xxx

Artem-B

; ModuleID = 'repro.ll'
source_filename = "blaze-out/k8-opt/genfiles/third_party/amdgpu/rocFFT/kernels/kernels_273.cpp"
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa"

%struct.ham = type { %struct.zot }
%struct.zot = type { %struct.quux }
%struct.quux = type { <2 x double> }

$_ZZ22fft_fwd_ip_len100_sbccI15HIP_vector_typeIdLj2EEL9StrideBin0ELb1EEvPKT_S5_mPKmS7_mPS3_E3lds = comdat any

Artem-B

; ModuleID = '/usr/local/google/home/tra/work/llvm/repo/clang/test/CodeGenCUDA/unused-global-var.cu'
source_filename = "/usr/local/google/home/tra/work/llvm/repo/clang/test/CodeGenCUDA/unused-global-var.cu"
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa"

@v1 = dso_local local_unnamed_addr addrspace(1) externally_initialized global i32 0, align 4
@v2 = dso_local local_unnamed_addr addrspace(4) externally_initialized global i32 0, align 4
@v4.managed = dso_local addrspace(1) externally_initialized global i32 0, align 4
@u1 = dso_local local_unnamed_addr addrspace(1) externally_initialized global i32 0, align 4
@u2 = dso_local local_unnamed_addr addrspace(4) externally_initialized global i32 0, align 4

Artem-B

.

Artem-B

/* Search in terms of symbol name and section name */
bool Elf::get_symbol(const std::string& name, const std::string& section_name,
                     Elf64_Addr& value, Elf_Xword& size, unsigned char& bind,
                     unsigned char& type, Elf_Half& section_index,
                     unsigned char& other) const {
  bool ret = false;
  symbol_section_accessor symbol_reader(_elfio, _elfio.sections[_symtab_ndx]);
  // Opportunistic lookup, in case the name is unique.
  if (symbol_reader.get_symbol(name, value, size, bind, type, section_index,
                               other)) {

Artem-B

Notes go here.

Artem-B

; ModuleID = 'x2.bc'
source_filename = "merged_12-9d953f.hip.i.cc"
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
target triple = "amdgcn-amd-amdhsa"

%struct.wibble = type { <4 x float> }
%struct.wombat = type { float }

@global = external hidden addrspace(1) global %struct.wibble*, align 8