Use the new asmjit to fix //deeplearning/fbgemm (pytorch#1077)

Summary: Pull Request resolved: pytorch#1077 Update asmjit dependencies within fbgemm Reviewed By: jianyuh Differential Revision: D35292923 fbshipit-source-id: d854a7d7772105085dad62d5e773dbe88816d78a
backyes · Jul 8, 2022 · 64a5c4a · 64a5c4a
1 parent cd9dceb
commit 64a5c4a
Show file tree

Hide file tree

Showing 18 changed files with 72 additions and 56 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -182,6 +182,10 @@ if(NOT TARGET asmjit)
 
   add_subdirectory("${ASMJIT_SRC_DIR}" "${FBGEMM_BINARY_DIR}/asmjit")
   set_property(TARGET asmjit PROPERTY POSITION_INDEPENDENT_CODE ON)
+  # add a flag required for mac build
+  if(NOT MSVC)
+    target_compile_options(asmjit PRIVATE "-Wno-sign-conversion")
+  endif()
 endif()
 
 if(NOT TARGET cpuinfo)

diff --git a/fbgemm_gpu/CMakeLists.txt b/fbgemm_gpu/CMakeLists.txt
@@ -197,7 +197,7 @@ set_source_files_properties(
   ${gen_cpu_source_files}
   PROPERTIES
     INCLUDE_DIRECTORIES
-    "${CMAKE_CURRENT_SOURCE_DIR};${CMAKE_CURRENT_SOURCE_DIR}/include;${CMAKE_CURRENT_SOURCE_DIR}/../include"
+    "${CMAKE_CURRENT_SOURCE_DIR};${CMAKE_CURRENT_SOURCE_DIR}/include;${CMAKE_CURRENT_SOURCE_DIR}/../include;${THIRDPARTY}/asmjit/src"
 )
 
 set_source_files_properties(

diff --git a/include/fbgemm/Utils.h b/include/fbgemm/Utils.h
@@ -5,22 +5,26 @@
  * LICENSE file in the root directory of this source tree.
  */
 #pragma once
+
+#include "./FbgemmBuild.h"
+#include "./UtilsAvx2.h"
+
 #include <algorithm>
 #include <array>
 #include <cmath>
 #include <string>
 #include <type_traits>
-#include "./FbgemmBuild.h"
-#include "./UtilsAvx2.h"
 
-// forward declarations to asmjit
-namespace asmjit {
-namespace x86 {
-class Xmm;
-class Ymm;
-class Zmm;
-} // namespace x86
-} // namespace asmjit
+#include <asmjit/asmjit.h>
+
+// // forward declarations to asmjit
+// namespace asmjit {
+// namespace x86 {
+// class Xmm;
+// class Ymm;
+// class Zmm;
+// } // namespace x86
+// } // namespace asmjit
 
 namespace fbgemm {
 

diff --git a/src/EmbeddingSpMDM.cc b/src/EmbeddingSpMDM.cc
@@ -313,7 +313,7 @@ GenEmbeddingSpMDMLookup<
                   const float*, // weights
                   outType*, // out
                   const int32_t*, // compressed_indices_table and then mask
-                  const int*>(asmjit::CallConv::kIdHost),
+                  const int*>(asmjit::CallConvId::kHost),
               a->environment());
         } else {
           func.init(
@@ -327,7 +327,7 @@ GenEmbeddingSpMDMLookup<
                   const offsetType*, // offsets or lengths
                   const float*, // weights
                   outType*, // out and then mask
-                  const int*>(asmjit::CallConv::kIdHost),
+                  const int*>(asmjit::CallConvId::kHost),
               a->environment());
         }
 
@@ -336,20 +336,20 @@ GenEmbeddingSpMDMLookup<
 
         if (instSet == inst_set_t::avx2) {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
         } else {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
                   asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
                   asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
         }
 
         frame.setDirtyRegs(
-            x86::Reg::kGroupGp,
+            asmjit::RegGroup::kGp,
             reg_id == 15
                 ? asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15)
                 : asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14));

diff --git a/src/EmbeddingSpMDMNBit.cc b/src/EmbeddingSpMDMNBit.cc
@@ -299,7 +299,7 @@ GenEmbeddingSpMDMNBitLookup<
                   const float*, // weights
                   float*, // out
                   const int32_t* /* compressed_indices_table */,
-                  const int* /* mask */>(asmjit::CallConv::kIdHost),
+                  const int* /* mask */>(asmjit::CallConvId::kHost),
               a->environment());
         } else {
           func.init(
@@ -313,22 +313,22 @@ GenEmbeddingSpMDMNBitLookup<
                   const offsetType*, // offsets or lengths
                   const float*, // weights
                   float*, // out
-                  const int* /* mask */>(asmjit::CallConv::kIdHost),
+                  const int* /* mask */>(asmjit::CallConvId::kHost),
               a->environment());
         }
 
         asmjit::FuncFrame frame;
         frame.init(func);
 
         frame.setDirtyRegs(
-            x86::Reg::kGroupVec,
+            asmjit::RegGroup::kVec,
             asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                 asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
                 asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
                 asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
 
         frame.setDirtyRegs(
-            x86::Reg::kGroupGp,
+            asmjit::RegGroup::kGp,
             reg_id == 15
                 ? asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15)
                 : asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14));

diff --git a/src/FbgemmI64.cc b/src/FbgemmI64.cc
@@ -182,20 +182,20 @@ CodeGenBase<int64_t, int64_t, int64_t, int64_t>::getOrCreate(
             int64_t*,
             int64_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
     frame.init(func);
 
     frame.setDirtyRegs(
-        x86::Reg::kGroupVec,
+        asmjit::RegGroup::kVec,
         asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
             asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
             asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GenerateI8Depthwise.cc b/src/GenerateI8Depthwise.cc
@@ -271,18 +271,18 @@ GenI8Depthwise::jit_kernel_signature GenI8Depthwise::getOrCreate(
             int,
             const int*,
             int,
-            const std::int32_t*>(asmjit::CallConv::kIdHost),
+            const std::int32_t*>(asmjit::CallConvId::kHost),
         e->environment());
 
     asmjit::FuncFrame frame;
     frame.init(func);
 
     frame.setDirtyRegs(
-        x86::Reg::kGroupVec,
+        asmjit::RegGroup::kVec,
         asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GenerateKernelDirectConvU8S8S32ACC32.cc b/src/GenerateKernelDirectConvU8S8S32ACC32.cc
@@ -233,7 +233,7 @@ DirectConvCodeGenBase<uint8_t, int8_t, int32_t, int32_t>::getOrCreateDirectConv(
             int8_t*,
             int32_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
@@ -246,9 +246,9 @@ DirectConvCodeGenBase<uint8_t, int8_t, int32_t, int32_t>::getOrCreateDirectConv(
           asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31);
     }
 
-    frame.setDirtyRegs(x86::Reg::kGroupVec, dirtyVecRegs);
+    frame.setDirtyRegs(asmjit::RegGroup::kVec, dirtyVecRegs);
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);
@@ -660,7 +660,7 @@ DirectConvCodeGenBase<uint8_t, int8_t, int32_t, int32_t>::
             int,
             int,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
@@ -673,9 +673,9 @@ DirectConvCodeGenBase<uint8_t, int8_t, int32_t, int32_t>::
           asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31);
     }
 
-    frame.setDirtyRegs(x86::Reg::kGroupVec, dirtyVecRegs);
+    frame.setDirtyRegs(asmjit::RegGroup::kVec, dirtyVecRegs);
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GenerateKernelU8S8S32ACC16.cc b/src/GenerateKernelU8S8S32ACC16.cc
@@ -189,17 +189,18 @@ CodeGenBase<uint8_t, int8_t, int32_t, int16_t>::getOrCreate<inst_set_t::avx2>(
             int8_t*,
             int32_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
     frame.init(func);
     frame.setDirtyRegs(
-        x86::Reg::kGroupVec,
+        asmjit::RegGroup::kVec,
         asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp, asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14));
+        asmjit::RegGroup::kGp,
+        asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14));
 
     asmjit::FuncArgsAssignment args(&func);
     args.assignAll(buffer_A, buffer_B, B_pf, CBase, kSize, ldcReg);

diff --git a/src/GenerateKernelU8S8S32ACC16Avx512.cc b/src/GenerateKernelU8S8S32ACC16Avx512.cc
@@ -151,20 +151,20 @@ CodeGenBase<uint8_t, int8_t, int32_t, int16_t>::getOrCreate(
             int8_t*,
             int32_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
     frame.init(func);
 
     frame.setDirtyRegs(
-        x86::Reg::kGroupVec,
+        asmjit::RegGroup::kVec,
         asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
             asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
             asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GenerateKernelU8S8S32ACC32.cc b/src/GenerateKernelU8S8S32ACC32.cc
@@ -188,7 +188,7 @@ CodeGenBase<uint8_t, int8_t, int32_t, int32_t>::getOrCreate(
             int8_t*,
             int32_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
@@ -201,9 +201,9 @@ CodeGenBase<uint8_t, int8_t, int32_t, int32_t>::getOrCreate(
           asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31);
     }
 
-    frame.setDirtyRegs(x86::Reg::kGroupVec, dirtyVecRegs);
+    frame.setDirtyRegs(asmjit::RegGroup::kVec, dirtyVecRegs);
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GenerateKernelU8S8S32ACC32Avx512VNNI.cc b/src/GenerateKernelU8S8S32ACC32Avx512VNNI.cc
@@ -142,20 +142,20 @@ CodeGenBase<uint8_t, int8_t, int32_t, int32_t>::getOrCreate(
             int8_t*,
             int32_t*,
             int,
-            int>(asmjit::CallConv::kIdHost),
+            int>(asmjit::CallConvId::kHost),
         a->environment());
 
     asmjit::FuncFrame frame;
     frame.init(func);
 
     frame.setDirtyRegs(
-        x86::Reg::kGroupVec,
+        asmjit::RegGroup::kVec,
         asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
             asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
             asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
     frame.setDirtyRegs(
-        x86::Reg::kGroupGp,
+        asmjit::RegGroup::kGp,
         asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
     asmjit::FuncArgsAssignment args(&func);

diff --git a/src/GroupwiseConv.cc b/src/GroupwiseConv.cc
@@ -224,17 +224,17 @@ jit_conv_kernel_fp GenConvKernel<SPATIAL_DIM, INST_SET>::getOrCreate() {
           int32_t,
           int32_t,
           int32_t,
-          int32_t*>(asmjit::CallConv::kIdHost),
+          int32_t*>(asmjit::CallConvId::kHost),
       a->environment());
 
   frame_.init(func_);
 
   frame_.setDirtyRegs(
-      x86::Reg::kGroupVec,
+      asmjit::RegGroup::kVec,
       asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
           asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
   frame_.setDirtyRegs(
-      x86::Reg::kGroupGp,
+      asmjit::RegGroup::kGp,
       asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
   asmjit::FuncArgsAssignment args(&func_);

diff --git a/src/RowWiseSparseAdagradFused.cc b/src/RowWiseSparseAdagradFused.cc
@@ -177,28 +177,28 @@ typename ReturnFunctionSignature<indxType, offsetType, dataType>::
                 const int*, // lengths
                 float, // epsilon
                 float, // lr then rand_buffer
-                uint32_t*>(asmjit::CallConv::kIdHost),
+                uint32_t*>(asmjit::CallConvId::kHost),
             a->environment());
 
         asmjit::FuncFrame frame;
         frame.init(func);
 
         if (instSet == inst_set_t::avx2) {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
         } else {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
                   asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
                   asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
         }
 
         frame.setDirtyRegs(
-            x86::Reg::kGroupGp,
+            asmjit::RegGroup::kGp,
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14));
 
         asmjit::FuncArgsAssignment args(&func);

diff --git a/src/SparseAdagrad.cc b/src/SparseAdagrad.cc
@@ -510,28 +510,28 @@ GenSparseAdagrad<indxType, instSet>::getOrCreate(
                 const int*, // mask_avx2
                 float, // weight_decay
                 const double*, // counter then counter_halflife
-                std::int64_t>(asmjit::CallConv::kIdHost),
+                std::int64_t>(asmjit::CallConvId::kHost),
             a->environment());
 
         asmjit::FuncFrame frame;
         frame.init(func);
 
         if (instSet == inst_set_t::avx2) {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
         } else {
           frame.setDirtyRegs(
-              x86::Reg::kGroupVec,
+              asmjit::RegGroup::kVec,
               asmjit::Support::bitMask(0, 1, 2, 3, 4, 5, 6, 7) |
                   asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15) |
                   asmjit::Support::bitMask(16, 17, 18, 19, 20, 21, 22, 23) |
                   asmjit::Support::bitMask(24, 25, 26, 27, 28, 29, 30, 31));
         }
 
         frame.setDirtyRegs(
-            x86::Reg::kGroupGp,
+            asmjit::RegGroup::kGp,
             asmjit::Support::bitMask(8, 9, 10, 11, 12, 13, 14, 15));
 
         asmjit::FuncArgsAssignment args(&func);