[Enhancement][Bugfix] Fix bug in warp specialized pass and add gemm_sr fallback support for Hopper (#712)

* bug fix and support gemm_sr fallback for hopper

* Update gemm.cc

---------

Co-authored-by: Lei Wang <34334180+LeiWang1999@users.noreply.github.com>
Co-authored-by: LeiWang1999 <leiwang1999@outlook.com>

Author: xwhzz

src/op/gemm.cc

@@ -241,6 +241,10 @@ std::pair<int, int> Gemm::ComputeWarpPartition(int block_size,
 }
 
 bool Gemm::CheckWGMMA() const {
+  if (B.scope() != "shared.dyn" && B.scope() != "shared") {
+    return false;
+  }
+
   if (C->dtype == DataType::Float(16)) {
     if (A->dtype == DataType::Float(16) && B->dtype == DataType::Float(16))
       return K % 16 == 0;
@@ -443,7 +447,9 @@ LayoutMap Gemm::InferLayout(const LayoutInferArgs &T, InferLevel level) {
                                  B->dtype.bits(), trans_B ? 2 : 1);
       results.Set(B, ABLayout);
     } else {
-      ICHECK(0) << "WGMMA only support B in shared.";
+      auto fragment =
+          makeGemmFragmentB(M, N, K, M / warp_m, N / warp_n, trans_B);
+      results.Set(B, fragment->BindThreadRange(thread_range));
     }
   } else if (TargetIsCDNA(T.target)) {
     auto fragment =
@@ -490,4 +496,4 @@ TIR_REGISTER_TL_OP(Gemm, gemm)
                                Integer(CallEffectKind::kOpaque));
 
 } // namespace tl
-} // namespace tvm
+} // namespace tvm

src/tl_templates/cuda/gemm_sm90.h

@@ -624,6 +624,19 @@ TL_DEVICE void gemm_rs(A_type *pA, B_type *pB, C_type *accum) {
   }
 }
 
+template <int M, int N, int K, int num_warp_m, int num_warp_n, bool trans_A,
+          bool trans_B, bool clear_accum = false, int lda = 0, int ldb = 0,
+          int offset_a = 0, int offset_b = 0, bool use_wgmma = true,
+          int wg_wait = 0, typename A_type, typename B_type, typename C_type>
+TL_DEVICE void gemm_sr(A_type *pA, B_type *pB, C_type *accum) {
+  static_assert(!use_wgmma, "wgmma doesn't support gemm_sr");
+  using MMA =
+      cute::tl_mma::GemmTensorOp<M, N, K, num_warp_m, num_warp_n, trans_A,
+                                 trans_B, clear_accum, lda, ldb, offset_a,
+                                 offset_b, A_type, B_type, C_type>;
+  MMA::body_sr(pA, pB, accum);
+}
+
 template <int num_mma> TL_DEVICE void wait_wgmma() {
   cute::warpgroup_wait<num_mma>();
 }

src/transform/warp_specialized_rewriter.cc

@@ -572,12 +572,11 @@ class WSCodeEmitter : public StmtMutator {
   WSCodeEmitter(bool is_emitting_producer, IterVar thread_iv,
                 Map<Var, Buffer> buffer_data_to_buffer,
                 const WarpSpecializedRoleMarker &marker,
-                bool mbarrier_only = false)
+                bool mbarrier_only = false, bool only_has_wgmma = false)
       : is_emitting_producer_(is_emitting_producer),
         buffer_data_to_buffer_(buffer_data_to_buffer), marker_(marker),
-        thread_var_(thread_iv->var), mbarrier_only_(mbarrier_only) {}
-
-  bool onlyHasWgMMA() const { return only_has_wgmma_; }
+        thread_var_(thread_iv->var), mbarrier_only_(mbarrier_only),
+        only_has_wgmma_(only_has_wgmma) {}
 
   bool hasSimtCopy() const { return has_simt_copy_; }
 
@@ -617,8 +616,6 @@ class WSCodeEmitter : public StmtMutator {
 
     auto map = ExtractSyncPattern(op->seq);
 
-    only_has_wgmma_ = WgMMACollector::HasWgMMA(SeqStmt(op->seq));
-
     /*
       std::cout << "Print ExtractSyncPattern" << std::endl;
       for (int i = 0; i < static_cast<int>(op->seq.size()); i++) {
@@ -1212,11 +1209,12 @@ class WarpSpecializedRewriter : public StmtExprMutator {
       block_realize.CopyOnWrite()->block = block;
       return block_realize;
     }
+    only_has_wgmma_ = WgMMACollector::HasWgMMA(block->body);
     WSCodeEmitter producer(true, thread_iv_, buffer_data_to_buffer_, marker);
-    WSCodeEmitter consumer(false, thread_iv_, buffer_data_to_buffer_, marker);
+    WSCodeEmitter consumer(false, thread_iv_, buffer_data_to_buffer_, marker,
+                           false, only_has_wgmma_);
     Stmt producer_code = producer(block->body);
     Stmt consumer_code = consumer(block->body);
-    bool only_has_wgmma = consumer.onlyHasWgMMA();
     PrimExpr consumer_thread_extent = thread_iv_->dom->extent;
     PrimExpr producer_thread_extent = thread_iv_->dom->extent;
     // Need one warp-group for bulk-copy only case
@@ -1259,8 +1257,8 @@ class WarpSpecializedRewriter : public StmtExprMutator {
       PrimExpr arrive_thread_count =
           producer.released_barrier_.count(i)
               ? (producer.hasSimtCopy() ? producer_thread_extent : 1)
-              : (only_has_wgmma ? FloorDiv(consumer_thread_extent, 128)
-                                : consumer_thread_extent);
+              : (only_has_wgmma_ ? FloorDiv(consumer_thread_extent, 128)
+                                 : consumer_thread_extent);
       barrier_num_threads.push_back(arrive_thread_count);
     }
 
@@ -1289,6 +1287,7 @@ class WarpSpecializedRewriter : public StmtExprMutator {
   bool disable_warp_specialized_ = false;
   bool disable_shuffle_elect_ = false;
   Array<IntImm> nreg_;
+  bool only_has_wgmma_ = false;
 };
 
 class WarpSpecializedDetector : public IRVisitorWithAnalyzer {