From 8f61c858800c6eb04ea57017a2b61c46326b0951 Mon Sep 17 00:00:00 2001
From: Georg Lehmann <dadschoorse@gmail.com>
Date: Wed, 23 Jul 2025 18:58:09 +0200
Subject: [PATCH] aco/statistics: add latency to WMMA

Assume the normal VALU latency of 4 cycles.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36328>
---
 src/amd/compiler/aco_statistics.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/compiler/aco_statistics.cpp b/src/amd/compiler/aco_statistics.cpp
index 0df5e8ab2d4..4a370215683 100644
--- a/src/amd/compiler/aco_statistics.cpp
+++ b/src/amd/compiler/aco_statistics.cpp
@@ -192,7 +192,7 @@ get_perf_info(const Program& program, const Instruction& instr)
       case instr_class::wmma: {
          /* int8 and (b)f16 have the same performance. */
          uint8_t cost = instr.opcode == aco_opcode::v_wmma_i32_16x16x16_iu4 ? 16 : 32;
-         return {cost, WAIT_USE(valu, cost)};
+         return {4 + cost, WAIT_USE(valu, cost)};
       }
       case instr_class::barrier:
       case instr_class::waitcnt: