[Mlir-commits] [mlir] [linalg] Add quantized version of `conv_3d_ncdhw_fcdhw` (PR #113953)

[Javed Absar]

================
@@ -694,3 +694,18 @@ func.func @conv2d_channel_first_q_promote(%img: tensor<100x3x224x224xi8>, %filt:
 // CHECK-LABEL: func @conv2d_channel_first_q_promote(
 // CHECK:   %[[arg0:[a-zA-z0-9]*]]: tensor<100x3x224x224xi8>, %[[arg1:[a-zA-z0-9]*]]: tensor<64x3x5x5xi8>, %[[arg2:[a-zA-z0-9]*]]: i8, %[[arg3:[a-zA-z0-9]*]]: i8)
 // CHECK:         linalg.conv_2d_nchw_fchw_q {dilations = dense<1> : tensor<2xi64>, strides = dense<1> : tensor<2xi64>} ins(%[[arg0]], %[[arg1]], %[[arg2]], %[[arg3]] : tensor<100x3x224x224xi8>, tensor<64x3x5x5xi8>, i8, i8) outs(%{{.*}} : tensor<100x64x220x220xi32>) -> tensor<100x64x220x220xi32>
+
+// -----
+
+func.func @conv3d_channel_first_q(%img: tensor<1x27x49x48x47xi8>, %filt: tensor<28x27x3x4x5xi8>, %a: i32, %b: i32) -> tensor<1x28x47x45x43xi32> {
+  %init = arith.constant dense<0> : tensor<1x28x47x45x43xi32>
+  %1 = linalg.conv_3d_ncdhw_fcdhw_q  {dilations = dense<1> : tensor<3xi64>,
+      strides = dense<1> : tensor<3xi64>}
+    ins(%img, %filt, %a, %b : tensor<1x27x49x48x47xi8>, tensor<28x27x3x4x5xi8>, i32, i32)
----------------
javedabsar1 wrote:

Do both (%image, %filter) have to be same type?

https://github.com/llvm/llvm-project/pull/113953