diff --git a/test/TritonGPU/loop-pipeline.mlir b/test/TritonGPU/loop-pipeline.mlir
index a5594b304309..80444b152616 100644
--- a/test/TritonGPU/loop-pipeline.mlir
+++ b/test/TritonGPU/loop-pipeline.mlir
@@ -55,7 +55,7 @@
 // CHECK-DAG: %[[NEXT_B:.*]] = triton_gpu.memdesc_subview %{{.+}}[%[[EXT_IDX_3]],
 // CHECK-DAG: triton_gpu.async_wait {{.*}} {num = 2 : i32}
 // CHECK:   scf.yield {{.*}}, %[[INS_IDX_3]], %[[EXT_IDX_3]], %[[NEXT_A]], %[[NEXT_B]]
-module attributes {"triton_gpu.num-warps" = 4 : i32, "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.target" = "cuda:80"} {
+module attributes {"triton_gpu.num-warps" = 4 : i32, "triton_gpu.num-ctas" = 1 : i32} {
 tt.func @matmul_loop(%lb : index, %ub : index, %step : index,
                   %A : !tt.ptr<f16> {tt.divisibility = 16 : i32},
                   %B : !tt.ptr<f16> {tt.divisibility = 16 : i32}) -> tensor<128x128xf32, #C> {
@@ -582,7 +582,7 @@ tt.func @dep_arg_two_uses(%arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32},
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 8]}>
 #shared = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [0, 1], hasLeadingOffset = false}>
 #shared1 = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0], hasLeadingOffset = false}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
 // CHECK-LABEL: tt.func @load_two_users
   tt.func @load_two_users(%arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f16> {tt.divisibility = 16 : i32}) -> (tensor<128x16xf32, #mma>, tensor<128x64xf32, #mma>) {
     %cst = arith.constant dense<0> : tensor<1x16xi32, #blocked>
@@ -643,7 +643,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 8]}>
 #shared = #triton_gpu.shared<{vec = 4, perPhase = 1, maxPhase = 2, order = [0, 1], hasLeadingOffset = false}>
 #shared1 = #triton_gpu.shared<{vec = 4, perPhase = 1, maxPhase = 2, order = [1, 0], hasLeadingOffset = false}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
 // CHECK-LABEL: tt.func @load_two_users_incompatible_layouts
   tt.func @load_two_users_incompatible_layouts(%arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f16> {tt.divisibility = 16 : i32}) -> (tensor<128x16xf32, #mma>, tensor<128x64xf32, #mma>) {
     %cst = arith.constant dense<0> : tensor<1x16xi32, #blocked>
@@ -728,7 +728,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 // would be pipelined.
 #blocked = #triton_gpu.blocked<{sizePerThread = [1, 4], threadsPerWarp = [4, 8], warpsPerCTA = [4, 1], order = [1, 0]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [2, 2], instrShape = [16, 8]}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @nested_loops(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg2: !tt.ptr<i32> {tt.divisibility = 16 : i32}, %arg3: !tt.ptr<f32> {tt.divisibility = 16 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<32x32xf32, #mma>
     %cst_0 = arith.constant dense<320> : tensor<32x1xi32, #blocked>
@@ -790,7 +790,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 8]}>
 #shared = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 4, order = [0, 1], hasLeadingOffset = false}>
 #shared1 = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 4, order = [1, 0], hasLeadingOffset = false}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @_jagged_hstu_attn_fwd_0d1d2d3d4d5de(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg2: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg3: !tt.ptr<i64> {tt.divisibility = 16 : i32}, %arg4: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg5: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 8 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<64x32xf32, #mma>
     %c64_i32 = arith.constant 64 : i32
@@ -903,7 +903,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #C = #triton_gpu.nvidia_mma<{versionMajor = 2, warpsPerCTA = [4, 1]}>
 #A = #triton_gpu.dot_op<{opIdx = 0, parent = #C, kWidth=2}>
 #B = #triton_gpu.dot_op<{opIdx = 1, parent = #C, kWidth=2}>
-module attributes {"triton_gpu.target" = "cuda:86", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
 tt.func @indirect_load_shared_layout(%77: tensor<16x16xi64, #BL> {tt.divisibility=16: i32, tt.constancy=16: i32},
                    %76: index,
                    %49: tensor<16x16x!tt.ptr<f16>, #AL> {tt.divisibility=16: i32, tt.contiguity=2 : i32},
@@ -948,7 +948,7 @@ tt.func @indirect_load_shared_layout(%77: tensor<16x16xi64, #BL> {tt.divisibilit
 // CHECK: tt.return
 #blocked = #triton_gpu.blocked<{sizePerThread = [1, 4], threadsPerWarp = [4, 8], warpsPerCTA = [4, 1], order = [1, 0]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [2, 2], instrShape = [16, 8]}>
-module attributes {"triton_gpu.target" = "cuda:86", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @kernel_yield_constant(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg2: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}, %arg3: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}, %arg4: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}, %arg5: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}, %arg6: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<32x32xf32, #mma>
     %cst1 = arith.constant dense<1.000000e+00> : tensor<32x32xf32, #mma>
@@ -1003,7 +1003,7 @@ module attributes {"triton_gpu.target" = "cuda:86", "triton_gpu.num-ctas" = 1 :
 // CHECK:   triton_gpu.async_copy_global_to_local {{.*}}, %[[B1BUFFER]]
 // CHECK:   scf.for
 #blocked = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
-module attributes {"triton_gpu.target" = "cuda:90", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @add_kernel(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg2: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg3: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}) attributes {noinline = false} {
     %c1024_i32 = arith.constant 1024 : i32
     %c0_i32 = arith.constant 0 : i32
@@ -1072,7 +1072,7 @@ module attributes {"triton_gpu.target" = "cuda:90", "triton_gpu.num-ctas" = 1 :
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [1, 2], instrShape = [16, 8]}>
 #shared = #triton_gpu.shared<{vec = 4, perPhase = 2, maxPhase = 4, order = [1, 0], hasLeadingOffset = false}>
 #shared1 = #triton_gpu.shared<{vec = 4, perPhase = 2, maxPhase = 4, order = [0, 1], hasLeadingOffset = false}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 2 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 2 : i32} {
   tt.func public @nested_loops(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<16x16xf32, #mma>
     %c1_i32 = arith.constant 1 : i32
@@ -1116,7 +1116,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #blocked4 = #triton_gpu.blocked<{sizePerThread = [16, 2, 1], threadsPerWarp = [4, 1, 8], warpsPerCTA = [1, 1, 8], order = [1, 0, 2]}>
 #blocked5 = #triton_gpu.blocked<{sizePerThread = [32, 1], threadsPerWarp = [4, 8], warpsPerCTA = [1, 8], order = [0, 1]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [1, 8], instrShape = [16, 8]}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 8 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 8 : i32} {
   tt.func public @int4_matmul_ampere(
     %arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32},
     %arg1: !tt.ptr<i8> {tt.divisibility = 16 : i32}
@@ -1191,7 +1191,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #A = #triton_gpu.dot_op<{opIdx = 0, parent = #C, kWidth=2}>
 #B = #triton_gpu.dot_op<{opIdx = 1, parent = #C, kWidth=2}>
 
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
 tt.func @load_convert_layout(%77: tensor<16x16xi64, #BL> {tt.divisibility=16: i32, tt.constancy=16: i32},
                    %76: index,
                    %49: tensor<16x16x!tt.ptr<f16>, #AL> {tt.divisibility=16: i32, tt.contiguity=2 : i32},
@@ -1235,7 +1235,7 @@ tt.func @load_convert_layout(%77: tensor<16x16xi64, #BL> {tt.divisibility=16: i3
 // CHECK-LABEL: @matmul_indirect_pipeline
 #blocked = #triton_gpu.blocked<{sizePerThread = [1, 1], threadsPerWarp = [32, 1], warpsPerCTA = [1, 2], order = [0, 1]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [2, 1], instrShape = [16, 8]}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 2 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 2 : i32} {
   tt.func public @matmul_indirect_pipeline(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<i64> {tt.divisibility = 16 : i32}, %arg2: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg3: !tt.ptr<f32> {tt.divisibility = 16 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<32x32xf32, #mma>
     %c1_i32 = arith.constant 1 : i32
@@ -1279,7 +1279,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 // CHECK-NOT: local_load{{.*}}128x1
 #blocked = #triton_gpu.blocked<{sizePerThread = [1, 1], threadsPerWarp = [32, 1], warpsPerCTA = [4, 1], order = [0, 1]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 8]}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @dont_pipeline_128x1(%arg6: !tt.ptr<i32> {tt.divisibility = 16 : i32}) attributes {noinline = false} {
     %cst = arith.constant dense<0.000000e+00> : tensor<128x64xf32, #mma>
     %c128_i32 = arith.constant 128 : i32
@@ -1330,7 +1330,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #A = #triton_gpu.dot_op<{opIdx = 0, parent = #C, kWidth=2}>
 #B = #triton_gpu.dot_op<{opIdx = 1, parent = #C, kWidth=2}>
 
-module attributes {"triton_gpu.num-warps" = 4 : i32, "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.target" = "cuda:80"} {
+module attributes {"triton_gpu.num-warps" = 4 : i32, "triton_gpu.num-ctas" = 1 : i32} {
 tt.func @matmul_nested_ops(%lb : index, %ub : index, %step : index,
                   %A : !tt.ptr<f16> {tt.divisibility = 16 : i32},
                   %B : !tt.ptr<f16> {tt.divisibility = 16 : i32},
@@ -1388,7 +1388,7 @@ tt.func @matmul_nested_ops(%lb : index, %ub : index, %step : index,
 #mma1 = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 16, 16]}>
 #shared = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0], hasLeadingOffset = true}>
 #shared1 = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [0, 1], hasLeadingOffset = true}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   // CHECK-LABEL: dot_prologue_epilogue
   // CHECK: {{.*}}, {{.*}}, %[[EXT:.*]]: i32, {{.*}}
   tt.func @dot_prologue_epilogue(%arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %ext: i32, %inc: tensor<64x16xi32, #blocked> {tt.divisibility = 16 : i32}) -> tensor<128x16xf32, #mma1> {
@@ -1460,7 +1460,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 #mma1 = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 16, 16]}>
 #shared = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0], hasLeadingOffset = true}>
 #shared1 = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [0, 1], hasLeadingOffset = true}>
-module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   // CHECK-NOCANON-LABEL: pipeline_downstream_dependencies
   // CHECK-NOCANON: {{.*}}, {{.*}}, %[[EXT:.*]]: i32, {{.*}}
   tt.func @pipeline_downstream_dependencies(%arg0: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f16> {tt.divisibility = 16 : i32}, %ext: i32, %inc: tensor<64x16xi32, #blocked> {tt.divisibility = 16 : i32}) -> tensor<128x16xf32, #mma1> {
@@ -1528,7 +1528,7 @@ module attributes {"triton_gpu.target" = "cuda:80", "triton_gpu.num-ctas" = 1 :
 // CHECK: arith.select {{.*}}, %[[B]], %[[CONSTANT]]
 
 #blocked = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
-module attributes {"triton_gpu.target" = "cuda:90", "triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
   tt.func public @masked_add_kernel(%arg0: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg1: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg2: !tt.ptr<f32> {tt.divisibility = 16 : i32}, %arg3: i32 {tt.divisibility = 16 : i32, tt.max_divisibility = 16 : i32}) attributes {noinline = false} {
     %c1024_i32 = arith.constant 1024 : i32
     %c0_i32 = arith.constant 0 : i32
@@ -1565,7 +1565,7 @@ module attributes {"triton_gpu.target" = "cuda:90", "triton_gpu.num-ctas" = 1 :
 #blocked1 = #triton_gpu.blocked<{sizePerThread = [1, 1], threadsPerWarp = [1, 32], warpsPerCTA = [1, 4], order = [1, 0]}>
 #mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 256, 16]}>
 #shared = #triton_gpu.shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0], hasLeadingOffset = true}>
-module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32, triton_gpu.target = "cuda:90", "triton_gpu.threads-per-warp" = 32 : i32} {
+module attributes {"triton_gpu.num-ctas" = 1 : i32, "triton_gpu.num-warps" = 4 : i32} {
 //   CHECK-LABEL: @matmul_tma
 //     CHECK-DAG:   triton_gpu.local_alloc  : () -> !tt.memdesc<3x128x64xf16, #{{.+}}, #triton_gpu.shared_memory, mutable>
 //     CHECK-DAG:   triton_gpu.local_alloc  : () -> !tt.memdesc<3x64x256xf16, #{{.+}}, #triton_gpu.shared_memory, mutable>