Lines Matching refs:gpu
5 "gpu.launch"(%sz, %sz, %sz, %sz, %sz) ({
6 gpu.return
15 "gpu.launch"(%sz, %sz, %sz, %sz, %sz, %sz) ({
18 gpu.terminator
26 // @expected-note@+1 {{in 'gpu.launch' body region}}
27 gpu.launch blocks(%bx, %by, %bz) in (%sbx = %sz, %sby = %sz, %sbz = %sz)
29 // @expected-error@+2 {{expected 'gpu.terminator' or a terminator with successors}}
31 "gpu.yield"(%one) : (i32) -> ()
40 "gpu.launch_func"(%sz, %sz, %sz, %sz, %sz)
49 …// expected-error@+1 {{expected the closest surrounding module to have the 'gpu.container_module' …
50 gpu.launch_func @foo::@bar blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
56 module attributes {gpu.container_module} {
58 // expected-error@+1 {{'gpu.launch_func' op requires attribute 'kernel'}}
59 "gpu.launch_func"(%sz, %sz, %sz, %sz, %sz, %sz)
68 module attributes {gpu.container_module} {
70 // expected-error@+1 {{custom op 'gpu.launch_func' invalid kind of attribute specified}}
71 gpu.launch_func "foo" blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
78 module attributes {gpu.container_module} {
81 gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
88 module attributes {gpu.container_module} {
90 // expected-error@+1 {{'gpu.func' op expects parent op 'gpu.module'}}
91 gpu.func @kernel_1(%arg1 : !llvm.ptr<f32>) {
92 gpu.return
99 module attributes {gpu.container_module} {
105 gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
112 module attributes {gpu.container_module} {
113 gpu.module @kernels { }
117 gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
124 module attributes {gpu.container_module} {
125 gpu.module @kernels {
132 gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz)
139 module attributes {gpu.container_module} {
141 gpu.func @kernel_1(%arg1 : !llvm.ptr<f32>) kernel {
142 gpu.return
148 …gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%arg…
155 module attributes {gpu.container_module} {
156 gpu.module @kernels {
157 gpu.func @kernel_1(%arg1 : !llvm.ptr<f32>) {
158 gpu.return
163 // expected-error@+1 {{kernel function is missing the 'gpu.kernel' attribute}}
164 …gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%arg…
171 module attributes {gpu.container_module} {
172 gpu.module @kernels {
173 gpu.func @kernel_1(%arg1 : !llvm.ptr<f32>) kernel {
174 gpu.return
180 …gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%arg…
187 module attributes {gpu.container_module} {
188 gpu.module @kernels {
189 gpu.func @kernel_1(%arg1 : f32) kernel {
190 gpu.return
196 …gpu.launch_func @kernels::@kernel_1 blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%arg…
203 module attributes {gpu.container_module} {
206 …gpu.launch_func @foo::@bar blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%sz : index {…
215 %res = "gpu.all_reduce"(%arg0) ({}) : (f32) -> (f32)
223 %res = "gpu.all_reduce"(%arg0) ({
225 "gpu.yield"(%lhs) : (f32) -> ()
226 }) {op = #gpu<all_reduce_op add>} : (f32) -> (f32)
234 %res = gpu.all_reduce foo %arg0 {} : (f32) -> (f32)
242 %res = gpu.all_reduce and %arg0 {} : (f32) -> (f32)
250 %res = gpu.all_reduce %arg0 {
252 "gpu.yield"(%lhs) : (f32) -> ()
261 %res = gpu.all_reduce %arg0 {
263 "gpu.yield"(%lhs) : (f32) -> ()
271 // expected-error@+1 {{expected one gpu.yield operand}}
272 %res = gpu.all_reduce %arg0 {
274 "gpu.yield"(%lhs, %rhs) : (f32, f32) -> ()
282 // expected-error@+1 {{incorrect gpu.yield type}}
283 %res = gpu.all_reduce %arg0 {
286 "gpu.yield"(%one) : (i32) -> ()
294 // expected-error@+1 {{expected gpu.yield op in region}}
295 %res = gpu.all_reduce %arg0 {
306 …%shfl, %pred = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = #gpu<shuffle_mode xor> } : (f32, i32, i…
314 %shfl, %pred = gpu.shuffle xor %arg0, %arg1, %arg2 : index
321 gpu.module @gpu_funcs {
322 // expected-error @+1 {{custom op 'gpu.func' gpu.func requires named arguments}}
323 gpu.func @kernel_1(f32, f32) {
325 gpu.return
333 gpu.module @gpu_funcs {
335 "gpu.func"() ({
336 gpu.return
344 gpu.module @gpu_funcs {
346 gpu.func @kernel() workgroup(%0: i32) {
347 gpu.return
355 gpu.module @gpu_funcs {
357 gpu.func @kernel() workgroup(%0: memref<4xf32>) {
358 gpu.return
366 gpu.module @gpu_funcs {
368 gpu.func @kernel() private(%0: memref<4xf32>) {
369 gpu.return
377 gpu.module @gpu_funcs {
379 gpu.func @kernel() private(%0: memref<4xf32>) {
380 gpu.return
388 gpu.module @gpu_funcs {
390 gpu.func @kernel() {
392 // expected-error @+1 {{'gpu.return' op expected 0 result operands}}
393 gpu.return %0 : index
401 gpu.module @gpu_funcs {
402 // expected-error @+1 {{'gpu.func' op expected void return type for kernel function}}
403 gpu.func @kernel() -> index kernel {
405 gpu.return
413 gpu.module @gpu_funcs {
414 // expected-error @+1 {{'gpu.func' op expected at least 5 arguments to body region}}
415 "gpu.func"() ({
417 "gpu.return"() : () -> ()
418 …} ) {function_type = (f32, memref<?xf32>) -> (), gpu.kernel, sym_name = "kernel_1", workgroup_attr…
426 %t = gpu.wait
432 // expected-error @+1 {{custom op 'gpu.wait' needs to be named when marked 'async'}}
433 gpu.wait async
439 // expected-error @+1 {{'gpu.memcpy' op arguments have incompatible element type}}
440 gpu.memcpy %dst, %src : memref<?xf32>, memref<?xi32>
446 // expected-error @+1 {{'gpu.memcpy' op arguments have incompatible shape}}
447 gpu.memcpy %dst, %src : memref<7xf32>, memref<9xf32>
453 …// expected-error @+1 {{'gpu.memset' op failed to verify that all of {dst, value} have same elemen…
454 gpu.memset %dst, %value : memref<?xf32>, i32
463 …%0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, 3> …
473 …%0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, 3> …
483 …%0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, 3> …
495 …%0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, #la…
505 …%0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, 5> …
513 func.func @wmmaStoreOp_invalid_map(%arg0 : !gpu.mma_matrix<16x16xf16, "COp">) -> () {
518 …gpu.subgroup_mma_store_matrix %arg0, %sg[%i,%j] {leadDimension= 32 : index} : !gpu.mma_matrix<16x1…
524 func.func @wmmaStoreOp_invalid_mem_space(%arg0 : !gpu.mma_matrix<16x16xf16, "COp">) -> () {
529 …gpu.subgroup_mma_store_matrix %arg0, %sg[%i,%j] {leadDimension= 32 : index} : !gpu.mma_matrix<16x1…
535 func.func @wmmaStoreOp_invalid_store_operand(%arg0 : !gpu.mma_matrix<16x16xf16, "AOp">) -> () {
540 …gpu.subgroup_mma_store_matrix %arg0, %sg[%i,%j] {leadDimension= 32 : index} : !gpu.mma_matrix<16x1…
546 …aMmaOp_invalid_operand_order(%A : !gpu.mma_matrix<16x16xf16, "AOp">, %B : !gpu.mma_matrix<16x16xf1…
548 …%D = gpu.subgroup_mma_compute %B, %A, %C : !gpu.mma_matrix<16x16xf16, "BOp">, !gpu.mma_matrix<16x1…
554 …MmaOp_invalid_operand_shapes(%A : !gpu.mma_matrix<16x32xf16, "AOp">, %B : !gpu.mma_matrix<16x16xf1…
556 …%D = gpu.subgroup_mma_compute %A, %B, %C : !gpu.mma_matrix<16x32xf16, "AOp">, !gpu.mma_matrix<16x1…
565 %1 = gpu.alloc() : memref<2x4xf32, affine_map<(d0, d1)[s0] -> ((d0 + s0), d1)>, 1>
575 %1 = gpu.alloc()[%0] : memref<2x4xf32, 1>
585 %1 = gpu.alloc(%0, %0) : memref<2x?xf32, 1>
595 %1 = gpu.alloc(%0) : memref<2x?x?xf32, 1>