experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[TPU] Don't pass host_shapes across the TPU API boundary. They can be computed from device shapes.

Browse Source 
PiperOrigin-RevId: 335996516
Change-Id: I702ff81ce311042399ef87bddb5e0d68b7464331
This commit is contained in:
Peter Hawkins 2020-10-07 18:43:53 -07:00 committed by TensorFlower Gardener
parent cc33e1f05b
commit 6a7391ca02
5 changed files with 5 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
tensorflow/compiler/xla/service/executable.h
View File

@ -153,6 +153,9 @@ class ExecutionOutput {
std::vector<se::OwningDeviceMemory> to_be_released) std::vector<se::OwningDeviceMemory> to_be_released)
: result_(std::move(result)), : result_(std::move(result)),
to_be_released_(std::move(to_be_released)) {} to_be_released_(std::move(to_be_released)) {}
ExecutionOutput(Shape on_device_shape, se::DeviceMemoryAllocator* allocator,
int device_ordinal)
: result_(std::move(on_device_shape), allocator, device_ordinal) {}
ExecutionOutput(Shape on_host_shape, Shape on_device_shape, ExecutionOutput(Shape on_host_shape, Shape on_device_shape,
se::DeviceMemoryAllocator* allocator, int device_ordinal) se::DeviceMemoryAllocator* allocator, int device_ordinal)
: result_(std::move(on_host_shape), std::move(on_device_shape), allocator, : result_(std::move(on_host_shape), std::move(on_device_shape), allocator,

2
tensorflow/core/tpu/tpu_on_demand_compiler.cc
View File

@ -119,7 +119,6 @@ class TpuExecutable : public TpuExecutableInterface {
} }
ApiConverter::ToC(arg.shape(), &se_args[i]->dynamic_shape); ApiConverter::ToC(arg.shape(), &se_args[i]->dynamic_shape);
ApiConverter::ToC(arg.host_shape(), &se_args[i]->host_shape);
const auto& unowned_indices = arg.unowned_indices(); const auto& unowned_indices = arg.unowned_indices();
se_args[i]->unowned_indices_size = unowned_indices.size(); se_args[i]->unowned_indices_size = unowned_indices.size();
se_args[i]->unowned_indices = new XLA_ShapeIndex[unowned_indices.size()]; se_args[i]->unowned_indices = new XLA_ShapeIndex[unowned_indices.size()];
@ -142,7 +141,6 @@ class TpuExecutable : public TpuExecutableInterface {
for (int i = 0; i < arguments.size(); ++i) { for (int i = 0; i < arguments.size(); ++i) {
ApiConverter::Free(&se_args[i]->shape_tree.shape); ApiConverter::Free(&se_args[i]->shape_tree.shape);
ApiConverter::Free(&se_args[i]->dynamic_shape); ApiConverter::Free(&se_args[i]->dynamic_shape);
ApiConverter::Free(&se_args[i]->host_shape);
delete[] se_args[i]->unowned_indices; delete[] se_args[i]->unowned_indices;
delete[] se_args[i]->shape_tree.buffers; delete[] se_args[i]->shape_tree.buffers;
delete se_args[i]; delete se_args[i];

5
tensorflow/stream_executor/tpu/c_api_conversions.cc
View File

@ -23,7 +23,6 @@ limitations under the License.
namespace ApiConverter { namespace ApiConverter {
xla::ShapedBuffer FromC(XLA_ShapedBuffer* c_buffer) { xla::ShapedBuffer FromC(XLA_ShapedBuffer* c_buffer) {
xla::Shape xla_on_host_shape = ApiConverter::FromC(&c_buffer->on_host_shape);
xla::Shape xla_on_device_shape = xla::Shape xla_on_device_shape =
ApiConverter::FromC(&c_buffer->on_device_shape); ApiConverter::FromC(&c_buffer->on_device_shape);
@ -36,7 +35,7 @@ xla::ShapedBuffer FromC(XLA_ShapedBuffer* c_buffer) {
} }
xla::ShapedBuffer xla_shaped_buffer( xla::ShapedBuffer xla_shaped_buffer(
xla_on_host_shape, xla_on_device_shape, xla_on_device_shape,
tensorflow::tpu::TpuPlatformInterface::GetRegisteredPlatform(), tensorflow::tpu::TpuPlatformInterface::GetRegisteredPlatform(),
c_buffer->device_ordinal); c_buffer->device_ordinal);
xla_shaped_buffer.set_buffers(xla_shape_tree); xla_shaped_buffer.set_buffers(xla_shape_tree);
@ -199,7 +198,6 @@ xla::MutableBorrowingLiteral FromC(XLA_Literal* c_literal) {
} }
void ToC(const xla::ShapedBuffer& buffer, XLA_ShapedBuffer* c_device_buffer) { void ToC(const xla::ShapedBuffer& buffer, XLA_ShapedBuffer* c_device_buffer) {
ApiConverter::ToC(buffer.on_host_shape(), &c_device_buffer->on_host_shape);
ApiConverter::ToC(buffer.on_device_shape(), ApiConverter::ToC(buffer.on_device_shape(),
&c_device_buffer->on_device_shape); &c_device_buffer->on_device_shape);
c_device_buffer->device_ordinal = buffer.device_ordinal(); c_device_buffer->device_ordinal = buffer.device_ordinal();
@ -226,7 +224,6 @@ void Free(XLA_Literal* c_literal) {
void Free(XLA_ShapedBuffer* c_buffer) { void Free(XLA_ShapedBuffer* c_buffer) {
ApiConverter::Free(&c_buffer->on_device_shape); ApiConverter::Free(&c_buffer->on_device_shape);
ApiConverter::Free(&c_buffer->on_host_shape);
delete[] c_buffer->bases; delete[] c_buffer->bases;
} }

2
tensorflow/stream_executor/tpu/c_api_decl.h
View File

@ -177,7 +177,6 @@ typedef struct XLA_Shape {
// Represents a leaf node for a XLA shaped buffer. // Represents a leaf node for a XLA shaped buffer.
typedef struct XLA_ShapedBuffer { typedef struct XLA_ShapedBuffer {
XLA_Shape on_host_shape;
XLA_Shape on_device_shape; XLA_Shape on_device_shape;
int device_ordinal; int device_ordinal;
@ -208,7 +207,6 @@ typedef struct SE_ExecutionInput {
XLA_ShapeIndex* unowned_indices; XLA_ShapeIndex* unowned_indices;
int unowned_indices_size; int unowned_indices_size;
XLA_Shape dynamic_shape; XLA_Shape dynamic_shape;
XLA_Shape host_shape;
} SE_ExecutionInput; } SE_ExecutionInput;
typedef struct SE_ExecutionOutput { typedef struct SE_ExecutionOutput {

3
tensorflow/stream_executor/tpu/tpu_executable_interface.cc
View File

@ -90,8 +90,7 @@ TpuExecutableInterface::AllocateOutputMemoryWithInputReuse(
} }
} }
ExecutionOutput result(host_shape, std::move(device_shape), allocator, ExecutionOutput result(std::move(device_shape), allocator, device_ordinal);
device_ordinal);
// Iterate through and allocate a buffer for each shape index, checking for // Iterate through and allocate a buffer for each shape index, checking for
// possible input buffer reuse. // possible input buffer reuse.
int64 reused_buffer_bytes = 0; int64 reused_buffer_bytes = 0;