OSSForks/tensorflow
1
0
Fork 0
mirror of https://github.com/zebrajr/tensorflow.git synced 2025-12-06 00:19:58 +01:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

[Upkeep] Resolve 4 instances of the following issue: Todo (resolved)

Browse Source 
PiperOrigin-RevId: 821824006
This commit is contained in:
Bill Varcho 2025-10-20 15:10:39 -07:00 committed by TensorFlower Gardener
parent 03819ab09c
commit a40f3bdebd
3 changed files with 0 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
third_party/xla/xla/hlo/transforms/collectives/collectives_schedule_linearizer.cc vendored
View File

@ -31,7 +31,6 @@ limitations under the License.
namespace xla {
// TODO(b/181653482): Fix for interprocedural collectives as well.
absl::StatusOr<bool> CollectivesScheduleLinearizer::Run(
HloModule* module,
const absl::flat_hash_set<absl::string_view>& execution_threads) {

2
third_party/xla/xla/tests/collective_pipeline_parallelism_test.cc vendored
View File

@ -1274,7 +1274,6 @@ TEST_P(CollectivePipelineParallelismTest,
// This is the partially pipelined version of
// NaiveBFSMicrobatch5CircularRepeat2Replica4 and should yield the same results.
// TODO(b/383868854): replace this with GPU pipeliner implementation.
TEST_P(CollectivePipelineParallelismTest,
NaiveBFSMb5Cr2Replica4SendRecvPartiallyPipelined) {
constexpr char kMoreComputationsStr[] = R"(
@ -1508,7 +1507,6 @@ TEST_P(CollectivePipelineParallelismTest,
// This is the async-grouped version of
// NaiveBFSMicrobatch5CircularRepeat2Replica4 and should yield the same results.
// TODO(b/383868854): replace this with GPU pipeliner implementation.
TEST_P(CollectivePipelineParallelismTest,
NaiveBFSMb5Cr2Replica4SendRecvAsyncGroup) {
constexpr char kMoreComputationsStr[] = R"(

2
third_party/xla/xla/tools/collective_perf_table_gen_main.cc vendored
View File

@ -158,8 +158,6 @@ std::string DefaultCollectiveDevicesIfEmpty(
} // namespace
// TODO(b/390097558): Add an option to generate perf table for collective which
// gets overlap to model resource contention.
int main(int argc, char* argv[]) {
// Default args.
int32_t num_nodes = 1;