- HloConstantSplitter pass will ensure that each constant has a single use by
  duplicating multi-use constants.
- Doing this before sharding propagation prevents inadvertent and unintended
  propagation of sharding. As an example, a shared constant between automatic
  sharded region and manually sharded region can cause the manual sharding to
  leak into the automatic sharded code.

PiperOrigin-RevId: 481203013

PiperOrigin-RevId: 481195613

PiperOrigin-RevId: 481195475

PiperOrigin-RevId: 481193247

BEFORE

---------------------------------------------------------------
Benchmark                     Time             CPU   Iterations
---------------------------------------------------------------
BM_FlatMemrefX12All        81.3 ns         81.3 ns      8617874
BM_FlatMemrefX12None       69.1 ns         69.1 ns     10129964

AFTER

---------------------------------------------------------------
Benchmark                     Time             CPU   Iterations
---------------------------------------------------------------
BM_FlatMemrefX12All        54.8 ns         54.8 ns     12683098
BM_FlatMemrefX12None       47.7 ns         47.7 ns     14667555

PiperOrigin-RevId: 481192231

PiperOrigin-RevId: 481192084

PiperOrigin-RevId: 481190338

PiperOrigin-RevId: 481184853

PiperOrigin-RevId: 481184687

http://github.com/tensorflow/runtime/commit/0e71dd2f3edc157ac9ef78643618ef2bdad9718c.

PiperOrigin-RevId: 481184177

Updates LLVM usage to match
[06da9b94ae37](https://github.com/llvm/llvm-project/commit/06da9b94ae37)

PiperOrigin-RevId: 481181270

Log when TPUReshardVariables is inserted. This information is often needed for performance optimization.

PiperOrigin-RevId: 481181240

PiperOrigin-RevId: 481177778

PiperOrigin-RevId: 481176034

PiperOrigin-RevId: 481174166

This is something that I forgot about when reviewing cl/481105084.

Importer needed to be changed manually, but exports is autogenerated and already works properly.

PiperOrigin-RevId: 481164618

PiperOrigin-RevId: 481159038

This error was caused by shifting a signed integer:
```
third_party/tensorflow/compiler/mlir/lite/utils/low_bit_utils.cc:44:38: runtime error: left shift of negative value -98
```

This is fixed by more intentional casting.

PiperOrigin-RevId: 481157662

PiperOrigin-RevId: 481153499

PiperOrigin-RevId: 481153073

PiperOrigin-RevId: 481152442

PiperOrigin-RevId: 481150776

PiperOrigin-RevId: 481150516

PiperOrigin-RevId: 481143415

PiperOrigin-RevId: 481142604

	update of g3doc/_includes/tf_passes.md
	update of g3doc/hlo_ops.md

PiperOrigin-RevId: 481140074

PiperOrigin-RevId: 481135234

PiperOrigin-RevId: 481126296

The former name conflicts with the regular index_vector_dim from other scatters, which is always 1 here, so it's a bit confusing.

PiperOrigin-RevId: 481111212

hlo_module_importer is used in many places in ways we don't fully
understand. The assumptions they make about layouts seem to be
incompatible with the ones we make in the MLIR CPU pipeline.

Verified that tensorflow/compiler/xla/tests:pad_test_cpu
and tensorflow/compiler/xla/tests:client_test_cpu pass
with XLA_FLAGS=--xla_cpu_enable_mlir_lowering=true. This also
fixes tensorflow/compiler/xla/tests:slice_test_cpu.

PiperOrigin-RevId: 481108763

http://github.com/tensorflow/runtime/commit/a2cf754d51e0db0aeab14285c0377308959bf28d.

PiperOrigin-RevId: 481105622

PiperOrigin-RevId: 481105084

Prepare to change kDynamicSize to int64_t::min.

PiperOrigin-RevId: 481104488

Updates LLVM usage to match
[2f357054d705](https://github.com/llvm/llvm-project/commit/2f357054d705)

PiperOrigin-RevId: 481103846

PiperOrigin-RevId: 481101180

PiperOrigin-RevId: 481095035

TFLite CompatibilityList: Ensure native code is available by calling GpuDelegateNative::init(). This allows for a different library name and external loading by client.

PiperOrigin-RevId: 481093953

PiperOrigin-RevId: 481093435

PiperOrigin-RevId: 481092113

PiperOrigin-RevId: 481092096