Bump TFLite Version

[boomanaiden154]

This patch bumps the TFLite version (along with versions of dependencies) so that we can get rid of the pthreadpool dependency.

[boomanaiden154]

Maybe this isn't ready. I'm seeing some LLVM test failures:

********************
FAIL: LLVM :: CodeGen/MLRegAlloc/dev-mode-prio-logging.ll (582 of 24983)
******************** TEST 'LLVM :: CodeGen/MLRegAlloc/dev-mode-prio-logging.ll' FAILED ********************
Exit Code: 134

Command Output (stdout):
--
Writing output spec to /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel/output_spec.json.

--
Command Output (stderr):
--
RUN: at line 6: /llvm-project/build/bin/llc -o /dev/null -mtriple=x86_64-linux-unknown -regalloc=greedy    -regalloc-enable-priority-advisor=development    -regalloc-priority-training-log=/llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1    < /llvm-project/llvm/test/CodeGen/MLRegAlloc/Inputs/input.ll
+ /llvm-project/build/bin/llc -o /dev/null -mtriple=x86_64-linux-unknown -regalloc=greedy -regalloc-enable-priority-advisor=development -regalloc-priority-training-log=/llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1
RUN: at line 10: "/usr/bin/python3.10" /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/log_reader.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1 > /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1.readable
+ /usr/bin/python3.10 /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/log_reader.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1
RUN: at line 11: /llvm-project/build/bin/FileCheck --input-file /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1.readable /llvm-project/llvm/test/CodeGen/MLRegAlloc/dev-mode-prio-logging.ll --check-prefixes=CHECK,NOML
+ /llvm-project/build/bin/FileCheck --input-file /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1.readable /llvm-project/llvm/test/CodeGen/MLRegAlloc/dev-mode-prio-logging.ll --check-prefixes=CHECK,NOML
RUN: at line 12: diff /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1.readable /llvm-project/llvm/test/CodeGen/MLRegAlloc/Inputs/reference-prio-log-noml.txt
+ diff /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp1.readable /llvm-project/llvm/test/CodeGen/MLRegAlloc/Inputs/reference-prio-log-noml.txt
RUN: at line 14: rm -rf /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp && mkdir /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp
+ rm -rf /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp
+ mkdir /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp
RUN: at line 15: "/usr/bin/python3.10" /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/gen-regalloc-priority-test-model.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel
+ /usr/bin/python3.10 /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/gen-regalloc-priority-test-model.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel
2024-07-04 17:31:49.868359: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:31:49.934849: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:31:49.935581: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-07-04 17:31:51.274415: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
RUN: at line 16: "/usr/bin/python3.10" /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp
+ /usr/bin/python3.10 /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp
2024-07-04 17:31:55.289130: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:31:55.361753: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:31:55.362503: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-07-04 17:31:56.767185: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
2024-07-04 17:31:59.944727: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:364] Ignored output_format.
2024-07-04 17:31:59.944897: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:367] Ignored drop_control_dependency.
2024-07-04 17:31:59.946121: I tensorflow/cc/saved_model/reader.cc:45] Reading SavedModel from: /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel
2024-07-04 17:31:59.946628: I tensorflow/cc/saved_model/reader.cc:89] Reading meta graph with tags { serve }
2024-07-04 17:31:59.946705: I tensorflow/cc/saved_model/reader.cc:130] Reading SavedModel debug info (if present) from: /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel
2024-07-04 17:31:59.949693: F tensorflow/tsl/platform/default/env.cc:74] Check failed: ret == 0 (11 vs. 0)Thread tf_Compute creation via pthread_create() failed.
/llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.script: line 10: 386571 Aborted                 "/usr/bin/python3.10" /llvm-project/llvm/test/CodeGen/MLRegAlloc/../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp_savedmodel /llvm-project/build/test/CodeGen/MLRegAlloc/Output/dev-mode-prio-logging.ll.tmp

--

********************
******************** TEST 'LLVM :: Transforms/Inline/ML/bounds-checks-rewards.ll' FAILED ********************
Exit Code: 134

Command Output (stdout):
--
Output model to: [/llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel]
The model will always return: 1
Writing output spec to /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel/output_spec.json.

--
Command Output (stderr):
--
RUN: at line 10: rm -rf /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp
+ rm -rf /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp
RUN: at line 11: rm -rf /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel
+ rm -rf /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel
RUN: at line 12: "/usr/bin/python3.10" /llvm-project/llvm/test/Transforms/Inline/ML/../../../../lib/Analysis/models/gen-inline-oz-test-model.py /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel
+ /usr/bin/python3.10 /llvm-project/llvm/test/Transforms/Inline/ML/../../../../lib/Analysis/models/gen-inline-oz-test-model.py /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel
2024-07-04 17:30:35.732872: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:30:35.804436: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:30:35.805205: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-07-04 17:30:37.138657: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
RUN: at line 13: "/usr/bin/python3.10" /llvm-project/llvm/test/Transforms/Inline/ML/../../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp
+ /usr/bin/python3.10 /llvm-project/llvm/test/Transforms/Inline/ML/../../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp
2024-07-04 17:30:40.305600: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:30:40.346275: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.
2024-07-04 17:30:40.346837: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-07-04 17:30:41.626710: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
2024-07-04 17:30:44.001516: F tensorflow/tsl/platform/default/env.cc:74] Check failed: ret == 0 (11 vs. 0)Thread tf_Compute creation via pthread_create() failed.
/llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.script: line 10: 363961 Aborted                 "/usr/bin/python3.10" /llvm-project/llvm/test/Transforms/Inline/ML/../../../../lib/Analysis/models/saved-model-to-tflite.py /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp_savedmodel /llvm-project/build/test/Transforms/Inline/ML/Output/bounds-checks-rewards.ll.tmp

--

********************

[boomanaiden154]

Nevermind. That's some thread limit issue on my cloudtop. It goes away when running the tests individually and comes up with the previous version of build_tflite.sh as well.

[mtrofin]

can you remove the plibthreadpool installation in the setup script?

[boomanaiden154]

can you remove the plibthreadpool installation in the setup script?

Yep. Removed.