chore: remove cpp benchmarks, tests, cmake path, as they are deprecated (#1345)

After discovering that these tests and benchmarks do not build, I
attempted to repair them
[here](#1336). I was
informed that this part of the code is deprecated, and that CMake is not
used to perform builds anymore.

To avoid anyone else attempting to use this code and spending time
trying to repair it, I suggest that it be removed.

Author: hypdeb

.gitignore

@@ -12,7 +12,6 @@ flashinfer/_build_meta.py
 flashinfer/data/
 flashinfer/jit/aot_config.py
 aot-ops/
-src/generated/
 csrc/aot_default_additional_params.h
 
 # DS_Store files

.gitmodules

@@ -1,9 +1,3 @@
-[submodule "3rdparty/nvbench"]
-	path = 3rdparty/nvbench
-	url = https://github.com/NVIDIA/nvbench.git
-[submodule "3rdparty/googletest"]
-	path = 3rdparty/googletest
-	url = https://github.com/google/googletest.git
 [submodule "3rdparty/cutlass"]
 	path = 3rdparty/cutlass
 	url = https://github.com/NVIDIA/cutlass.git

.pre-commit-config.yaml

@@ -53,10 +53,4 @@ repos:
       - id: clang-format
         types_or: [c++, c, cuda]
         exclude: |
-          (?x)^(3rdparty/.* src/generated/.* flashinfer/jit/aot_config.py)$
-
-  - repo: https://github.com/cheshirekow/cmake-format-precommit
-    rev: v0.6.13
-    hooks:
-      - id: cmake-format
-        additional_dependencies: [pyyaml>=5.1]
+          (?x)^(3rdparty/.* flashinfer/jit/aot_config.py)$

3rdparty/googletest

@@ -134,20 +134,6 @@ Check out [documentation](https://docs.flashinfer.ai/) for usage of batch decode
 
 Starting from FlashInfer v0.2, users can customize their own attention variants with additional parameters. For more details, refer to our [JIT examples](https://github.com/flashinfer-ai/flashinfer/blob/main/tests/test_jit_example.py).
 
-## Run Benchmarks
-
-We profile FlashInfer kernel performance with [nvbench](https://github.com/NVIDIA/nvbench) and you can compile and run the benchmarks with the following commands:
-
-```bash
-mkdir build
-cp cmake/config.cmake build # you can modify the config.cmake to enable/disable benchmarks and change CUDA architectures
-cd build
-cmake ..
-make -j12
-```
-
-You can run `./bench_{single/batch}_{prefill/decode}` to benchmark the performance (e.g. `./bench_single_prefill` for single-request prefill attention). `./bench_{single/batch}_{prefill/decode} --help` will show you the available options.
-
 ## C++ API and TVM Bindings
 
 FlashInfer also provides C++ API and TVM bindings, please refer to [documentation](https://docs.flashinfer.ai/) for more details.