Squashed commit of the following:

commit 6973d819fda5b763e62f5a54c41bdf9049b90e9c
Author: Ash Vardanian <[email protected]>
Date:   Mon May 13 01:39:14 2024 +0000

    Chore: Format benchmark

commit 13ad19a
Author: Victor Zverovich <[email protected]>
Date:   Sun May 12 18:13:02 2024 -0700

    Fix: Running C++ & Py benchmarks

examples/README.md

@@ -17,55 +17,14 @@ All of them implement identical endpoints:
 - `resize` - batch-capable image processing operation for Base64-encoded images
 - `dot_product` - batch-capable matrix vector-vector multiplication operation
 
-## Echo and Summation
-
-The `echo` example is a simple demonstration of a server that echoes back the request.
-To make it a bit more challenging, the same server also contains the `sum` endpoint, which accepts two numbers and returns their aggregate.
-
-- `echo_server_ucall.cpp` - UCall server implementation in C++
-- `echo_server_ucall.py` - UCall server implementation in Python
-- `echo_server_fastapi.py` - UCall server implementation in FastAPI
-- `echo_client.py` - shared client implementation for benchmarking
-
-To run those:
-
-```sh
-
-```
-
-## Redis
-
-Redis example is a basic in-memory key-value store implementation with just two endpoints - `set` and `get`.
-The server implementations are available in C++ and Python for UCall.
-
-- `redis_server_ucall.cpp` - UCall server implementation in C++
-- `redis_server_ucall.py` - UCall server implementation in Python
-- `redis_server_fastapi.py` - UCall server implementation in FastAPI
-- `redis_client.py` - shared client implementation for benchmarking
-
-## Multimedia
-
-Multimedia example showcases the more edvanced UCall functionality, like built-in support for NumPy arrays and Pillow images.
-It also uses batch endpoints for AI inference and image processing, the functionality not available in FastAPI.
-
-- `multimedia_server_ucall.cpp` - UCall server implementation in C++
-- `multimedia_server_ucall.py` - UCall server implementation in Python
-- `multimedia_client.py` - shared client implementation for benchmarking
-
-
-===
-
-The simplest possible endpoint after `hello-world` and `echo`, is probably `sum`.
-We would just accept two numbers and return their aggregate.
-Packets are tiny, so it is great for benchmarking the request latency.
 
 ## Reproducing Benchmarks
 
 ### FastAPI
 
 ```sh
 pip install uvicorn fastapi websocket-client requests tqdm fire
-cd examples && uvicorn sum.fastapi_server:app --log-level critical &
+cd examples && uvicorn fastapi_server:app --log-level critical &
 cd ..
 python examples/bench.py "fastapi_client.ClientREST" --progress
 python examples/bench.py "fastapi_client.ClientWebSocket" --progress
@@ -82,14 +41,14 @@ python examples/bench.py "fastapi_client.ClientWebSocket" --threads 8
 ### UCall
 
 UCall can produce both a POSIX compliant old-school server, and a modern `io_uring`-based version for Linux kernel 5.19 and newer.
-You would either run `ucall_example_sum_posix` or `ucall_example_sum_uring`.
+You would either run `ucall_example_server_posix` or `ucall_example_server_uring`.
 
 ```sh
 sudo apt-get install cmake g++ build-essential
 cmake -DCMAKE_BUILD_TYPE=Release -B build_release
 cmake --build build_release --config Release
-build_release/build/bin/ucall_example_sum_posix &
-build_release/build/bin/ucall_example_sum_uring &
+build_release/build/bin/ucall_example_server_posix &
+build_release/build/bin/ucall_example_server_uring &
 python examples/bench.py "jsonrpc_client.CaseTCP" --progress
 python examples/bench.py "jsonrpc_client.CaseHTTP" --progress
 python examples/bench.py "jsonrpc_client.CaseHTTPBatches" --progress
@@ -99,7 +58,7 @@ kill %%
 Want to customize server settings?
 
 ```sh
-build_release/build/bin/ucall_example_sum_uring --nic=127.0.0.1 --port=8545 --threads=16 --silent=false
+build_release/build/bin/ucall_example_server_uring --nic=127.0.0.1 --port=8545 --threads=16 --silent=false
 ```
 
 Want to dispatch more clients and aggregate more accurate statistics?

examples/bench.py

@@ -140,7 +140,10 @@ def main(
     progress: bool = False,
 ):
     script_dir = os.path.dirname(os.path.abspath(__file__))
+    project_dir = os.path.dirname(script_dir)
     sys.path.append(f"{script_dir}/login")
+    sys.path.append(os.path.join(project_dir, "src"))
+
     class_ = locate(class_name)
     stats = bench_parallel(
         callable=class_(),