https://github.com/ocervell/benchtools

Benchmarking software for CPU, GPU, Memory, Processor and Network evaluations [C, Python, Shell]
https://github.com/ocervell/benchtools

Last synced: about 1 month ago
JSON representation

Benchmarking software for CPU, GPU, Memory, Processor and Network evaluations [C, Python, Shell]

• Host: GitHub
• URL: https://github.com/ocervell/benchtools
• Owner: ocervell
• Created: 2015-06-09T19:09:00.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2015-06-09T19:37:38.000Z (over 9 years ago)
• Last Synced: 2023-03-02T08:41:20.442Z (over 1 year ago)
• Language: Python
• Homepage:
• Size: 156 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# BenchTools

## Benchmarking software for CPU, GPU, Memory and Network evaluations

Run a full benchmark

--------------------

The complete benchmark has been automated using a shell script, *auto.sh*, to allow simplicity of use.

**bash auto.sh**

Reports will be generated in the current folder for each test, under the form report_TEST_TIME

*TEST* is *CPU*, *MEMORY*, *DISK* or *NETWORK*. *GPU* is not included in this automated test tool, you'll have to run it manually.

*TIME* is the current time to the second

The shell script will automatically compile the files. You don't have to do it manually.

It will then automatically run all the experiments. 

*IMPORTANT* : Near the end of the benchmark, the network benchmark **needs your input**.

You need to :

* *Open* a new terminal

* *Go* to the *src* directory

* *Enter* the command line  **python client.py tcp**

* *Choose* between the three file sizes proposed (1B, 1kB, 64kB).

* After it's done, *hit* CTRL + C (Windows/Linux) or CMD + C (Mac) to exit the network and go to the next test, which is UDP.

* Repeat the previous operation for UDP (with for 3. the command line **python client.py udp** this time) and the program should finish successfully after that.

Separate compilation and run instruction

----------------------------------------

CPU

---

* **make -f make_cpu**

* **./cpu nb_threads**

* *nb_threads* is the number of threads you want.

GPU

---

* *Make sure* you have a GPU and all the libraries needed

* **bash gpu_script.sh**

* **./gpu blocksize nb_threads**

* *nb_threads*: Specify any number of threads you want to be run.

* *blocksize*: Common blocksize is 4kB (default), but you can put any value you want.

MEMORY

------

* **gcc -pthread memory.c -o memory**

* **./memory mode blocksize nb_threads**

  * mode 	   : Specify the mode you want among : s, r

	           * s = sequential

	           * r = random

  * blocksize  : Specify any blocksize you want (in bytes).

  * nb_threads : Specify any number of threads you want to be run

DISK

----

* **gcc -pthread disk.c -o disk**

* **./disk mode blocksize nb_threads**

  * mode	      : Specify the mode you want among *ws*, *wr*, *rs*, *rr*.

  	* *w* = write

  	* *r* = read

  	* *s* = sequential

  	* *r* = random

  * *blocksize* : Specify any blocksize you want (in bytes). 1MB = 1000000B

	* *nb_threads*: Specify any number of threads you want to be run

NETWORK

-------

* TCP

  * **python server.py tcp**

  * *Open* a new terminal

	* *Go* to the *src* directory

	* **python client.py tcp**

* UDP

  * python server.py udp

  * Open a new terminal

  * Go to the src directory

  * python client.py udp

Additional information

----------------------

The CPU module will run 3 tests in a row whereas the others only generate one test (so you would

have to runthem more than once if you want more test to be done).