Docker-in-Docker (DinD) capabilities of public runners deactivated. More info

Commit e3e48d02 authored by Pierre Aubert's avatar Pierre Aubert
Browse files

Update doc

parent 2abeacc7
Pipeline #95483 passed with stages
in 6 minutes and 57 seconds
......@@ -29,6 +29,53 @@ The program **phoenix_hpc_proxy** can generates automatically a proxy library to
![proxy principle](doc/proxy_principle.png)
Program parameters :
```
Parameters :
--header=FILENAME , -i FILENAME : list of input headers to be used to describe the main library
This argument has to be set
--library=STRING , -l STRING : name of the library to be created
Default value : lib_shadok
This argument is optional
--libdir=DIRECTORY , -d DIRECTORY : directory where the sub libraries are installed
This argument is optional
--anyarch=FILENAME : file name of the library to be used for any architectures (kind of security if the other architectures are not found)
This argument is optional
--sse=FILENAME : file name of the library to be used for sse architecture
This argument is optional
--sse2=FILENAME : file name of the library to be used for sse2 architecture
This argument is optional
--ssse3=FILENAME : file name of the library to be used for ssse3 architecture
This argument is optional
--sse4.1=FILENAME : file name of the library to be used for sse4.1 architecture
This argument is optional
--sse4.2=FILENAME : file name of the library to be used for sse4.2 architecture
This argument is optional
--avx=FILENAME : file name of the library to be used for avx architecture
This argument is optional
--avx2=FILENAME : file name of the library to be used for avx2 architecture
This argument is optional
--avx512f=FILENAME : file name of the library to be used for avx512f architecture
This argument is optional
--avx512wb=FILENAME : file name of the library to be used for avx512wb architecture
This argument is optional
--output=STRING , -o STRING : Output directory
Default value : .
This argument is optional
```
# Performance example
The performance example is [here](TESTS/PERFORMANCE_TESTS)
The following figure shows performance between a **specialised hadamard product** and the **generated proxy hadamard product**.
We notice, proxy performance are always faster than specialised one, no matter the compilation option.
![proxy perf](doc/Performances/cmpHadamardProxyBaseElapsedTimeCyEl.png)
All the detailed performances are in [here](doc/Performances/)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment