Dockerizing Linux-Bench Q2 2017

[Patrick]

Spent some time doing this today making some notes on this.

Overall Status
Components: 13 Usable, 2 Running Regressions, 1 Not Running (may exclude)
Automated Upload: Not Started
Automated Parsing: Not Started

c-ray 1.1 Updated and Expanded
Build Status: Running
Usage Status: Usable

Kernel Compile Benchmark
Build Status: Running
Usage Status: Usable

Notes:
-e iterats=x where x is number of iterations. Will default to 1 if not set
-e numthreads=y where y is number of threads. Will default to $nproc if not set

Redis Legacy
Build Status: Running
Usage Status: Usable

Redis Updated and Expanded
Build Status: Running
Usage Status: Usable

GROMACS w/ AVX-512
Build Status: Running
Usage Status: Running Regressions

NAMD
Build Status: Running
Usage Status: Usable

UnixBench
Build Status: Running
Usage Status: Usable

Chess
Build Status: Not Running
Usage Status: N/A

Sysbench
Build Status: Running
Usage Status: Usable

hardinfo
Build Status: Running
Usage Status: Usable

STREAM
Build Status: Running
Usage Status: Usable

OpenSSL Legacy
Build Status: Running
Usage Status: Usable

OpenSSL Updated and Expanded
Build Status: Running
Usage Status: Usable

7-Zip
Build Status: Running
Usage Status: Usable

Elasticsearch Benchmarks
Build Status: Running
Usage Status: Running Regressions

 

[Jeggs101]

What about 7-zip?

 

[MiniKnight]

Is this going to be the path forward for Linux-bench? Are these images uploading results?

Why are you using docker not sticking with bare metal?

 

[Patrick]

7-zip and NAMD updated. NAMD had to make a slight config tweak for.

 

[Patrick]

Updated the status on this. @Patriot has tested the build without elasticsearch.

Likely will need help on building a new upload, log parsing, and viewer functionality.

Still working on coremark and SPEC CPU 2017 but both have special license concerns.

 

[Patriot]

Just running it on a R7 1700 @ 3.9ghz with 3066 ram.

Might bring something old and Opteron back to life for some Epyc comparisons

 

[Patrick]

Here is the next question:

One log file or multiple log files for this?

Trying to figure out what is the easiest to parse later. Output is now tens of thousands of lines per run. There are delimiters in the log between different tests.

Looks like the next step is getting an automated upload and parsing system setup.

 

[Patrick]

Just wanted to say thank you to @Patriot for doing a full run.... just need to finish the new GROMACS tests...

For those who have not heard, the 4P 8180 system handled so many cores/ threads that we needed to move to multiple test difficulties like we did with c-ray.