Chapel Nightly Performance Graphs
The following graphs are used by the Chapel development team in order to track Chapel performance over time. It should be noted that the graphs show changes not only in the Chapel compiler, runtime, and libraries, but also to the tests themselves, and the systems on which the tests are run. To the extent possible, the graphs contain annotations indicating causes of nontrivial changes.
Since Chapel is open-source, community members are encouraged to submit their own Chapel performance tests to be included in these graphs. For an introduction to what is needed to submit a new performance test, see our GitHub How to Contribute as well as the developer's documentation for creating Chapel performance tests.
Execution-Time/Performance:
chapcs is a 64-bit Linux machine with 2 12-core (24 HT) Intel Xeon processors for a total of 24 cores (48 threads) and 256 GB of ram. It runs our single-locale performance suite.
16-node-xc is 16 nodes of a Cray XC that runs our multi-locale performance suite for various communication and tasking layer configurations.
chapcs release-over-release compares the compiler's current performance with that of previous releases run on recent sources and hardware. This permits us to compare how performance has changed over time, factoring out changes to the test itself and the underlying system and software. Since the historical timings are run manually and infrequently, they may not always be 100% up-to-date.
comm-counts runs communication count tests to track the number of communication events required for various benchmarks.
no-local runs our single-locale performance suite with --no-local on a chapcs machine
llvm runs our single-locale performance suite with --llvm on a chapcs machine
clang runs our single-locale performance suite with clang as the backend C compiler on a chapcs machine
1-node-xc runs our single-locale performance suite with intel, cray, and gcc backend C compilers on a single node of a Cray XC.
numa runs our single-locale performance suite with the numa locale model on a chapcs machine
chap01 is a 64 bit Linux machine with 2 quad core (8 HT) Intel Xeon processors for a total of 8 cores (16 threads) and 48 GB of ram. It runs the full test suite with default settings. This reflects compilation times for single-locale programs with Chapel's default settings.
chap08 is a 64 bit Linux machine with 2 quad core (8 HT) Intel Xeon processors for a total of 8 cores (16 threads) and 96 GB of ram. It runs the full test suite with --no-local thrown. This reflects compilation times for multi-locale programs with Chapel's default settings.
chap12 is a 64 bit Linux machine with 2 quad core (8 HT) Intel Xeon processors for a total of 8 cores (16 threads) and 96 GB of ram. It runs the full test suite with --fast thrown. This reflects compilation times for single-locale programs with all optimizations (including back-end) turned on.
chap04 is a 64 bit Linux machine with 2 quad core (8 HT) Intel Xeon processors for a total of 8 cores (16 threads) and 48 GB of ram.