test our implementations against gcc_s #67
Conversation
if it exposes the same intrinsics that we implement -- gcc_s doesn't implement all the intrinsics for all the architectures. closes #65
|
@homunkulus try |
|
💔 Test failed - status-appveyor |
|
@homunkulus try |
1 similar comment
|
@homunkulus try |
|
💔 Test failed - status-appveyor |
|
@homunkulus try |
|
💔 Test failed - status-travis |
|
@homunkulus try |
|
💔 Test failed - status-travis |
|
re: the add*sf3 failures on hard float arm I think the gcc_s implementation of these intrinsics for hard float arm may be wrong (?). Our implementation vs native operations passes the unit tests and our implementation vs the gcc_s implementation of these intrinsics but for soft float arm also passes the unit tests ... So I don't know what else could explain these results. |
instead test half of the time against gcc_s and the other half test against the native operation (\*). (\*) Not all the targets have available a native version of the intrinsics under test. On those targets we'll end up testing our implementation against itself half of the time. This is not much of a problem because we do several quickcheck runs per intrinsic.
|
@homunkulus try |
|
💔 Test failed - status-appveyor |
we don't want to match musleabihf targets
|
The changes look good to me! It does seem like there could be a better solution to testing all of these intrinsics using some sort of custom testing harness, but that can wait for a later PR of course. I'm starting to brainstorm ideas for this. It'd be nice to have results like: ... Values made up, of course. |
|
Thanks for taking a look @mattico! I'm going to (@homunkulus r+) this. re: a custom test harness. Love the idea! In particular, benchmarking is something important that we are missing. Could you open an issue for it so it doesn't get lost? Thanks! |
|
@homunkulus r+ I said ... |
|
📌 Commit dafe47b has been approved by |
test our implementations against gcc_s if it exposes the same intrinsics that we implement -- gcc_s doesn't implement all the intrinsics for all the architectures. closes #65 r? @Amanieu Tested on Linux x86_64 and against the x86_64 and the arm-gnueabi targets. Unclear whether this works on osx or windows.
|
☀️ Test successful - status-appveyor, status-travis |
if it exposes the same intrinsics that we implement -- gcc_s doesn't
implement all the intrinsics for all the architectures.
closes #65
r? @Amanieu
Tested on Linux x86_64 and against the x86_64 and the arm-gnueabi targets. Unclear whether this works on osx or windows.