Closed jaycedowell closed 6 years ago
This approach to building the unpacking and quantization modules works on ADP but it seems to cause problems with travis. Is there a more robust way to deal with this that doesn't lead to maintaining two sets of files?
Would passing "-x c++" to g++ help?
Alternatively, can you move the cuda code into a separate .cu + .hpp and guard calls to it from the .cpp with ifndef NOCUDA?
I had something similar to the "-x c++" before that did work right either. Maybe I should revisit this approach before splitting things up.
What's up the these travis builds? The "details" link indicates that they have finished but here they are listed as still pending.
Merging #103 into master will increase coverage by
0.6%
. The diff coverage isn/a
.
@@ Coverage Diff @@
## master #103 +/- ##
=========================================
+ Coverage 67.17% 67.77% +0.6%
=========================================
Files 39 39
Lines 3555 3643 +88
=========================================
+ Hits 2388 2469 +81
- Misses 1167 1174 +7
Impacted Files | Coverage Δ | |
---|---|---|
python/bifrost/blocks/serialize.py | 91.27% <0%> (+1.11%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 8489177...1dc3f2f. Read the comment docs.
I clicked "restart build" just now and it passed. Not sure what that was - it could be because I was editing the travis settings at the same time as the build.
Add support for GPU-based unpacking and quantization (third attempt). This includes: