jaycedowell
jaycedowell
This PR provides a unified packet capture interface that makes it easier to add new packet formats to Bifrost and to read packet formats from the network (unicast or multicast)...
This PR adds a new mmap()'d space, named "mapped", for Bifrost rings. This will allow for larger disk-based rings for buffering data.
This PR addresses recent (CUDA 10.1+) problems with the FFT test suite. These problems turned out to be caused by two things: 1. There was a change to cuFFT that...
This PR cleans up the code base by removing unused and commented out code. It may also be an appropriate place to address #156.
This PR enables support for the `astype()` method on `bifrost.ndarray`'s. This was previously disabled because of problems with arrays that were in the "cuda" space. As implmented: * If an...
This PR addresses #155 where `bifrost.reduce()` throws a BF_STATUS_DEVICE_ERROR when given a slice. This is done my making sure that the input data/slice is aligned along a vector-size boundary before...
@telegraphic has noted that both `bifrost.block` and `bifrost.pipeline` provide the same functionality. There are also several other places in Bifrost where this seems to happen: * DataType.py vs. dtype.py *...
A couple of times now I have run into problems passing data between blocks using CUDA-based rings. If I don't force a `bifrost.device.synchronize_stream()` within the reserve context for the ring,...
We upgraded ADP to Ubuntu 18 (gcc 7.4.0), CUDA 10.2, and libvma 8.9.5 and started running into problems with packet capture. Specifically, the first packet captured with Bifrost compiled with...
I've recently started seeing failures on the self-hosted Python 3.8 test cases. They seem to be from a problem building/installing `pycuda`. On master I see: ``` Building wheels for collected...