search

c3sr / comm_scope

NUMA-aware multi-CPU multi-GPU data transfer benchmarks
https://github.com/c3sr/scope
Apache License 2.0
21 stars 3 forks source link

Make sure data is random to foil compression #34

Open cwpearson opened 4 years ago

cwpearson commented 4 years ago

Some allocations are zeroed right now. We should instead fill them with random data to prevent any automatic GPU memory compression.