Issues

hipMemcpyHostToDevice -> hipMemcpyAsyncDeviceToDevice
c3sr/comm_scope#47 · opened on Aug 7, 2023
PCIe A100 can deliver incorrect results when GPU is idle due to low clocks
c3sr/comm_scope#43 · opened on Mar 5, 2021
ThetaGPU: error in src/cudaMemcpyPeerAsync_Duplex_GPUGPUPeer: a PTX JIT compilation failed
c3sr/comm_scope#42 · opened on Jan 19, 2021
use cudaEvent to measure empty kernel time
c3sr/comm_scope#40 · opened on Sep 16, 2020
add NVSHMEM benchmarks
c3sr/comm_scope#39 · opened on Sep 7, 2020
Break benchmarks out into latency and bandwidth
c3sr/comm_scope#38 · opened on Sep 3, 2020
Make sure data is random to foil compression
c3sr/comm_scope#34 · opened on Jun 4, 2020
Any lessons to be learned from EasyPerf?
c3sr/comm_scope#29 · opened on Aug 26, 2019
1
Investigate using cudaLaunchHostFunc for getting wall time when a stream operation ends
c3sr/comm_scope#28 · opened on Apr 10, 2019
prefetch-duplex GPU/GPU may be able to associate both streams with a single device
c3sr/comm_scope#22 · opened on Apr 3, 2019
Anything we should learn from NVIDIA/multi-gpu-programming-models
c3sr/comm_scope#21 · opened on Feb 22, 2019
nvcc 7.5 has incompatibility with spdlog 0.16.3-p1
c3sr/comm_scope#17 · opened on Dec 19, 2018
add multi-threaded explicit transfer benchmarks
c3sr/comm_scope#16 · opened on Oct 11, 2018
1
Flush caches in unified memory host-to-gpu
c3sr/comm_scope#15 · opened on Oct 11, 2018