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

mfence should clobber memory #26

Closed cwpearson closed 5 years ago

cwpearson commented 5 years ago

https://github.com/c3sr/comm_scope/blob/48e1ce32b3f0f45faa68fda81b551f4f4822feab/src/utils/cache_control.hpp#L61

From the linux kernel https://github.com/torvalds/linux/blob/a978a5b8d83f795e107a2ff759b28643739be70e/tools/arch/x86/include/asm/barrier.h#L24

#elif defined(__x86_64__)
#define mb() asm volatile("mfence":::"memory")
cwpearson commented 5 years ago

fixed in 004d90b00f6c65a7d143b3c87c507235bcc677a7