You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The reason is due to finding as a last kernel a MemcpyHtoD,0x00007f2483b08a00,4 instead of a kernel-X.traceg. This behavior has been observed in parboil-bfs with an Ampere trace.
Hopefully, this can be fixed by replacing an if by: if ( (finished_kernel_uid || m_gpgpu_sim->cycle_insn_cta_max_hit() || !m_gpgpu_sim->active()) && !kernels_info.empty() )
However, I don't know if you are thinking in a different solution.
The text was updated successfully, but these errors were encountered:
I'm more curious about why the last kernel is MemcpyHtoD.
I'm worrying that maybe there was a kernel launch after that, but the tracer crashed and was unable to catch it. MemcpyHtoD to GPU without consuming the data sounds like a weird use case to me.
Thanks
sun-lingyu
added a commit
to sun-lingyu/accel-sim-framework
that referenced
this issue
Nov 13, 2024
The simulator is crashing with assert(k) in main.cc
The reason is due to finding as a last kernel a
MemcpyHtoD,0x00007f2483b08a00,4
instead of akernel-X.traceg
. This behavior has been observed inparboil-bfs
with an Ampere trace.Hopefully, this can be fixed by replacing an if by:
if ( (finished_kernel_uid || m_gpgpu_sim->cycle_insn_cta_max_hit() || !m_gpgpu_sim->active()) && !kernels_info.empty() )
However, I don't know if you are thinking in a different solution.
The text was updated successfully, but these errors were encountered: