Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Cudastuff #6

Open
wants to merge 38 commits into
base: master
Choose a base branch
from
Open

Cudastuff #6

wants to merge 38 commits into from

Conversation

drossetti
Copy link
Owner

dummy PR, just to ease browsing

drossetti and others added 30 commits April 12, 2018 09:58
@drossetti
update to version in MOFED 4.2-1.2.0.0, perftest-4.2-0.2.g4f56894.4212
07a9f94
@drossetti
WAR bug: don't try to initialize GPU memory
3277b33
@drossetti
don't parse --use_cuda if HAVE_CUDA is not defined
9d77bb6
@drossetti
add help string for CUDA_H_PATH configure var
e1d3299
@drossetti
protect use_cuda member declaration
c700929
@drossetti
add comment for use_cuda case
f022bce
@drossetti
enable testing of CUDA Unified Memory: introduce --use_cuda_um param
307c5bf
@sjeaugey @drossetti
Enable latency benchmarks with CUDA. Not working with ib_write_lat.
029d4d4
@drossetti
print when --odp is passed
a8baddf
@drossetti
dump GPU BDF infos, fix use of GPU id instead of CUdevice
12342a0
@drossetti
pad BDF appropriately
d6dec93 
add error checking for cuMemFree
@drossetti
fix !CUDA build
59e2f14
@drossetti
pre-populate UM on either CPU or GPU memories
d7d2803
@drossetti
warn when using ODP on GPU device memory
f541ee0
@drossetti
do not consider HAVE_EX_ODP/HAVE_EXP_ODP for setting ODP flags.
c60f176 
patch from Huy Nguyen <[email protected]>
@drossetti
add affinity selection to --use_cuda_um param
99d4397
@drossetti
fix warning in printf
c6da118
@drossetti
add error checking to cuMemAdvise & cuMemPrefetchAsync calls
fd3ea56
@drossetti
replace managed memory options --use_cuda_um with --cuda_mem_type, in…
76031f1 
…troduce --cuda_mem_hints
@drossetti
remove pid and modify printing of GPU info
d2f8295
@drossetti
replace CPU with host in usage
530fa45
@drossetti
multiple fixes mainly for gpu hints
5c56523
@drossetti
add --cuda_use_gpu parameter to select a particular GPU by ordinal
a0ed0b6
@drossetti
bug fix: use DIRECT_MANAGED_MEM_ACCESS_FROM_HOST rather than CONCURRE…
ad41028 
…NT_MANAGED_ACCESS as proxy to ATS
@drossetti
add support for --cuda_mem_hints=3 (read-mostly)
86e6917
@drossetti
add some more CUDA related details to usage()
c1bfd91
@drossetti
forbid hints for HOSTREGISTER memory. add memory initialization calls…
591194a 
…. use separate error string for the different pp_init_ functions.
@drossetti
return an error when setting hints on unsupport memory types
0c1f775
@drossetti
report error if memory hints are non default on unsupported types
bfdd14f
@drossetti
remove ODP enabled message
a831136
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@drossetti @sjeaugey