some profiling problems
Hi, I 'm trying to profile load efficiency of MapGraph by using nvprof --metrics branch_efficiency there is a warning: "Warning: No CUDA application was profiled, exiting". I do not know why this is happening. I used the Redhat 6.5 OS and CUDA 7.5 on K20M GPU card. Thanks for your kindly response.
George, We are no longer supporting that project. Thanks for your understanding....
CsrProblem()~GraphSlice
How to increase the no of iterations in Pagerank
Mapgraph 0.3.3 Release
Update for 0.3.3 release for minor compilation ...
Update to resolve compilation error introduced ...
Mapgraph moves to Git
v3 release.
Modified top-level make file and ufl.mk to supp...
remove enum
clang fix
Removing BFS executable that had been checked i...
warning fix
template error on Mac test
add modern GPU
mgpu version added.
Modify threshold for performance
add gather_out, gather_all, expand_in, expand_all
add b40c
Algorithms: SSSP, BFS, CC, PR
added more comments
add edge list
added Makefile
add new API
added the small regression graphs. they are ref...
add CPU implementation and correctness test for...
change PageRank executable so that it has the s...
Add CPU implementation of BFS
Creating a branch for the new API. The files i...
add pagerank correctness check
updating the changelog in preparation for a new...
Modify usage information
Add top level Makefile
Change the organization of the algorithm files
bug fix for gather stage
added SSSP new implementation based on Merrill'...
add CC CPU implementation and correctness test
add new API
change options for Pagerank
Adding some scripts as a starting point for bat...
Added a parameterized Makefile that includes a ...
updated license headers.
delete old version
Adding code for wave
refactor
Bug fix for wave: message size error
assigned each device to one rank and generate g...
rmat generation and partitioning
wave/frontier broadcasts
Changes to match the changes in generator
testing code
using GPUDirect, improved stats
Adding the Generator and Partitioner
bitmap_in to frontier
Convert extended bitmap to regular bitmap
gernate compace bitmap for output frontier for ...
bit operations for MPI
refactor done and tested. One Harish is done, w...
Add or correct copyright headers for MULTI_GPU ...
wave performance fixes and timers added
Added support to read flat files from generator
Bug Fix for file read
user can specify warmup times
Creating branch for Multi-GPU work.
GPU bit operations
debug
Add per iteration stat
Catching up the multi-GPU version with the trunk.
Fixed some issues in Makefiles and generator is...
check frontier size to make sure it is not empty
bug fix in wave
usr can choose to compress msg or not
local starting vertex fix
Generator Bugfix + frontier size + Pseudo-rando...
timing for wave with compression
compressed size added
correct test
added wait time and wave removed in last iteration
custom warmup number
local predecessor computed
compute number of traversed edges
bitunion time and optimization for wave: remove...
Frontier Size computation
wave timing fix
Fix for validation
1-based pred indices