UPC Operations Microbenchmarking Suite
V1.1 November 2010
----------------------------------------------------
Damian Alvarez Mallon:
Added benchmarking of upc_forall statement (read, write, read+write) of a shared array
Added benchmarking of for statement (read, write, read+write) of a shared array
Dynamic threads environment is now supported
Added an option to set the maximum run time for each size
V1.0 March/April/May 2010
----------------------------------------------------
Damian Alvarez Mallon:
Added benchmarking of upc_all_alloc and upc_free
Added benchmarking of all the reduce and prefix reduce operations
Bug fix in reduce operations
Now the number of bytes indicates the block size rather than the problem size
Cache invalidation switchable at runtime
Warmup iteration switchable at runtime
User selectable message sizes
User selectable minimum and maximum sizes
User selectable list of benchmarks
User selectable synchronization mode
Help flag added
Version flag added
Code reorganization
Number of iterations now depends on THREADS and cache invalidation
Added documentation in PDF
Changed the "times" variable name in UOMS.upc. It seems to interfere with the math library in some platforms
Removed the <upc_strict.h> header
Minor changes in the manual
V1.0rc1 November 2009
----------------------------------------------------
Damian Alvarez Mallon:
Code cleanup and reorganization.
Automatic number of cores detection.
Local operations included for all the core affinity setups.
Barrier added after timming synchronization
V0.3 March 2009
----------------------------------------------------
Damian Alvarez Mallon:
UPC asynchronous memory copy operations test.
V0.2 December 2008
----------------------------------------------------
Damian Alvarez Mallon:
Efficient Timers for IA64
Switch to nanoseconds timers
V0.1 November 2008
----------------------------------------------------
Guillermo Lopez Taboada:
Initial release for internal use.
Additional contributions by Carlos Teijeiro and Jose Manuel Andion.
|
