Loop-dedicated benchmark suites?This is a discussion on Loop-dedicated benchmark suites? within the Compilers forums in Theory and Concepts category; Hi, I want to evaluate the effectiveness of a research C source-to-source optimizer on benchmarks. Especially, I'm interested in the supported loop transformations. Which freely available testsuites contain a large number of loops? Are there possibly benchmark suites that focus on loops? One of the supported loop transformations I want to evaluate is loop unswitching. However, the benchmarks from the UTDSP testsuite I recently considered have no potential for this transformation, i.e. the loop structures do not allow to "unswitch" loops. Is this a normal situation for unswitching? Regards, Christian... | ||||||||
| | ||||||||
| ||||||||
| Register | FAQ | Calendar | Search | Today's Posts | Mark Forums Read |
|
#1
07-23-2008, 03:09 PM
| |||
| |||
 Loop-dedicated benchmark suites? Hi, I want to evaluate the effectiveness of a research C source-to-source optimizer on benchmarks. Especially, I'm interested in the supported loop transformations. Which freely available testsuites contain a large number of loops? Are there possibly benchmark suites that focus on loops? One of the supported loop transformations I want to evaluate is loop unswitching. However, the benchmarks from the UTDSP testsuite I recently considered have no potential for this transformation, i.e. the loop structures do not allow to "unswitch" loops. Is this a normal situation for unswitching? Regards, Christian  |
|
#2
07-30-2008, 08:12 AM
| |||
| |||
| Re: Loop-dedicated benchmark suites? The efficiency of source to source transformations highly depends on the underlying compiler, the compilation options used to generate the code, and on the experimented processor. You cannot have any general trend, even if you do thousands of experiments. |
 |
« Previous Thread
|
Next Thread »
| Thread Tools | |
| Display Modes | |
Linear Mode
All times are GMT -5. The time now is 06:04 AM.
