[macstl-dev] Note about the YDL benchmarks

Ilya Lipovsky lipovsky at skycomputers.com
Fri Jul 8 04:55:40 WST 2005

  • Previous message: [macstl-dev] not all rosey in the gcc-4.0.0 land
  • Next message: [macstl-dev] Question about sized booleans
  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]


It looks like certain older versions of PPC Linux kernel do not have a 
proper handler to deal with the Java-mode Altivec exception. Version 
2.4.23 of the kernel does not have this problem, so I *assume* that all 
versions that are younger are fine. For versions that are older there are 
no guarantees: in fact, I know that 2.4.22 kicks the programs out of 
execution with the "Trace/BPT trap" message. So you have to be aware of 
those things in order to not feel lost, if you get your benchmarks stopped 
with a message like the one above. Unless you're an IEE754 freak, feel 
free to disable the setting of the Altivec Java-mode bit in the MacSTL's 
benchmark code as a temporary workaround. Happy benching!

-Ilya



  • Previous message: [macstl-dev] not all rosey in the gcc-4.0.0 land
  • Next message: [macstl-dev] Question about sized booleans
  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

More information about the macstl-dev mailing list