Re: [m5-dev] libm5_* is huge

2009-02-02 Thread Steve Reinhardt
On Sun, Feb 1, 2009 at 2:13 PM, nathan binkert n...@binkert.org wrote: It'd also be easy for us to just not use libm5_foo.a as an intermediate step to m5.foo. We could easily just skip the archive step unless you explicitly ask for it. I'd be in favor of this. Steve

[m5-dev] libm5_* is huge

2009-02-01 Thread Gabe Black
I just noticed that while m5.debug is 54MB, but libm5_debug.a 197MB. Do we know why it's so huge? That -can't- be good for performance. Gabe ___ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev

Re: [m5-dev] libm5_* is huge

2009-02-01 Thread nathan binkert
I just noticed that while m5.debug is 54MB, but libm5_debug.a 197MB. Do we know why it's so huge? That -can't- be good for performance. I'm pretty sure it's because there are multiple copies of many functions in there. When C++ sees a template class member function, it emits code for it in