An HTML attachment was scrubbed... URL: http://sos.ece.cmu.edu/pipermail/simflex/attachments/20060328/f7ebbd10/attachment.html From twenisch at ece.cmu.edu Tue Mar 28 16:12:41 2006 From: twenisch at ece.cmu.edu (Thomas Wenisch) List-Post: [email protected] Date: Tue Mar 28 16:12:46 2006 Subject: [Simflex] Re: A question regarding to SPLASH2 In-Reply-To: <[email protected]> References: <[email protected]> Message-ID: <[email protected]>
On Tue, 28 Mar 2006, lu peng wrote: > > > > >Second, you may want to try modifying the freq_mhz setting of each > >CPU in your checkpoint (you can do this by manually editing the > >Simics checkpoints with a text editor). Lowering this frequency > >(e.g. to 75MHz) will reduce the distance between timer interrupts, > >which will make Solaris get around to scheduling your apps sooner > >(on the other hand, it also increases the frequency of timer > >interrupts, which slow ocean down slightly). > > So you mean that I should try higher frequency? If so, simply change the > "freq_mhz: 75" to "freq_mhz: 1000" is enough? Or do I need reboot the target > machine with > higher frequency? > I was suggesting lowering it to 75, but you can try experimenting with this setting - we have found that it can have a significant impact on workloads that do I/O. However, it shouldn't really affect ocean much, except for right at the start (while the shell in the simulated machine is still loading the binary). We have found that it is generally not neccessary to reboot the machine - the setting takes effect by the next timer interrupt (within hundreds of thousands of cycles). Perhaps you can try single-stepping the ocean code in the simulator for a while in the to see if it "looks" like OS code (all the virtual addresses will be big) or like ocean (low values for PC, probably only the V8 subset of the V9 instruction set, so no unusual opcodes). The real question we're trying to get at is whether your simulation is actually measuring ocean or not. > > > > > >Regards, > >-Tom Wenisch > >Computer Architecture Lab > >Carnegie Mellon University > > >
