Date: Sat, 07 Nov 1998 19:30:47 -0800 From: Mike Smith <mike@smith.net.au> To: Peter Wemm <peter@netplex.com.au> Cc: Bruce Evans <bde@zeta.org.au>, jc@irbs.com, mike@smith.net.au, current@FreeBSD.ORG, smp@FreeBSD.ORG Subject: Re: Dog Sloooow SMP Message-ID: <199811080330.TAA01507@dingo.cdrom.com> In-Reply-To: Your message of "Sun, 08 Nov 1998 11:25:48 %2B0800." <199811080325.LAA22362@spinner.netplex.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
> Bruce Evans wrote: > > >No idea. I've received verification that fixing this for all 686-class > > >CPUs seems to work (ie. it's OK on the Cyrix MII and doesn't appear to > > >impact performance there), so the tests are now generalised for the > > >entire 686-class. > > > > It's only OK for MII's because of various `#if 0's and `#ifdef SMP's > > that prevent non-OK code from running on MII's. > > I think it should be CPU specific, not cpu class specific. The > model-specific-registers are very specific to the Intel family. I'd be a > lot happier if it was 'if (cpu == CPU_686 || cpu == CPU_PII) ...' Of > course, feature tests would be better. 'if (cpu_features & CF_PPRO_MSR)...' > The problem is that there is a 'cpu_feature' already for the CPUID. We > need more general flags than what Intel choose to tell us. *shrug* If you have better documentation for what should and shouldn't be based on the CPU class vs. CPU functionality, please illuminate us poor mortals. Meanwhile I'm simply trying to get back some of the performance that seems to have been lost; I'll pull it back to two comparisons against the P6 and PII if that's considered safer. -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199811080330.TAA01507>