Date: Tue, 01 Oct 1996 22:47:16 -0700 From: "Michael L. VanLoon -- HeadCandy.com" <michaelv@MindBender.serv.net> To: BRETT_GLASS@infoworld.com Cc: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com>, narvi@haldjas.folklore.ee, obrien@nuxi.cs.ucdavis.edu, freebsd-hardware@freebsd.org Subject: Re: H/W recommendation Message-ID: <199610020547.WAA21553@MindBender.serv.net> In-Reply-To: Your message of Tue, 01 Oct 96 19:15:27 -0800. <9609018442.AA844219269@ccgate.infoworld.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>> As can be seen the best parts to be using are the 100, 133, 166 and 200,
>> with the exception that at a mulitplier of 3 the CPU starves for memory.
>Depends on the cache, and whether you get a Pentium or a Pentium Pro.
Please don't group these together, because they don't belong in the
same category when talking about this.
It doesn't much depend on the cache, with the Pentium, because at 3x
the bus speed, it's going to saturate the bus no matter how much cache
you have. Because the cache is on the far side of the bus. No matter
how much or how fast your cache is, you still are only running it at
~66MHz (75MHz in some of the Cyrix motherboards -- which isn't
"standard" [yet?]).
The Pentium Pro is a whole different ball game precisely because of
its on-chip cache. The cache speed runs at the CPU speed, which means
its always a single-cycle access, no matter how fast the CPU. This
was done specifically to address this problem. Otherwise, the P6
wouldn't be able to scale much higher than 200MHz (with a ~66MHz
memory bus, anyway), either. Coupled with the faster cache is the
fact that it is non-blocking at up to four cache misses.
>A
>Pentium Pro with a built-in 512 KB level 2 cache usually won't starve, even
>on UNIX boxes. (To put things in perspective, a typical FreeBSD kernel,
>with unnecessary drivers removed, is about that size.)
I think you need to do some more detailed testing. It *can* starve,
given the right mix of work to do. Even with 16MB of cache. Granted,
the misses won't take as long to fill, but it will still stall.
>But the bargain
>basement version of the Pentium Pro, with the 256 KB cache, will drag in
>the same configuration. Unfortunately, far too many clone vendors just
>HAPPEN not to mention in their ads that they're including the cheaper CPU.
Do you have any proof of this? I think you're speculating.
And, I don't know how you can call the 256K P6 the "bargain basement"
when it costs as much as the *highest* speed Pentium. Plus the fact
that 512K P6s are *twice* as expensive, and hard to get. I would say
the 512K model is *definitely* a _premium_ model. But if you like
paying $1200 for your processors, you can think of them any way you
like. :-)
Besides, as already pointed out above, the 256K Pentium Pro cache is
already *way* faster than any size cache on a Pentium. Plus, with a
256K cache you're already taking about hits in the 90+% range. 512K
would be nice, but its definitely at the far side of diminishing
returns. Where the 512K model *would* be very valuable is in a
four-way SMP box, since you start getting a lot of bus contention in
that kind of environment.
>I'd like to see a megabyte cache on board.
I would too. But I have better things to spend my money on, first.
-----------------------------------------------------------------------------
Michael L. VanLoon michaelv@MindBender.serv.net
--< Free your mind and your machine -- NetBSD free un*x >--
NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3,
Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32...
NetBSD ports in progress: PICA, others...
-----------------------------------------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199610020547.WAA21553>