Date: Wed, 12 Feb 97 13:17:48 -0600 From: Ben Black <black@gage.com> To: "Jonathan M. Bresler" <jmb@freefall.freebsd.org> Cc: mcwong@hotmail.com (M.C Wong), questions@freebsd.org Subject: Re: UltraSPARC and MicroSPARC vs Pentium Pro ? Message-ID: <9702121917.AA22714@squid.gage.com> In-Reply-To: <199702121833.KAA18506@freefall.freebsd.org> References: <199702121833.KAA18506@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
>quick dirty answer:
> Integer: the intel boxes kill the snot out of *all* suns
> Float: the ultras outperform intel boxes.
see numbers below which indicate that a P6-200 slightly outperforms a 200MHz
UltraSPARC in integer, while the UltraSPARC actually does "kill the snot"
out of the P6 in floating point.
>long answer:
> the sparc architecture is limited in its ability to perform
> integer operations. my suspicion is that the memory bandwidth
> is not up to the task. (surely, its not the cpu itself, but
> rather feeding data and instructions to the cpu that is the
> limiting factor.)
thank you, mr. wizard. unfortunately, completely wrong. the UltraSPARC
machines all use the UPA crossbar switch which gives a bandwidth of over
1GB/s. compare that to a typical PC bus which is usually around 400MB/s (and
is not switched, as the UPA is). quite simply, current mass produced Intel
boards just can't compete for I/O bandwidth.
if memory bandwidth *were* the problem, you'd expect it to show up in the
floating point benchmarks.
as for "killing the snot out of Sun" in integer performance, the SPECint95
numbers just don't show it:
SPECint95 SPECfp95
Intel Alder 200MHz P6 8.09 6.75
Sun Ultra 2 1200 200MHz 7.72 11.1
if i get a chance, i will run the HiNT benchmarks on an UltraSPARC and a P6
here. i doubt i will see any snot flying from the Sun.
benchmarks, of course, can always say what you want them to.
b3n
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9702121917.AA22714>