> > No disadvanages of GX over BX. GX is a superior chipset.
>
> Specifically the GX seems to benchmark at 2x-4x the memory throughput of
> the BX. I'd definatly recommend getting it over BX, where available.
what benchmark are you referrring to? Stream is pretty much the normal
one (or perhaps the lmbench memory components). here's a pretty normal
Stream result from a BX board:
Function Rate (MB/s) RMS time Min time Max time
Assignment: 318.5662 0.0503 0.0502 0.0508
Scaling : 323.1670 0.0496 0.0495 0.0497
Summing : 400.5210 0.0600 0.0599 0.0600
SAXPYing : 292.8187 0.0820 0.0820 0.0820
that's for a celeron 300/450, bx, cas2 pc100.
the reason I ask is that GX does many-way interleaving, but it's not
clear that this does more than make up for the poor bandwidth of EDO.
the issue is whether the P6 bus can _ever_ beat half its peak rate
(800 MB/s). Stream results for various clock speeds of PII's indicate
that 400 MB/s is about as far as it goes.
thanks, mark hahn.
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
