[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: 440FX PCI CHIPSET performance
- To: Mailing List Recipients <pci-sig-request@znyx.com>
- Subject: RE: 440FX PCI CHIPSET performance
- From: Graeme Gill <graeme@digideas.com.au>
- Date: Fri, 27 Jun 1997 13:34:35 +1000
- Resent-Date: Fri, 27 Jun 1997 13:34:35 +1000
- Resent-From: pci-sig-request@znyx.com
- Resent-Message-Id: <"W8LiD.0.7E.T8qip"@dart>
- Resent-Sender: pci-sig-request@znyx.com
Dave Feng <dfeng@iced.com> wrote:
> If you are using the AMCC S5933, there is an undocumented bit that
> enables the PCI read line/multiple cycle. If you don't use this
> explicitly, depending on the motherboard chipset, you may or may not be
> bursting more than a cache line.
No, I'm using the 9060SD, and already using MRM and MWI.
> My experience is the same as you guys with the 440FX. I was getting
> sustain of 55(112 peak) for writes and 30(60 peak) for reads. On the
> 430TX I was getting 55(same) for writes but 64(> 2x) for reads.
The problem I explained is that I haven't been able to get anywhere
near 55 Mb/sec writes or 30 Mb/sec reads sustained using the 440FX.
The best I have seen sustained is less than 10 Mb/sec.
> Once I enabled the hidden bit the 440FX behaved much like the 430TX.
> So(drum roll please), the bit you need to enable is bit 15 of the MCSR.
I can't find any register in the 440FX called MCSR.
Graeme Gill
Ö € o