[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: PLX 9060SD
- To: Mailing List Recipients <pci-sig-request@znyx.com>
- Subject: Re: PLX 9060SD
- From: holeman@austx.tandem.com (Jim Holeman)
- Date: Thu, 9 Jan 97 10:34:49 CST
- Resent-Date: Thu, 9 Jan 97 10:34:49 CST
- Resent-From: pci-sig-request@znyx.com
- Resent-Message-Id: <"fAG2L1.0.g4.HwHro"@dart>
- Resent-Sender: pci-sig-request@znyx.com
> Steve Deiss wrote:
>
> >> Anyone tried the 9060SD with insights to share?
> >> Has the DMA channel throughput been measured on
> >> Pentium and Pro motherboards for data moving
> >> from target to host? Know where I can find a
> >> current errata sheet? Got a favorite gripe
> >> about the chp?
Adam Barnes replied:
>
> We have a lot of experience with 9060 and some
> with the SD. As far as I know the differences between
> the parts are very minor in the DMA area (apart from 1
> channel vs. 2).
>
> Anyway, we have seen some very good results both
> between 9060s (>100Mbyte/sec aggregate b/w sustained
> with both DMAing to each other at the same time),
> and from 9060 to host memory. On a Triton 430 FX
> M/B with a P133 + 256k PB cache + 16Mb EDO
> I saw 110-120Mbyte/sec sustained. The trick seemed
> to be programming a large value into the latency register.
> This caused the 9060 to stick around after losing GNT#
> so that it could do a nice long burst. I saw plenty of
> cases where the burst length hit 256 words. The local
> side was 0ws interface to a FIFO which was being
> pumped with data from a SHARC DSP chip.
>
1) I thought it was a no-no for the driver SW to reprogram the latency
register!
2) Anybody have any experience with Pentium Pro's in this same scenario?
holeman
____________________________________________________________________________
Jim Holeman Tandem Computers, Inc.
(512) 432-8755 (fax 8247) 14231 Tandem Boulevard
holeman@austx.tandem.com Austin, Tx 78728-6699
× õ