Mediocre Fasttrak Performance?

[Brian Ellis]

Out of curiosity, I checked what difference the S.M.A.R.T feature makes: exactly 1 Mb/s for both seq read and write and 3 Mb/s for the two buffered speeds. Not much to write home about!

[Brian Ellis]

Further tests:
I disabled the vcache. Windows hogged 331 Mb (out of 384 Mb) for current disk cache, instead of my usual 30 Mb. As expected, performance dropped, but only on write. The seq read remained the same, pitiful, 19 Mb/s, while the seq write dropped from 43 to 26 Mb/s. The strange thing is that the disk access whizzed up to a stunning 4 ms.

There's one thing I don't understand, though: I can't get the FastCheck utility to stick through reboots. If I disable S.M.A.R.T Check, it comes up again each time, even though the Start minimized box is checked. It doesn't appear in the SysTray, until I invoke it manually. There is nothing relevant in the StartUp menu, nor can I see anything specific (quick glance) in any of the ini files.

[Dr Mordrid]

When you installed FastCheck did you mark the option for it to start at bootup?

I find your results very odd and unlike any system I have here. Something in your system must be amiss. Presuming a 2 drive array are they on separate channels and not on the same cable?

Dr. Mordrid

[Brian Ellis]

Doc

Yes, I can assure you that the array is correctly installed, using the two cables that came with the Promise card (I haven't tried the Maxtor ones).

I admit that I can't remember how I installed the software: it was 3 or 4 weeks ago that I did it. However, although it does make a small difference, it is just that, small, and doesn't tackle the basic problem. The one thing I haven't tried yet is to attack the system BIOS Latency setting, as mentioned in the manual (if there is an adjustment). I might try uninstalling the caboodle and starting again.

[Brian Ellis]

Unfortunately, the PCI Latency is not adjustable on my system. I've tweaked one or two minor things in the BIOS and have gained another 1 Mb/s on both seq settings. But it still ain't good enough.

[Brian Ellis]

Thank the Pope! I've now swapped m/bs and on the default BIOS settings the Seq read and write have shot up to 37 and 47 Mb/s resp.

This is funny, because the old board was super fast with video stuff, considering it's a PIII/450.

My Snailtrack is now Fastrack: hurrah!

[Dr Mordrid]

GREAT!!. Welcome to the speed demons club

Bravo-Baby mainboard?? Sounds like something that might have originated at PC-Chips. If so.....

Dr. Modrid

[Brian Ellis]

Doc

I must admit I'm still flummoxed as to why the Bravo Baby performed so badly in this respect, when it was good in other respects, including faster writing directly to UDMA disks than through the Fastrack. I removed every card, the D: drive, the CD-ROM and the DVD drives, including the Marvel and replaced that with a different AGP 2x and it was still atrocious. As the RAID was then the only thing on the PCI bus and this was driven by a standard 443BX chipset, I can assume only that the PCI to ISA bridge was the culprit.

Incidentally, one of the things I tried was to master/slave the two drives on the same Promise port. This made literally zero difference from 2x master on 2 ports.

Anyway, BB is now in my office computer, working very well, with Marvel as graphics card (no video drivers installed), receiving this message as I type it The only thing I haven't yet tried is voice recognition, as I'll probably have to re-enroll it, as the sound card is different.