Page 1 of 2

Elite Group (ECS) si54p aio

Posted: Fri Jan 24, 2003 10:30 pm
by schvankus
A friend of mine asked me to upgrade his pc. At the time it had a 850MB HDD that had been repeatedly compressed, 16MB parity ram and running 95B. I thought things could definitely run better so I agreed to help.

After installing 128MB ram and a 40GB HDD (partitioned into 2 5GB and 3 10GB partitions) with OnTrack's DDO, the computer STILL runs poorly. For example, when I go into Windows Explorer I see the contents of my C: drive displayed. But when I click on one of the partitions (most are nearly empty), it literally takes 15 seconds to display its contents.

I called Seagate tech support and they suggested it might be the IDE drivers. I went to the motherboard manufacturer's websites:

http://www.ecsusa.com/downloads/si54p.html
http://www.ecs.com.tw/download/si54p.htm

There are no drivers I could find at the sites, but I did find a BIOS upgrade. I downloaded it but haven't flashed the BIOS yet since I'm not certain it's for the correct PCB. The current BIOS version is 2.1 dated 9/1/95. I downloaded the 2.1i. Does this sound like the correct upgrade?

These may sound like stupid questions, but I'd hate to destroy my friend's pc!

What I'm trying to determine is if it's possible to just be stuck with a poor performing mb? I've always thought that there has to be a logical reason why something is occurring, but I just can't figure it out here.

So, do you think this is the correct BIOS upgrade and might it solve the IDE problem if that's indeed what it is?

Thanks so much for your help.

Posted: Fri Jan 24, 2003 11:24 pm
by NickS
It may help if you can post the BIOS ID and the OEM message. Thanks for posting the links, that's always helpful.
but I'd hate to destroy my friend's pc!
I don't know, you could be doing him a favour! You can get a complete base unit with a P200MMX for under 50 pounds UK....

During the 15 seconds it takes to show the contents, is the machine just sitting there, is the HDD light off, on solid or flashing ?

Posted: Sat Jan 25, 2003 6:01 am
by schvankus
This is the result of a BIOS ID utility:

Program: Unicore BIOS Wizard Version 1.5
Program: Unicore Chip Detect v0.72 2000.03.10
BIOS Date: 09/01/95
BIOS Type: PhoenixBIOS Version 4.04 PLUG and PLAY
BIOS ID: unknown
Chipset: SiS 501 rev 0
Superio: SMC 665GT rev 2 found at port 3F0h

When I start the system, I just see the message "SI54P AIO v2.1". There's no long string.

Seagate tech support asked the same question. They thought it might be waking from a power saving mode. But this is immediately after starting Windows.

The HDD light is solid.

I have a Compaq 166MHz system that I'm prepared to give him if I can't get this clunker running better, so I imagine it won't be too great a loss if it gets stuffed during flashing!

Thanks

g

Posted: Sat Jan 25, 2003 12:11 pm
by NickS
Errrk. Phoenix ? Hmm. Further things you can do...check if there is a PCB revision no. 2.x printed on the board; peel back the label on the BIOS rom to find out what type of chip is used (they mention SST and MX). From the dat I wonder whether you've got a beta for the v1

To be honest, I don't know that changing the BIOS is going to help as they're both so old. I wonder if the problem is to do with the HDD I/O mode or with the DDO. I take it that the old drive didn't have this problem ?

Posted: Sat Jan 25, 2003 2:49 pm
by Denniss
What's the OS you are using ?
Have you enabled DMa for your HDD ?
Maybe there are some usable drivers on www.sis.com

If you still have these HDD problems you might consider of getting a new mainboard or a PCI-IDE Controller

Posted: Sun Jan 26, 2003 12:15 pm
by schvankus
Running 95B.

Tried installing 98SE but it ran like a dog.

Yes, DMA is enabled.

The old 850MB drive didn't have partitions, so I couldn't reproduce the problem. I did a reformat on it and reinstalled 95B and it did some strange things too (took 15-20 minutes to add a 1MB update in Windows setup!). I did enable DMA and it seemed to speed things up.

I know this is an old system, but I've upgraded similar ones with good results. I just feel that maybe I'm missing something simple here, but I can't think what it could be.

Nick, you mentioned it might be a beta. The board lists a few jumper settings that aren't there; they're hard-wired. Could this thing have just been bad out of the box?

As this computer isn't mine, it's hard to know if it has ever been normal.

I do appreciate all your suggestions and thoughts.

Cheers

g

Posted: Sun Jan 26, 2003 2:30 pm
by NickS
You reformatted the 850MB and reinstalled Win95 and it now runs slow, or normal after some funny behaviour ?
I don't know enough about the Seagate DDO to know what it does about different drive modes, but the fact that the drive light is on solid for 15 seconds does sound as though there may be some sort of time-out occurring beacuse the transfer is not wrking properly, which is why I was wondering what disk mode is bing used. Presumably turning DMA off on the new drive makes no difference ? Certainly we have had experience of Award BIOSes trying to set up UDMA66 when the chipset would not do it, requiring the use of the UDMA mode setting tool.

Posted: Mon Jan 27, 2003 10:52 am
by schvankus
It took a good 90 minutes for 95B to install on the 850MB drive. Afterwards when I went to add "games" in Windows Setup (in Add/Delete programs), it took about 20 minutes! This is about a 1MB file and generally installs in seconds (I copied the .cab files to the c: drive so it's not a CD-ROM issue). Windows Setup seemed to improve when I enabled DMA, but I wouldn't say it was running great.

Is there a benchmarking utility that you might suggest i run to give me an idea of how the system is performing so I can pass it on to you? It seems slow but maybe it's in the acceptable range for this era system.
requiring the use of the UDMA mode setting tool.
where do I find that? i'll try anything!

i'm going over to his house tomorrow to work on the beast. wish me luck.

cheers

g

Posted: Mon Jan 27, 2003 11:51 am
by NickS
Mode setting: see "Collected wisdom", "links to UDMA mode setting tools". However, if you have the same problem with the old drive I don't think that's it. Have you tried a known good cable ? That can screw up transfers, although at PIO Mode 3 it would have to be pretty dodgy.

HDTach is widely recommended as a drive benchmarking tool for Windows and can be found on a lot of download sites; a disk benchmark is also included in SiSoft Sandra (look for http://www.sisoftware.co.uk, not Sandra.co.uk - you can guess why!)

Posted: Tue Jan 28, 2003 11:24 pm
by schvankus
Seeing as I've nothing to lose, I thought I'd flash the bios.

I've never used their flash utility, flash36, before and can't find any documentation (searched the wimbios forum as well).

When I boot from the flash disk, I run flash36.exe. The screen reads:
Usage: Flash [drive:][path] file_name

Note: Please release the HIMEM.SYS and EMM386.EXE device driver before programming.
hmmm. i haven't a clue how to release the device drivers.

the file name is si54paio.21i so i try typing:

flash a: si54paio.21i

says bad command

could you either point me to some documentation or help me with this?

thanks again

g

Posted: Wed Jan 29, 2003 12:24 am
by schvankus
I see what I did wrong: flash36 si54paio.21i no "a:"

It worked! And guess what? The problem seems to be fixed!

I was about to scrap it, so I'd taken out the 128MB of RAM and tossed in 24MB. I flashed it and it works better now with just the 24MB!

Thanks so much for all your help. I really appreciate your efforts.

g

Posted: Wed Jan 29, 2003 2:59 pm
by NickS
Well done! Now we know that BIOS can be used on that board. Are you still using DDO, or does the new BIOS support big HDDs ?

Posted: Wed Jan 29, 2003 10:05 pm
by schvankus
I think I spoke to soon.

The flash definitely updated the BIOS and the computer worked much better with 24MB of RAM. But when I put the 128MB of parity FPM back in, it exhibited the same problem! I tried the same amount of non-parity EDO RAM and only 64 in bank 0 and still the same thing.

Why would this computer work so much worse with more RAM???? (There's currently a pair of 8MBs and 4MBs in. One might be parity, one not.)

I've left the DDO on as the BIOS date is still quite old. Now I think there's something else going on. I reloaded defaults after the flash. Are there any changes that I might need to make in the BIOS settings?

This has definitely been the strangest pc I've ever worked on!

Posted: Wed Jan 29, 2003 11:07 pm
by NickS
Idea 1: some BIOSes disable cache when more than the cacheable amount of RAM is put into the machine.

The amount of RAM that can be supported by on-board cache depends on the CPU type and the amount of cache RAM, which may be limited by the chipset. For example, my PC-Chips M577 has 1Mbyte of cache RAM which will support 128 or 256 MB of RAM with the AMD K6 CPUs depending on whether the cache mode is write-back or write-through, I forget which way round. On this board the BIOS does not disable the cache, but any RAM above 256 Mybtes is not cached and so runs slower.

However, you mention that you tried it with only 64 Mb and it ran slowly, so it does not sound very likely.

Idea 2: It is reading the RAM speed from the links on the SIMM and getting it wrong. Some SIMMs didn't have the links fitted. Have you checked what the memory speed settings in the CMOS setup say ?

One thing you could try is running the memory bandwidth test in SiSoft Sandra. Sandra will also give you as much information as it can find about the type of RAM it thinks is in the machine. Note: it isn't always 100% right.

Posted: Sun Feb 02, 2003 8:10 pm
by schvankus
As far as I know, these are all my memory settings:

Advanced Chipset Control

DRAM CAS precharge time: 2T
DRAM RAS precharge time: 5T
RAS to CAS delay time: 4T
DRAM write push to CAS delay: 2T
PCI master access shadow RAM: Disabled
GateA20 emulation Enabled
Fast reset emulation: Enabled
Fast reset latency: 2 us
Slow refresh: Disabled
PCI clock frequency: CPUCLK/2
CAS# width in PCI master write: 1T
Latency for CPU to PCI write: 1T
CPU to PCI burst memory write: Enabled

Anything look suspicious?

I disconnected most of the cables to find the PCB version. It's 2.1 which means I have flashed the BIOS with the correct and latest version.

I formatted and reinstalled the old hard drive and loaded 95B. Everything is running fine with 24MB. I gave my friend the P166 machine running 98SE and 128MB of RAM. He's happy. I'll try to sell this one. I was just hoping to learn why this is happening with the memory so I could fix it the next time I see it (which will hopefuly be never!).