...more like...what can this motherboard do...M5ATC Rev 1.0

[KachiWachi]

Yes...AWARD BIOS...patched version for 32GB HD bug is even available here...

OK...

So then if you boot to DOS (mode), what Mode is the drive in, assuming it is set in the BIOS to UDMA?

How can the drive force a Mode on a BIOS that it doesn't know about? The drive doesn't "send data" to the BIOS/Chipset...it just reports what it is capable of and then the BIOS does the setup...right? If it can't do UDMA5, wouldn't it just go up to UDMA2 and set that...assuming it is intelligent enough not to mis-interpret things and set it up for PIO-1...LOL

If the drive is only MWDMA1 capable, what does setting Windows DMA do?

I have 2 utilities that I use...ATAPROBE and ATAINF...that run out of DOS to see what Mode my drives are in. With the Seagate, it is too new a drive, and the results are not entirely accurate (the programs are 1996 era). But I do have a drive that posts a result not what it is capable of...I'll put the result printout here when I get home...thought it was SWDMA2...but in any case...enabling DMA on that drive would make it run slower than in PIO-3...correct??


Back to the UDMA33 bug...is it all UDMA modes or just UDMA33? Can UDMA0 (16.7 MB/s) and UDMA1 (25.0 MB/s) still be set, if I had a drive that utilized it (from what I have read, only UDMA2 (33.3 MB/s was ever implemented in a drive...)?

[KachiWachi]

Just to update...

Flashed the board up with the patched BIOS from here (Large HD Support) and tried the 40GB Seagate. Detected fine as such. Have it locked at PIO-4 until I get some more feedback on my previous response to UDMA.

Still having weird memory errors when using GoldMemory, but I e-mailed the author...we'll see what he says...

[edwin]

Weird or not, if goldmemory reports something wrong with your memory you better throw out that stick and try some tht pass the test. Believe me, it will save you from a lot of stress...

[Rainbow]

Can be also memory timing set too fast in BIOS or bad L2 cache...
About the UDMA bug - have you TRIED if the chipset is buggy? Install OS, enable DMA and do some heavy HDD work (e.g. copying a lot of large files and then compare them if they copied OK).

[KachiWachi]

edwin - thing is MemTest86 (v3.0) says there are no errors...is one program more effective than the other here?? I'm running overnight tests with many passes.

What is weird is this (with GoldMemory)... I have a matched set of 64MB EDO in 4 sticks. As a set of 4, they fail all over the place. But in sets of 2, in any combo, and in any bank, they pass!! How do we explain that??

With my other matched set of 64MB FPM in 2 sticks, these guys fail like 35 mins into the first pass, then no errors in all subsequent passes...can we explain this too??

Rainbow - haven't tried UDMA yet since I'm sorting out the memory bug. Memory timings are set to Auto at present, but appear to be nothing out of the ordinary, assuming what I see in the greyed out area is what has been set... Just been going on what others have said about the bug, since I can't seem to find out if it was just a particular revision of the chipset or ones that were sent to a particular motherboard manufacturer or...???

[KachiWachi]

Interesting results here...

Just installed HDTach 2.61 and tested the combination. HD was not limited by mfgr's software, 40-pin cable used, Windows98SE drivers used (did not install the ALi downloadable ones).

BIOS set to PIO-3, Windows DMA enabled:
14.9 ms Random Access Time
10 mbps Read Burst Stregnth
9855.9 kps Average Read Speed

BIOS set to PIO-4, Windows DMA enabled:
14.9 ms Random Access Time
14.6 mbps Read Burst Stregnth
12,441.4 kps Average Read Speed
(CPU Utilization 9.3%)

BIOS set to UDMA-2, Windows DMA enabled:
15 ms Random Access Time
14.6 mbps Read Burst Stregnth
12,461.9 kps Average Read Speed
(CPU Utilization 9.4%)

Attempt at disaster -
BIOS set to Auto (UDMA-5 posted), Windows DMA enabled:
14.8 ms Random Access Time
14.6 mbps Read Burst Stregnth
12,474.2 kps Average Read Speed
(CPU Utilization 9.8%)


Seems something is stopping the drive from going into UDMA mode. Also noticed that around 21GB, the graph has a step-up in read speed...what is going on there??

** Just downloaded the Seagate Uata100 (DOS Mode) utility...it says the Drive is configured to UltraDMA-5, but the System is only in MWDMA2, even though it is set at UDMA-2 (or even Auto) in the BIOS. So that is why I seem to be locked at the same transfer rate...but why???

Is this BIOS limited (fixed) not to allow UDMA? Is this revision of the Chipset not UDMA capable (Rev. B1)??

[KachiWachi]

Interesting occurance...

I completely gutted this guy so I can set it up new for my girlfriend. Just have some RAM, the WD AC28400 8.4GB HD (port 1), a TORiSAN CDR_S18 (8x, port 2), and a Cirrus Logic PCI Video Card installed. Loaded Win98SE on a fresh format. BIOS set for no UDMA. Everything comes up fine...but...

...when I set DMA on the CDROM (which I know worked fine when it was installed in my DFI) the drive name comes up corrupted in the Device Manager...looks like C$ or like that...and if you look further, it is almost like parts of the device name are showing up in the firmware area, etc... and DMA is not set, and cannot be set after this. HD DMA is OK.

A look in the registry does show a CDROM with this corrupted info there, as well as the correct TORiSAN info below it. Even though I remove the drive from the system, and even delete all references to it in the Registry, it will not post the correct info again till I reinstall Windows.

Any thoughts on what might be causing this to happen??

[Rainbow]

Looks like CD-ROM without DMA support.

[KachiWachi]

Does have DMA support...was in the DFI...as I said in my previous response here...

[KachiWachi]

OK...more news...

I installed the ALi Integrated Driver Package 1.091 just to see if that would cure the freeky CD problem. When I run the IDE Cache Utility, it says my chip is too old...so can I assume it is not a M1543C-B1 but really a M1543-B1?

As soon as the machine rebooted, it made the CD funny again. The driver remained the Microsoft Driver Esdi_506.pdr as well.

Thoughts??

[KachiWachi]

Results of CTPCI -

c't PCI-Optimierer Version 3.03.00
(c) Georg Schnurer, Redaktion c't, 1993..1998

Searching PCI-Devices (CFG_Type 1)
Searching 805 glue... Getting Vendor and Device Info... Reading PCI CFG Space...

Bus Nr. 00h
--------------
00 0 Vendor=Acer Labs (ALi) No PCI-INT
Device=ALi M1531 Aladdin IV 586-Chipset, Rev=B2h

02 0 Vendor=Acer Labs (ALi) No PCI-INT
Device=ALi M1543 Function 0: PCI to ISA Bridge , Rev=B4h

05 0 Vendor=Cirrus Logic No PCI-INT
Device=00B8h -- new device ID, Rev=45h

0B 0 Vendor=Acer Labs (ALi) INT A=IRQ 0
IRQ routing Error, check setup
Device=ALi M1543 PCI Enhanced IDE Controller, Sub-Device, Rev=20h

OK...so what have I got here...a plain 1543-B1 or a 1543C-B1??? ...and what might that error be from?

[Denniss]

You still have the Southbridge with the UDMA Bug - I remeber a post from Peter missel in Pcchips Lottery Board from Rev 22h Bug is fixed .
IT's a plain M1543 - the M1543C has this Bug fixed and supports UDMA66 AFAIK

[Rainbow]

M1543C is UDMA/33 only. M1543C B1 is UDMA/66.

[KachiWachi]

From The PCChips lottery...

Peter Missel: "It wasn't until September 1999, when ALi finally admitted that the pre-C 1543 can't do UDMA33 right - and they didn't do so by press release or something, but because they published their Linux IDE driver.

Herein, the interested audience could read the below piece of art (hope it comes out halfway readable here), which avoids enabling UDMA modes with devices that are known for not doing it right: WDC harddrives and ALi IDE controller revisions 20h or earlier (i.e. before 1543C)."

And from The PCChips lottery...

Peter Missel: "Exactly - as a reminder, the 1543 rev. B4h with embedded 5229 IDE controller rev. 20h does not do UDMA33. It was meant to, but has later been showing timing problems when UDMA33 drives actually appeared.

It was not until the 1543C that we got working UDMA33 on ALi chipsets - and 1543C-B1 added UDMA66. 1543C isn't pin compatible with 1543, so you won't find a 726 with a 1543C."


So if I read the CTPCI right, I have a Rev 20h, which has the bug.
Now that that is settled, and since I have had the UDMA shut off in the BIOS anyway throughout all this, what is going on with the CDROM???

Any thoughts on that error too btw??

[Rainbow]

No M726 with M1543C? If M1543 and M1543C are not pin compatible then ALL M726s have M1543C!