K6-2+, K6-3+ support for BCM IN5598 w/ added functionality

Recommended by Wim's BIOS...

by **Uranium235** » Thu Jul 29, 2004 7:45 am

Greetings. I have an old socket 7 motherboard manufactured by BCM/GVC known as the IN5598 or FR500. This motherboard originally came from a NEC Ready 9851 system. It is also known as the Packard Bell PB850 motherboard from the Packard Bell Club40 systems. I believe this board may have found it's way into some Compaq systems also.
I've been using K6-2 380 at 5.5x75 with some success but I can tell you there is no BIOS support for the CXT core Write Allocation. I have been using Rawpower to enable the CXT features in Windows.
I'd like to upgrade to a K6-2+ 450 I recently picked up. I checked Jan Steunebrink's excellent website for a compatible BIOS but found none. I did a Google search and the best I could come up with was a patched FR500 BIOS at wims.rainbow-software.org for the 32GB HD limitation. I took a peak at the patched BIOS with MODBIN and noticed there are many BIOS settings that are hidden. The RAM timings could be very useful if unhidden as well as some other settings.
This may be a tall order but would it be possible to see a patch with K6-2+, K6-3+ support, CXT core support, the 32GB HD limitation patch, and BIOS settings unhidden.
Here are some useful links I've found:

Manufacturer link:
http://www.bcmcom.com/tech/in5598/IN5598.htm

Packard Bell link:
http://www.uktsupport.co.uk/pb/mb/850.htm

Newest BIOS version available with some bug fixes:
http://www.pbnec.nl/support/itemnr/REFBIOS00300200.html

Patched BIOS for 32 GB HD limitation: (FR500)
http://wims.rainbow-software.org/

I'd truly appreciate any help that can be provided.

Brian

BiosMan · by **KachiWachi** » Thu Jul 29, 2004 1:39 pm

Contact Biosman (Jan Steunebrink) from this site, or at his homepage. I'm sure he will add this BIOS to his ever growing list.

by **Uranium235** » Fri Jul 30, 2004 4:50 am

Will do. Thanks.

by **BiosMan** » Sat Oct 23, 2004 10:40 pm

Uranium235,

Patched BIOS sent to your e-mail address...

I've added K6-2CXT, K6-III, K6-2+, and K6-III+ support up to 500MHz (6x83). Also did the 32GB and 64GB bugfixes for 128GB IDE support and enabled a large number of previously hidden BIOS options.

Let us know how it works.
When tested, I will make it available on my k6plus page. See link below.

BiosMan.

by **Uranium235** » Thu Oct 28, 2004 5:20 am

BiosMan,
I received the patched BIOS (IN5598J2.zip) and installed it a few nights ago. I flashed it to my motherboard successfully using Uniflash.exe with a K6-2 380 installed. I then entered the BIOS and verified all the unhidden BIOS settings. The motherboard than booted successfully into Windows 98SE without any problems. I was able to verify that the patched BIOS had successfully enabled Write Allocation for the CXT core.

The next day I installed a 40GB Western Digital hardrive and did a clean install of Windows 98SE. The 32GB hardrive limitation appears to have been successfully patched. I don't have a larger hardrive available to test if the 64GB limitation had been successfully patched.
There is another bug I am experiencing with this hardrive however. The BIOS is seeing the hardrive as UDMA-5 when the motherboard is only capable of UDMA-2. This causes the hardrive to choke when DMA is enabled in Windows. I remember Petr Soucek had mentioned this bug here:
http://www.ryston.cz/petr/bios/award.html
"3. UDMA mode limitation
In the older Award BIOSes was the HDD set to highest supported mode, it was not problem when only UDMA33 HDDs were produced, but later problem arised: HDD was programmed to e.g. 66 MHz and chipset maximum was 33 MHz, the result was unreliable operation. There are some solutions, switch off UDMA in BIOS setup at all, or use special program supplied by HDD manufacturer and switch off ATA/66 support. Better way is to implement this directly into the BIOS code. FYI:
UDMA mode 2 = ATA/33
UDMA mode 4 = ATA/66
UDMA mode 5 = ATA/100
UDMA mode 6 = ATA/133 "
I tried to use the Western Digital Data Lifeguard Tools to force UDMA-2 on this 40GB hardrive but it won't accept it. I know this hardrive is UDMA-2 backwards compatible. Can this be patched?

Tonight I installed the K6-2+ 450. It successfully posted as a K6-2+/450 (6x75) on the display. I clocked it at 500(6x83), 412(5.5x75) and 400(6x66) and it posted everytime showing the correct clock speed.
Now the major problem begins. This machine, with the K6-2+, will only boot into Windows only if I disable "External Cache" in the BIOS. It will not even boot into DOS with a floppy bootdisc. I've tried nearly every combination of BIOS settings to get around this. Any ideas to fix this?

A minor bug that must have pre-existed is the setting for "Quick Power On Self Test". Enabled is disable and disabled is enabled. They are switched.

There is an IRQ steering bug in this BCM BIOS that I didn't catch until now. It appears to have been patched in the Packard Bell BIOS version 2.00.
Here's the link:
http://www.pbnec.nl/support/itemnr/REFBIOS00300200.html
"On some FR500-based systems it may happen that the Windows98 Device Manager shows an error in loading the IRQ Table from the PCI BIOS. This can be verified by checking the IRQ Steering Tab in the PCI Bus properties screen. To do so, go to Device Manager, System Devices, PCI Bus and click the Properties button. Note that this error has been reported on systems with BIOS versions 2.06 or 2.07. If your system currently uses version 2.06 or 2.07 and does not show any errors as described above, there is NO need to flash to BIOS version 2.00."
I looked at 2.00 BIOS with Modbin and it seems the only change they made was a new setting for "ACPI" which they disable by default.
Can this bug fix be implemented in the IN5598J2 patch?

BiosMan · by **KachiWachi** » Thu Oct 28, 2004 1:55 pm

Uranium235 -

I would check with Western Digital about why their tool cannot set UDMA-2 on your drive. Perhaps you need a different version of their tool, or your drive does not have that feature available. What model drive do you have?

Have you tried turning off System BIOS Cacheable (and/or Video BIOS Cacheable) in your BIOS? I had to do that for my patched BIOS.

It sounds easy to fix the Quick Power On Self Test...just switch the displayed names! (Can you do this in MODBIN??)

Once you can boot to DOS, you can run Craig Hart's PCI program with the switches "-d -b -p >data.txt". This will dump the Device Configuration Registers and the PCI Routing Table to the file "data.txt". Note any errors it provides.

by **Uranium235** » Sat Oct 30, 2004 10:29 pm

Thanks KachiWachi,
I turned off all cacheable and shadowing options in the BIOS and was at least able to boot into Memtest86+ which ran until it piled up over 2000 errors before freezing.
Keeping in mind that board's L2 cache (L3 cache with K6-2+) runs at fsb speed I dropped the fsb down to 60 and clocked the CPU at 6x60=360Mhz. This got the board stable enough to run Memtest86+ without errors although the machine still seems abit flakey.
I can now safely boot into DOS and will run Cachechk V7 and Readmsr shortly.

This is the hardrive I'm running with the IN5598:
http://www.wdc.com/en/products/products.asp?DriveID=36
I checked Western Digitals knowledge base regarding the UDMA-2 issue and couldn't find any answers. The newest Western Digital Dlg tool version 11.0 doesn't even have a ATA manager utility. I had to use version 2.8 to try to force UDMA-2 without success. Perhaps they've engineered this support out since these older boards are approaching obsolescence?

It looks like I can use Modbin to switch the displayed names for Quick Power On Self Test.

Can you point me to a direct link for Craig Hart's PCI program and what specifically I should be running? I've never used it before.

by **Uranium235** » Sun Oct 31, 2004 12:43 am

Never mind. I got the right PCI program from Craig Hart's website. I will post the results after I run it.

BiosMan · by **KachiWachi** » Sun Oct 31, 2004 2:28 pm

I needed to turn off the System Cacheable so the system would boot. The Video Cacheable just seemed to make the board work for shutdown and sleep modes, with Win98SE. I have not tried turning it back on with W2K, my current OS. Shadowing is currently on, and I have no issues with that.

Use Memtest to sort out your RAM errors first. Tests #5 and #8 will be where your errors will most likely show up. Once you sort those out, run the Extended tests as well, to make sure you have your refresh down correctly.

Check your on-board cache and TAG chips, and see if they are 75MHz compatible. They should be, since your board link mentions that it can run at 75MHz. Also check your SIMMS...how much RAM do you have installed, and what chips do they contain? 60nS?

I've been fooling recently with my Amptron PM-7900 (PCChips M520) i430VX board to learn if it can stand the overclock from 66 to 75MHz bus, so I can eventually try it on the DFI. So far I've found that I need to back the RAM timings down so that the clocks occur when they are supposed to, otherwise I will get a TON of errors, or just a few. See my DFI post for more on what I have done so far.

One issue I will still have with the DFI is with my SCSI card...but see my post for more info on that as well.

by **BiosMan** » Mon Nov 08, 2004 9:29 pm

KachiWachi,
Thanks for helping-out in my abcense.

Uranium235,
Good to hear the patched BIOS and K6-2+ work. About your problem reports:

UDMA-2:
This should work oke when you use a regular 40-wire IDE flatcable. The drive will "see" that a 80-wire cable, required for higher modes, is not present and limit the timing to UDMA Mode 2. The UDMA-5 indication of the BIOS is just cosmetic and indicates the cabability of the drive; not the actual timing.
If with a 40-wire cable the PIO4 mode is stable but the UDMA-2 is not, then this is usually a problem with the Windows Bus Master driver.

Problems running the K6+ with External cache Enabled:
The requirement to Disable the "System BIOS Cacheable" BIOS option for running a K6+ is a known issue with the i430VX, i430TX, and Aladdin IV(+) chipsets. I've never seen this problem with other chipsets but it is possible the SiS5598 is affected too. So keep this option Disabled, while leaving the other Chipset Features at their Setup Default setting when running the test utilities.

"Quick Power On Self Test" bug:
I'm not sure if you can change the Disabled/Enabled labeling via Modbin because these labels are used by many other options as well. But you could change "Quick Power On Self Test" into "Long Power On Self Test". :wink:

Please mail me the full PCISniff, Cachechk, and Readmsr logs for analysis and I will report back here with my findings.

Greetings,
BiosMan.

BiosMan · by **KachiWachi** » Tue Nov 09, 2004 3:20 pm

NP Jan.

I didn't think about the 40/80-wire cable method you mentioned...that is a slick trick that will let the drive adjust the UDMA on it's own...cool!!

Why does his BIOS show UDMA-5 when the chipset is only UDMA-2 capable? My VX BIOS does not show my UDMA-5 capable drive that way. I though this was dispalyed from the BIOS HD tables...do you mean to say this particular item is read directly from the HD, and posted to the summary screen?

Side note - would using an 80-wire cable on a UDMA-5 capable drive, with a PIO-4/MWDMA-2 capable chipset, do anything useful? More noise immunity perhaps??

by **Uranium235** » Tue Nov 09, 2004 4:12 pm

Thanks for the response Biosman. Yes, your patches appear to be working flawlessly.

I've been tied up with work and unable to test this motherboard anymore lately, hopefully tonight maybe...

I will try your suggestions and get the test results to you. Thanks.

by **Uranium235** » Fri Nov 12, 2004 5:11 am

Biosman,
The test results have been E-mailed to you.

by **Uranium235** » Thu Dec 02, 2004 10:25 pm

Just a few updates.
I checked the on-board cache and TAG RAM and they seem to be able to handle the 75 fsb. I've had this board stable all the way up to the 83 fsb with a regular K6-2 so I should be OK there.
The RAM I was using was 4x32MB 60ns EDO which shouldn't have been a problem. I have just acquired 4x64MB 50ns EDO which has already increased the stability at the 66fsb. I just need some time to run Memtest86 to tweak the timings.
This board seems to like to run this K6-2+ 450 at only 1.8v instead of 2.0v. I get alot fewer errors with the lower v's.
A little burn-in seems to be helping. The longer I run it the better.
Biosman has finished analyzing the PCISniff, Cachechk and Readmsr results and has found no irregularities.
I'm shooting for stable operation at 6x75 at 450Mhz.

BiosMan · by **KachiWachi** » Fri Dec 03, 2004 2:42 pm

50ns EDO...boy would I kill for some of that... :cry: