Change PSU?

on tha subject, my asus p5a is very similar, it has a jumper setting for which power supply to use. While the asus would work either way, if it diddnt match what I was using, sometimes it would work wonkey. Also, memory, you arent mixing modules are you? is the bios up to date?

I didn't see a K6-2 mentioned in the manual, only a K6. What happens if you try a regular K6? I have a TX board which mentions a K6-300 in the manual, but it does not work with a K6-2 without a hacked BIOS update.

What are your BIOS advanced chipset settings like? I've noticed this behavior on a TX board if SDRAM speculative read is enabled, or CAS is set to 2/2 instead of 3/3. Also, if your BIOS has an enabled cache pipeline read/write option, this has caused similar problems. Disable it.

I have had some odd issues whereby an S370 motherboard worked out of the case, but in the case it did not work. There was some grounding issue with one of the offsets.

You may also want to play with an Intel P55C at 166 MHz first to establish motherboard stability.

Are you using the AT or ATX power connector? What happens if you switch connectors? Switch RAM modules? Increase CPU voltage? Try a Virge DX or something of established compatability. Does MemTest pass?

Good luck! This can take days to track down.

I wonder if one of those ~$5 POST cards you see on ebay would help diagnosing such problems.

Well, I'm using an AT power supply, which so far gave me no reason to suspect of anything bad (it holds a full slot1 system just fine); I will try ATX tonight (no jumper for that here).

I'm not mixing RAM modules (each time I used either a pair of identical DIMMs, or just one DIMM - and tried both DIMM slots).

The BIOS is this one here (first in list, K6-2 support mentioned) - http://www.elhvb.com/mboards/fordlian/Garner.html - and both the 333 and 400 MHz K6-2s are properly identified at POST time (that is, when it does POST).

I didn't pay attention to most of the Advanced Chipset BIOS settings, I can only remember CAS being definitely set to 3 (I tested with HYS64V4120GU-10 Siemens DIMMs and some no-name PC100 DIMMs).

I do have a P55C/166 for testing, and the video cards used were two ViRGE/DX-based PCI cards and an ISA TVGA9000i. Increasing voltage for the K6-2 CPUs might be a bit tricky though - the manual shows settings for 2.2V and 2.8-2.9V, but nothing in between. I'll also try memtesting tonight.

A diagnostic card might also be a good idea, since the speaker is mute 🙁...

Well, tonight's debugging session went pretty weird (to a good end though, hopefully permanent).

Armed with a multimeter borrowed from work, I started the board on a different PSU (still AT, not ATX) - it worked like a charm through two passes of memtest, and would still work after doing all sorts of things to it (changing RAM, video card type and location, taking the cooler off and putting it back on); while doing this I was able to measure steady 12.15V at the input of the voltage regulator chip and steady 2.21V at its output.

I then switched back to the old PSU, and again I wasn't able to make the system not POST, no matter what I did (except for a hang in memtest when I probably touched something while measuring voltages - this time a bit higher 12.30V/2.21V, but still steady).

Then I put the system together (using the new PSU, just in case), and to my surprise the Win98SE previously installed on Intel CPU / VIA chipset worked just fine on AMD CPU / Intel chipset. And pretty snappy too, with only 64MB of RAM (played AoE, Unreal, surfed the net), although I think / hope Win95 OSR2.5 will feel better in that amount of memory...

The voltage regulator's (HIP6008CB) datasheet (attached) showed me how to set the Vcore from 2.0V to 3.5V in 0.1V increments (its VID0-VID3 pins correspond to JP10's four pairs of pins) - I wonder how a K6-III would behave...

Another thing I noticed is that, while the two passively-cooled transistors next to the CPU socket stay almost cold, the one between the PCI and ISA slots could melt my skin in seconds - can anyone guess what it does there?

And yet another interesting thing is that Everest Home Edition identifies the board as Acorp 5TX29 which, except for the green PCB, looks identical to this one - I wonder if an Acorp BIOS would work on mine...

I finally found an unofficial BIOS for Acorp 5TX29 with support for K6-2+/III/III+:

http://web.archive.org/web/20021007192939/htt … orp/socket7.htm

The "Review" link (in Russian, but Google Translate does a decent enough job) shows how to set the undocumented 83MHz FSB, so if I had been born with the overclocker's gene I could have tested my K6-III/400AFR at 450 (6*75) or even 500 (6*83) 😀.

I'm pretty happy with what I've done and learned today, which in the end might be all that matters about this hobby...

iulianv wrote:

I'm pretty happy with what I've done and learned today, which in the end might be all that matters about this hobby...

We pretend to like gaming, but it's only an excuse 😀

I thought the testing *was* the game 🙁

Well, I installed Win95 OSR2.5 last night, and it didn't blow me away at all in terms of speed, compared to Win98SE. So I guess I'll throw in 128MB of RAM or even max it to 256 (despite 430TX's limit in cacheable size), put a K6-2/450 (6*75) instead of the 400, and go back to Win98SE... 2.7GB of HDD capacity is a bit tight though, so I'll also need to look for a bigger drive.

I have always liked the 430TX board for its stability and feature set. For the cacheable limit, have you tried a K6-2+/3+ so that you can cache all 256 MB? If these don't work, what about a regular K6-III? I have not had any trouble running a K6-3+ at 83x6 in my 430TX board, though I replaced the cache TAG with a 10 ns piece instead of the regular 12 ns one.

83MHz FSB might be too much for the PCI bus, I think I'll stick to the moderate 75 😀.

Isn't K6-III "the same" as 2+/III+ when it comes to RAM cacheability? I mean, K6-III has on-die L2 cache too, right? I don't have any +, but I do have a K6-III/400AFR that I could try, and if the BIOS acts weirdly on it there's always the hacked Acorp 5TX29 BIOS on that Russian site. All that just for fun though - for more regular use I'd save the K6-III for a SS7 board 😀...

If the motherboard uses a 1/2 FSB-to-PCI multiplier, I suppose the PCI bus would be at 41.7 MHz. I haven't had any problems with a Matrox Millennium G200, Promise ATA card, nor Intel Pro PCI card with an 83 MHz bus.

To my knowledge, the K6-III and K6-III+ are very similar, except that some BIOSes may not like the + chips for whatever reason.

According to the latest plan, I'm back to Win98SE (this time without the unofficial service pack), on K6-2/450 (6*75) and 128MB of RAM.

It's running pretty well, but obviously something always has to be wrong - this time is Device Manager's DMA setting for the hard-drive: when enabled the whole thing crawls - front HDD LED is constantly on, booting lasts forever and even a minor operation like right-click/Properties on a desktop icon can take minutes.

I tried several IDE cables, I tried using the secondary IDE port, I tried leaving the hard-drive (Maxtor 90288D2) alone on the cable, I tried the latest chipset drivers for 430TX from intel, nothing worked. The BIOS sees both the HDD and the CD-ROM as UDMA33, and I think I could get a pretty significant increase in performance if I could just make that DMA setting work...

I personaly use PCI ATA cards of some sort. Use the vendor supplied drivers. See if this works first with your current HDD, CD-ROM, and cable.

Well, I do have a floppy disk that came with the mainboard and contains a BMIDE folder, but I suspect it's for Win95... I did try the Triones v3.70d drivers though, which didn't even allow me to boot Windows (well, at least not within reasonable time).

The interesting thing is that going back from 75 to 66 FSB makes the problem go away (I can leave the DMA setting on - it doesn't seem to help (probably due to the slow hard drive), but it doesn't hurt anymore either). Linux, however, seems to have no problem running FSB 75 with both the HDD and the CD-ROM drive at UDMA2 (ATA-33) speed (I tried some older versions of two LiveCD distros - Knoppix and systemrescuecd).

As usual, more information brings more questions 😀:

- assuming that I get a faster drive (somewhere in the 8-20GB generation), is it safe to say that the DMA setting in Device Manager would bring me more responsiveness from the system during office/net use, while 50MHz extra for the CPU would increase gaming performance?

- is there any chance that a different BIOS (I'm thinking about that identical Acorp 5TX29) might get me both the DMA and the 50MHz in Win98SE?

- is it worth max-ing the RAM to 256MB and replacing Win98SE with Win2K?