@skyscraper

I'm testing a Dfi socket 370 mother. P3 550E oced to 733 mhz running at stock vcore without any issues. Good motherboard and good cpu. Hopefully this will be a rig to test cards under wxp.

luckybob wrote:

@skyscraper […]
Show full quote

@skyscraper

Close but no cigar. The board gets stuck at code "28" "initialization before setting the video mode is complete" the next step is "2A" "Bus initialization system, static, output devices will be done next, if present."

The behaviour is exactly the same with one or two CPUs, registered memory or normal memory, AGP or PCI video.

The board will not post with one or two 900 MHz 2M Xeons, at least not with this BIOS and I do not have any other Slot-2 CPUs to try. I think both the board and CPUs work? Without a CPU in the socket my diagnostic post code card shows nothing while with one or two CPUs it at least showns signs of life. The board does not beep though.

As always I ran in to some other issues.

This is not good.

Temporary fix.

It also looked like some pins in the SCSI card slot might be shorting so I inserted a piece of paper.

#1: slot 2 works identical to slot 1. You need a terminator in the 2nd slot or a 2nd cpu.
#2: 28 bios code for ami bios seems to be a early video error. Its stopping just is it starts to init the video. Are you using an agp 2x video card?

luckybob wrote:

#1: slot 2 works identical to slot 1. You need a terminator in the 2nd slot or a 2nd cpu.
#2: 28 bios code for ami bios seems to be a early video error. Its stopping just is it starts to init the video. Are you using an agp 2x video card?

Im using a known to be working S3 Trio 64 PCI but I have also tried both a Geforce 2 MX 400 and a Matrox card.

My Supermicro Super P6DGU dual Slot-1 board do not need a terminator when running with one CPU, there is nothing in the manual for the Super S2DG2 that says I need a terminator but I have tested with both CPUs installed and it stops at the same code.

I will try connecting two PSUs with both CPUs installed to exclude the terminator issue in combination with too little power, I just need to eat something first.

Using two PSUS with both CPUs installed did not change the situation, the system gets stuck at code 28.

I also tried yet another video card, a Riva 128 and the system got stuck at code 28, then i tried no video card and it got stuck at code 28. Then I added my second diagnostic post card that shows both the the second to last code and the last code, it showed 27 28. The diagnostic cards at least agree.

I have also checked all SCSI termination jumpers and cleared the BIOS a couple of times. I have tried jumpering the multiplier to the two fail safe modes, all on and all off but this should not matter as the Xeon CPUs multiplier is locked, I have also tested to force 66 MHz with the FSB jumper in case the caps are borderline bad and wont do 900 MHz but this did not work either. Code 28 seems to be the flavor of the day.

I'm willing to bet the entire board needs to be re-capped. The agp card might not be getting proper voltage. maybe try a pci card for giggles?

I assume you have also tried changing the memory modules?

luckybob wrote:

I'm willing to bet the entire board needs to be re-capped. The agp card might not be getting proper voltage. maybe try a pci card for giggles?

A few posts up he wrote that a PCI video card is used for testing.

luckybob wrote:

I'm willing to bet the entire board needs to be re-capped. The agp card might not be getting proper voltage. maybe try a pci card for giggles?

As I wrote above I have tried a known to be working PCI S3 Trio 64 card and I have also tried without a video card, this board should complete POST without a video card.

Yea there is a good chance the board needs a full recap but the board was sold as working when I bought it half a year ago, the seller is a reputable guy with perfect feedback score.

At least I got a good indication that the CPUs do work so I can leave a positive feedback on those.

sorry, missed the part about it being pci.

Anyway, this board uses the 440gx chipset, so you want to use non-buffered ram. ecc is optional.

luckybob wrote:

sorry, missed the part about it being pci.

Anyway, this board uses the 440gx chipset, so you want to use non-buffered ram. ecc is optional.

I have tried the memory from my other Supermicro i440GX board (the P6PGU), it's unbufferd ECC but I did also try the memory from my Compaq Deskro EP/EN it's standard SDRAM. The manual says that all memory works as long as it's PC100. I guess the performance should be same when all error corrction is disabled in the BIOS setup.

Sounds like you're doing a thorough job (I did not expect otherwise). Frustrating but also a bit of fun to do I hope 😀

Cyrix200+ wrote:

Sounds like you're doing a thorough job (I did not expect otherwise). Frustrating but also a bit of fun to do I hope 😀

Yea, the time money is never totally wasted as you always learn something.

The next step will be to get two officially supported CPUs just to make sure it isnt some wierd compatibility issue, if those dont work i will recap the board. As the board is littered with shady "NRSY green" caps it's the last resort. I think I counted 31 "NRSY green" if you include the odd already replaced ones. The 8 Sanyo caps should be ok though.

Skyscraper wrote:

The 8 Sanyo caps should be ok though.

Never assume that. You're already replacing a few caps, buy a few more and do a complete job. Besides, IT'S FUN! I think this site ships worldwide: http://www.mouser.com/nichicon/

luckybob wrote:

Skyscraper wrote:

The 8 Sanyo caps should be ok though.

Never assume that. You're already replacing a few caps, buy a few more and do a complete job. Besides, IT'S FUN! I think this site ships worldwide: http://www.mouser.com/nichicon/

I recounted the caps, there are 33 "NRSY green" (odd already replaced ones included) and 8 Sanyo caps, very fun replacing them all indeed. At least I have a desoldering station so I wont damage any traces when desoldering. I took a closer look at my other i440GX board the Super P6DGU as I need to replace some "NRSY green" caps on that board aswell. On the P6DGU all solder pads for the caps are nice and large, not so on the Super S2DG2, it's cap solder pads are like grains of sand.

In the last two weeks or so, I was mainly tinkering with my newly found 486 system. This included cleaning the case, dusting the PSU, fixing the 5.25" floppy drive, replacing an ISA video card with a VLB one, replacing the hard drives with a 128MB CF card and finally messing with a ST351A/X which turns out to be perfectly healthy.

I've been quite busy, as you can see 😀

Finalized and posted my project on testing the effects of disabling L1 and L2 cache on 486 and higher processors.
Let's benchmark our systems with caches disabled

luckybob wrote:

You need a terminator in the 2nd slot or a 2nd cpu.

Out of quriosity... Do all Slot1-CPU-Terminators work the same, electrically speaking.

brostenen wrote:

luckybob wrote:

You need a terminator in the 2nd slot or a 2nd cpu.

Out of quriosity... Do all Slot1-CPU-Terminators work the same, electrically speaking.

The principal is the same. They are used to make the signals work properly. Cpu terminators are just like scsi terminators.

examples:
Slot 2: http://www.ebay.com/itm/391254846918
Slot 1: http://www.ebay.com/itm/281892170303

Socket 370 works the same way as well, again just in a different package.