I celebrated too early, the Windows 98 SE crashes when trying to start to copy files. With aggressive timing parameters even earlier.
I tried several things: Replaced the VLB graphics with an ISA graphic card, set one VESA wait state + bus speed > 33. Setup default values in Bios.
So I will try either 33MHz to stablize everything or switch CPU to L1 WT or switch back to the original Bios. What I don't want to replace is the setup with the VLB controller + CF card + 3Com ethernetcard with XTIDE Bios.
Maybe still the setup for L1 WB is wrong, so I have to double check the jumpers. But that should be solved when switching to L1 WT, so maybe I try that first.
But another time, not today.

Ok, this is heading in the right direction now.
The fact that you are able to get all BIOS settings on max, except the DRAM SPEED, is certainly nice.
I don't remember checking the AWARD BIOS. For AMI the only setting that i had to set to below optimal value was CACHE BURST READ CYCLES 2:4 = X-2-2-2 (best is X-1-1-1).
It will become clear what's what once you run the perf tests.
EDIT: You posted concurrently. Ok, looks like there is still flakiness you have to sort out.

Btw, somebody who bought a previous version of this motherboard from me sent this message through Ebay:

Hey, i bought one of the synphony wagner chipset 486 VLB boards from you back in October and I just saw the discussion about a version of this board on vogons which i read often but am not a member.

Fyi, I got the one i bought to work with and properly recognize 128mb RAM (including beyond the BIOS, in such things as memtest 86 4.3.7 or speedsys where it no longer allows probing for the memory size), by using the newest award bios from the retro web 's page on the Data Expert EXP4045 motherboard which uses the same symphony chipset. This was done with a winbond w27c512 prom chip. Anyway its just a thought in case you wanted to suggest it to the other guy.

I am late to pass the message since you already stepped on the AWARD BIOS path, but it is nice to see others participating indirectly from "outside".
In general, i am noticing an increase in Ebay communication (related to buying/selling the junk we play with) where people bring up topics discussed in Vogons.

I have an update:
I tried the Windows 98SE installation with lowered bus of 33MHz. Also switched back to the original 2800 Ami Bios. Then set the settings to Default, which seems to be ultra slow, I think cache write 2-2-2-2 and cache read 3-2-2-2. Still crashing at the beginning of the Win98SE installation. Omg. I replaced then the RAM with other modules, with 2 times 16MB or 2 times 32MB (but different modules) things started to work.
The Windows 98 installation finished. But it really needed those slow settings as it seems.

Today I tried with 40MHz again. Short summary: no dice!
Even with slowest (!) settings and L2 cache completely disabled => Not working.
Then I replaced the VLB graphics card with an ISA graphics card (so only one VL card in the system) => Not working.
I lowered the clock to 120MHz, still the same.
I checked the voltage on the voltage regulator when the board was running: 3.52 volts, that is good.
Last thing was changing both JP20 and 21 to 2-3, which is recommended for 40MHz (1 WS and Bus speed > 33) => Still the same.

The only explanation I have, that the board really does not like the VLB I/O controller at 40MHz.
I was thinking to try out 50MHz, but it makes no sense, when I use already the slowest timing settings at 40 and it still does not work. So did not make that test.

But as I wrote in the beginning, there are huge problems also at 33MHz. Horrific.

EDIT: I also switched also back to the Award Bios, using the original AMI Bios did not improve something.

More bad news:
I removed JP4 (which is obviously for the WB signal) and the board starts in L1 WT. Same crash in the Windows 98SE installation (Still slowest setting and L2 disabled!) So it is hopeless, it is not about a missing signal HITM, CACHE or whatever 🙁
So I can only run the board with 33MHz, I am afraid that is not what I want.

The weirdness continues.
So because 40MHz does not seem to work, I try to install Win98SE with 33MHz again. Not working either with the Abit Bios.
So change back to the Ami 2800 Bios, with setup default. Scrambled screen!
OK, change to an ISA ET4000. System boots, but scrambled picture during the Windows 98 installation, 🤣.
What the fuck?
I changed back to the original 1800 Bios: Also "optimal values" which disables cache burst. Is probably ultra slow.
But now the Win98 SE installation runs with 4*40 with the Ark 1000 VLB. omfg.
Never change the Bios, haha.

EDIT: Installation went through. L2 is enabled. L1 in WB. 160MHz.
Memory and cache timings are crappy.
After the installation I activated cache burst, but only x-2-2-2 for both read and write., that works. x-1-1-1 for both immediately crashed the Windows 98 boot process.
So timing is still bad.
Shall I try again the Sprint Bios 2200? I have no clue.
Also I am a bit curious, if it would work now with 128MB RAM from the beginning, but that is minor.
Not sure how to continue, maybe I try to get the network card running and transfer the usual programs: Philsbenchmark Suite, ....
So even the 2800 Bios does something wrong to the board, also the A-Bit Bios. Simply unstable.

I can't get L2 cache working on mine, mine lacks the voltage regulator though

These boards are not a picnic, but the number of problems @CoffeeOne mentions is out of ordinary.
Especially this last of line design was pretty ok here. No major drama or anything. Things just clicked.
Not sure what to think. Can pull mine and give it another spin. Do you want me to check something in particular - some verification that may put you on better path ?

pshipkov wrote on 2024-05-09, 03:16:

These boards are not a picnic, but the number of problems @CoffeeOne mentions is out of ordinary.
Especially this last of line design was pretty ok here. No major drama or anything. Things just clicked.
Not sure what to think. Can pull mine and give it another spin. Do you want me to check something in particular - some verification that may put you on better path ?

Hi.
Now after I made those tests, it is not sooooo bad.
The first major problem was to find out the correct settings for L1 WB. Not 100% sure, but I think I have it now.
The second major problem is that the newer BIOSes do not work: So neither the original 2800 nor the modified Abit Bios. So there must be a register setting inside the Bios, my board does not like.
The installation of Windows 98SE is a good stability test. I believe you did not have 100% stable settings in your testings with that board, because the Quake benchmark did not behave correctly.
Obviously your board must be different (at least slightly!), because your Board came and works with the 2800 one.

I changed now to the 2200 Bios which works, too. Timing is a tiny bit faster in comparison to the 1800. So the only real problem is that the performance values (I am talking of the speedsys showings for memory throughput) are bad. I will do some real world benchmarks soon.

EDIT: Now when I think about it:
I will try now the Bios of the Data Expert EXP4045, too 😁