I’ll give it a shot, just need to find a reliable vlb test board.

what VLB mobo you use now ?
the other day i scanned briefly ebay -there were at least 2-3 good ones in the $100-$150 range.

pshipkov wrote on 2022-02-20, 00:55:

what VLB mobo you use now ?
the other day i scanned briefly ebay -there were at least 2-3 good ones in the $100-$150 range.

By some stroke of luck, and some diligence, I’ve found myself with two GX4’s. I like having a backup, but I probably shouldn’t be testing random eBay junk on either.

Probably a dumb question by why can I see a 32GB partition as drive D in Windows but not in DOS? I’m using the onboard IDE controller of the Biostar 8433UUD.

That is a lot of luck these days.
But I suspect diligence paired with the right amount of green backs was the key. ; )

Even you got two VLIs now it is not wise to use them as a test site.

Latest DOS versions use FAT16 which can address up to 2Gb partitions.
Your 32Gb partition implies FAT32 or later, or *nix file allocation table - none of them are recognizible in DOS.

Few more notes:
Big clusters = bigger partitions.
Only one file can be stored in a cluster.
So big % of the clusters are wasted since large part of DOS era files are smaller than 32-64kb.
In other words even the partition is big in size, space will start coming down much more rapidly that smaller partition with smaller clusters.
For example, 4 512Mb partitions with 8kb clusters > 1 2Gb partition with 64kb clusters.
So, i use 510Mb partitions for DOS.

Remember back then running some kid-style analysis of the files on my hdd only to arrive to the conclusion that 8kb cluster size is optimal, which was already formulated by wise people. This reduces partition size to 512mb.
Anyhow.

pshipkov wrote on 2022-02-20, 17:54:

Latest DOS versions use FAT16 which can address up to 2Gb partitions.
Your 32Gb partition implies FAT32 or later, or *nix file allocation table - none of them are recognizible in DOS.

I can def create partitions greater than 2GB in DOS 7, up to 8GB I think. It’s curious because if I use a PCI Promise controller I can read the ~32GB partition as drive D in dos, at least I think I can, but it’s not recognized by the onboard controller. I’m sure we’ve had this conversation previously but I fail to remember the details.

Some good points on the cluster size.

Oh, we are talking diff problem here.
Dos 7 (win95) can see fat32 from dos shell, but the limitation is on the on board bios.
Need to verify that. Details escape me.

pshipkov wrote on 2022-02-21, 16:22:

Oh, we are talking diff problem here.
Dos 7 (win95) can see fat32 from dos shell, but the limitation is on the on board bios.
Need to verify that. Details escape me.

Yes I think you are correct something to do with the bios. I want to use the larger partition for ISO’s so this actually isn’t too much of an issue since Windows can see it.

Hi pshipkov, curious if you’ve tried any SIS based socket 3 PCI boards with 1024k cache or have I missed it?

Have 2 boards "documented" here.
Asus VL/I-486SV2GX4 revisions 2.0 and 2.1 (whole PC)
DTK PKM-0039S.

There were 2 more that are not with me anymore - "Green Board" and another nameless one.
They all worked will with 1024Kb L2 cache, but it was always trouble finding the right combination of chips for each one of them.

pshipkov wrote on 2022-02-23, 22:51:

Have 2 boards "documented" here. Asus VL/I-486SV2GX4 revisions 2.0 and 2.1 (whole PC) DTK PKM-0039S. […]
Show full quote

Have 2 boards "documented" here.
Asus VL/I-486SV2GX4 revisions 2.0 and 2.1 (whole PC)
DTK PKM-0039S.

There were 2 more that are not with me anymore - "Green Board" and another nameless one.
They all worked will with 1024Kb L2 cache, but it was always trouble finding the right combination of chips for each one of them.

Gotcha, I think specifically I was curious if any SIS boards with PCI slots did well with 1024k cache at 50MHz or better.

Wanted to share some results here with an AM5x86@200MHz (air cooling) on a Biostar UUD 3.1. It turns out that 50x4 beats 66x3 with my scenario. One in which the CPU:PCI divider of 1:1 is unstable at 66MHz. I’ve actually tried in two other SIS based boards with the same results. Both scenarios are very stable, the Biostar is a great board for stability once dialed in.

AMD 5x86 @200MHz (50x4)
Biostar UUD 3.1
CPU:PCI 1:1
256k L1 12ns (2111)
32MB EDO 60ns (0/0 wait states)
Riva 128 PCI (Screen corruption w/Tseng)
Voodoo2 12MB

Phils Dosbench:
3D Bench 1.0c: 117.7
Chris 3D Bench VGA: 86.8
Chris 3D Bench SVGA: 20.4
PCP VGA: 31
PCP SVGA: 10.2
Doom Max: 1110 Realtics
Quake: 21.7

GLQuake: 30.5
Turok: 23.1

AMD 5x86 @200MHz (66x3)
Biostar UUD 3.1
CPU:PCI 1:1/2
256k L1 12ns (3222)
32MB EDO 60ns (0/1 wait states) 1 on read
Tseng 6100
Voodoo2 12MB

Phils Dosbench:
3D Bench 1.0c: 100.1
Chris 3D Bench VGA: 85.9
Chris 3D Bench SVGA: 19.1
PCP VGA: 30
PCP SVGA: 9.4
Doom Max: 1316 Realtics
Quake: 21.2

GLQuake: 29.5
Turok: 21.9

Thanks for sharing.
Confirmed from my side.
UMC chipsets cannot do 3x66mhz reliably withuth cpu:pci divider which hurts perf, but makes the system stable.
Will link your post in the directory of the thread because until now this was not "documented" properly in one place but scattered in several posts here and in the fastest 486 thread.

Can you confirm if dos ide drivers work ok on the Biostar board ?
They were unstable for me at rx50 with tight timings, which halved the ide perf.

If you decide to base your 200mhz PC on the UUD mobo with 4x50 config, consider the 1024Kb L2 cache mode. It should be possible. Will be huge PITA to find the chips tho.

SIS based boards can reliably handle 3x66mhz with cpu:pci set to 1:1.
This is how I run it on LSD for example.

---

Spent time examining DRAM SPEED = FASTEST on LSD running at 180mhz with all BIOS settings on max, or 200mhz with 313 L2 cache timings.
We already confirmed that single LGS branded 16Mb 60ns rated EDO module can deliver that, but 2x16Mb is not fully stable.
It turns out that the memory size threshold is around 20 to 24 Mb. After that some issue on signal level starts to get in the way.
Tested a lot of configurations. Certain about it.

In addition to the finding above it is evident that these LGS modules are the best 72 pin SIMM memory for overclocking.

I don't know if this is a known fact, but after looking around about solutions for retro brightening of yellowed plastics and realizing that it will be work and trouble, i decided to first sanitize a Tandy monitor before consider my options for brightening it.
Rubbed few times the front panel with paper towel dipped in Acetone and ... see for yourself:

I'll take it - sanitization and brightening in one simple step.

The yellow tint is all transferred on the paper towel.

Acetone de-yellowed ABS with a single wipe? How hard did you press? After the acetone dried, did you take a close look at the surface finish with a magnifying glass or 10x gem magnifier? I'd be concerned that the textured element to the plastic, or even the sheen would be altered.

See for yourself:
video

Camera shake from unsteady hand may be misleading that i am applying a lot of force, but i am not pressing hard at all.
But i am sure that if i overdo the fine bumpy texture will eventually get smooth.

Had you already cleaned the surface of the ABS with isopropyl first to ensure none of that was dirt, dust, and grime?

once the acetone did its thing initially, i tried with alcohol thinking that it may be just dirt, but the alcohol did nothing to it.
then tried with wet cleaning wipes - no change.
acetone disolves the top layer of the plastic and the paper towel absorbs the yellow color revealing the beige underneath.
i am sure that under microscope we will be able to see difference in the plastic's texture, but there is no visible change to naked eye even after close inspection.

What about the feel of the plastic as you your finger across it?

I am shocked. I will have to do some tests with this. All this time I've been using a 400 watt grow lamp in the garage with a hydrogen peroxide salon cream. It is slow and the plastic wrap is tedious.

Might be one of those alchemical magic polar attraction things. ... though acetone doesn't look that far off from vinegar or isopropyl in that regard. Possibly it's restoring a lost acetyl bond or some crap, I dunno, too long since I took chemistry.