Cant say that I have seen or not seen this before. I guess it was designed this way, because it did not matter were cards are mounted per se' and on these old systems, serial and parallel ports are on them pcb-less brackets. Guess this is just designed because they felt doing it this way. Just a thought that I have, no real proof of it anyway. I remember that back in the day's of 1989/1993, we used to mount EGA/VGA cards in the bottom slot. Then controllers in the second lowest slot. Then stuff like soundcards and CD-Rom controllers and finally the two serial/parallel brackets. This configuration scheme all changed with the rise of AGP cards, onboard devices and stuff like that.

clueless1 wrote:

I've been meaning to ask, was it fairly common for BIOS chips to block ISA slots? It seems all my ISA cards will not seat fully in slots 2 and 3 because of the AMIBIOS chip there.

I don't think I have any motherboards with BIOS chips in that place, but I have some with the keyboard controller chip there, but it's not very tall, so it doesn't impact the slots.

That BIOS chip doesn't look all that tall. It seems to be about the same height as the slots, but maybe the picture is misleading.

Anyway, you still have a lot of them free, so no worries.

brostenen wrote:

I remember that back in the day's of 1989/1993, we used to mount EGA/VGA cards in the bottom slot. Then controllers in the second lowest slot. Then stuff like soundcards and CD-Rom controllers and finally the two serial/parallel brackets.

Now that you say that, I remember exactly the same, though I remember putting the I/O as high up as possible to try and keep other cards from being on top of it since they would obstruct the I/O cables.

But yeah, with that configuration, you'd just have serial/parallel slot brackets in that area, so no harm, no foul.

I'm dealing with an interesting problem with this build. The case I'm using seems to have a grounding issue. Intermittently, it won't POST when in the case (usually after some movement or add/removal of components), but whenever I remove the components and run it on the bench, it POSTS every time. 😀

clueless1 wrote:

I've been meaning to ask, was it fairly common for BIOS chips to block ISA slots? It seems all my ISA cards will not seat fully in slots 2 and 3 because of the AMIBIOS chip there.

Yep, certainly NOT uncommon and indeed very annoying. Not that much back then when you only had the necessary cards and one could easily match the slots to their design, but today building a super PC of that era needing all the slots available/adding many cards it's an issue.

Found the source of the grounding issue: the tabs where the expansion card slot covers are supposed to touch when inserted were not making contact. I just had to bend them back up a bit so when you insert expansion cards in, the bottom of the slot cover is in contact with them. Whew!

I'm liking this system!
Similar to you I went from a Apple IIe which never really interested me at the time to a DX2 66 VLB around 1995, and got into computers in a big way.
I also lived in Genoa Italy for a short while so really want a video card for other nostalgic reasons 😀

Actually really like the black case as well! get a black 5 1/4 floppy drive and think it would look awesome
Looking forward to seeing the system up and running

chinny22 wrote:

I'm liking this system! Similar to you I went from a Apple IIe which never really interested me at the time to a DX2 66 VLB arou […]
Show full quote

I'm liking this system!
Similar to you I went from a Apple IIe which never really interested me at the time to a DX2 66 VLB around 1995, and got into computers in a big way.
I also lived in Genoa Italy for a short while so really want a video card for other nostalgic reasons 😀

Actually really like the black case as well! get a black 5 1/4 floppy drive and think it would look awesome
Looking forward to seeing the system up and running

So cool that you lived in Genoa for a bit! There's not a ton of info on this brand, but I do like that both of their components are in such good shape, and it turns out, perform admirably.

I was very tempted to move back to the black case after spending a couple of hours working out the grounding issue I was having with this beige one (it would go away when I ran the system on the bench, then come back when I assembled into the case). Really? I've never had such an issue until now. On the case, if you look on the opposite side of where the motherboard screws in,you will see little slits where the bottom of each expansion card comes through when you plug it into the board. There are little grounding tabs there that were bent to the point where they were making intermittent contact with the expansion card slot covers, which in turn would cause the system to not POST intermittently when I moved the system around a bit. Pushing those tabs up enough that they made solid contact fixed the problem.

Note the hard-to-see red rectangle outlining the problem area.

So I've got the system together (minus an ATX shield, mounted power switch, turbo, etc) and have been benchmarking it. I couldn't be happier with the performance so far. Here's a progression of BIOS cache timing adjustments:

I had stability issues with DRAM Write CAS at 1T, and since it made no difference in performance, I leave it at 2T. Same with DRAM Write Cycle at 0 W/S. Where I saw good increases was going to 'Fastest' DRAM Speed Option, dropping Cache Write Cycle from 3T to 2T, and finally dropping Cache Burst Read Cycle from 2T to 1T. Just with those three settings, Doom FPS went up 11.8%. If you look at Phil's VGA Benchmark Database, that improves this system from an average 486 to one of the fastest.

But, part of the magic of 486 systems is the variety of speed profiles you can run by manipulating Turbo, L1 and L2 cache, and BIOS timings. I'm not going to use this system to run demanding games. My Pentium already does this. I'm going to use the 486 to try to hit speed profiles that my Pentium can't. That includes 486 33-66Mhz and speeds between a fast 286 and a 386sx-40.

Temporarily, I've got the case power switch connected to the Turbo header, which lets me turn turbo on as long as I hold the power button in. 😀 Not very convenient for benchmarking Doom with caches disabled...

Anyway, some photos of its current state... (note the homage to a Vogoner who sold me an I/O card at a sane price)

It's not a grounding problem - at the latest when you tighten the screw for the slot plate it makes contact with the case and thus is grounded. The metal down there is electrically same as the rest of the case.

You have cracked solder joints, broken traces, corroded contacts or something along these lines - the little force caused by the bent plates will put it back to operational state. This is no reliable fix though.

The mainboard or affected expansion card have to be considered trash unless you find and fix the cause.

h-a-l-9000 wrote:

It's not a grounding problem - at the latest when you tighten the screw for the slot plate it makes contact with the case and thus is grounded. The metal down there is electrically same as the rest of the case.

You have cracked solder joints, broken traces, corroded contacts or something along these lines - the little force caused by the bent plates will put it back to operational state. This is no reliable fix though.

The mainboard or affected expansion card have to be considered trash unless you find and fix the cause.

Hmm. I'll keep an eye on it. But it's been working consistently since yesterday when I bent those tabs up, whereas it was failing regularly prior to that.

clueless1 wrote:

I couldn't be happier with the performance so far. Here's a progression of BIOS cache timing adjustments:

Very interesting! That's a very healthy increase in performance.

clueless1 wrote:

But, part of the magic of 486 systems is the variety of speed profiles you can run by manipulating Turbo, L1 and L2 cache, and BIOS timings...

That's even more interesting! I don't think the Turbo button does anything in my late PCI 486 motherboard, but it's definitely working in your case. Also interesting is the not so great impact disabling the L2 cache produced. That explains why my first ever PC (a Compaq Presario with 486DX2-66) didn't have any cache. They could save costs for a little performance decrease that wouldn't be noticeable to the naked eye. If they improved the memory and L1 cache timings like you did then it was possible to maybe offset the lack of L2 cache and have the system performing at the level of other similar PCs.

Great job with those benchmarks 😀

devius wrote:

clueless1 wrote:

But, part of the magic of 486 systems is the variety of speed profiles you can run by manipulating Turbo, L1 and L2 cache, and BIOS timings...

That's even more interesting! I don't think the Turbo button does anything in my late PCI 486 motherboard, but it's definitely working in your case. Also interesting is the not so great impact disabling the L2 cache produced. That explains why my first ever PC (a Compaq Presario with 486DX2-66) didn't have any cache. They could save costs for a little performance decrease that wouldn't be noticeable to the naked eye. If they improved the memory and L1 cache timings like you did then it was possible to maybe offset the lack of L2 cache and have the system performing at the level of other similar PCs.

Great job with those benchmarks 😀

Thanks. 😀 I discovered that when I disable L1, L2 AND hold the Turbo button in, the speed drops to what I guess is low to mid 286 speeds (maybe slower?). I base that on the fact that Ultima 2 and Might and Magic 1 both run at normal speeds, which I've never witnessed before on my systems. You can actually see the intro scenes when loading the games. 🤣.

I expected that when caches were disabled, that the Turbo button would have no effect.

edit: this system bottoms out at 1.8 fps in 3dbench, 0.4 fps in pcpbench, and 1.95 in speedsys. I'm not sure I have the constitution to stick out a Doom run.

edit2: speedtst score is 10563 (3.9 XT), which is between a i8088-9.54 and an i80286-10.

Norton Sysinfo 6 score is 3.3. A 286/8 gets 4.4. 😀

Here's official benchmark results of all the different performance levels with disabling combinations of L1, L2 and Turbo. That .65 fps Doom run was pretty painful to endure. 😵

^Wow, that's a pretty smooth sweep down the range.

TheMobRules wrote:

Another option you can consider for the power switch is to replace the momentary ATX power button of the case with a self-lockin […]
Show full quote

Another option you can consider for the power switch is to replace the momentary ATX power button of the case with a self-locking on/off button of the same type, and then solder the new button to the adapter green & black wires.

Of course, this depends on the type of button your case uses, but I have noticed that many ATX cases use these little 8mm square buttons:

http://www.ebay.com/itm/10Pcs-8x8-mm-Square-B … R0AAOSwqu9VPR2Q#

If you can do this, it would allow you to cleanly re-use your case power button without having to drill any extra holes. For the turbo button you would need another kind of solution though, such as what brostenen suggested.

I ended up ordering these (paid $3 shipped from US seller instead of $1 shipped from overseas):
http://www.ebay.com/itm/15-x-Push-Button-Swit … PoAAOxycmBSvdh2
The switches in my case are 8x8mm, so I'm hoping these work without problems. I plan on putting one at the power switch (obviously) and wiring the Turbo header to the RESET switch. I see no reason why I need both power and reset switches 😉 At least this way I'll be able to close up the case and have it look clean. Will report back after the install is finished.

is there any use for the lowest rating there? what game that You play needs that?
also: its my guess the results without l2 cache will very strongly depend on memory config, no?

mrau wrote:

is there any use for the lowest rating there? what game that You play needs that?
also: its my guess the results without l2 cache will very strongly depend on memory config, no?

Here are the games that I have that need those slowest speeds:
Lode Runner (1983)
Might and Magic 1 & 2 (1986, 1988)
Ultimas 1-5 (1981-1988)

Not sure what you mean by the other question.

what i mean is that without cache memory speed will decide overall system speed, so more fine tuning should be possible?

Yes. Here ya go 😀

The slower the memory timings are set in the BIOS, the bigger the slowdown when you disable L2. These are based on Doom scores.

So yes, I have a performance profile for each BIOS memory timing:
Fastest
Faster
Default
Slower
Slowest

Fastest+ is the Fastest memory timings plus the tightest timings on Cache Burst and Cache Write Cycle.

nice
when looking at these numbers i wonder if processors even improved over time except for getting caches and wider machine word
imagine what happens on a xeon processor when suddenly 40MB of cache get turned off 😁