Let me check that really quick.

-edit- the jumper for IRQ9 was not bridged on the card, so I went ahead and did that. MSD.exe shows no difference in the IRQ reservations, and the HardMPU still reports the same error.

Okay I have tried to do a bunch of troubleshooting. It seems that IRQ is reserved by the BIOS, which is probably for the integrated video chip, a Cirrus Logic chip. I would guess that there is a jumper somewhere that I can disable that with? I cannot seem to find it though. And the excellent links shared here earlier don't seem to have a listing in the jumpers for it either. Hmmm.

Deksor wrote on 2021-09-06, 21:49:

Sure you can, use this program : http://cd.textfiles.com/microhaus/mhblackbox3 … MORY/GETROM.ZIP

Okay. It took me some time to get back to this but I have made a rom file from this BIOS now, I think? I followed the instructions and I have three files. An ATROM, TEMPODD and TEMPEVN. Which ones do you need?

Send all three, I'll figure it out (I think it's ATROM though).

Okay, I sent a PM with the files. Thank you.

Happy Weekend everyone. I hope you are all doing well. I am trying to tighten timings to squeeze more performance out of the 100MHz DX4 before trying to find something like a AMD 5x86-133 chip. My biggest bottleneck seems to be memory. I moved the Awe64 Value card to a different slot on the ISA bus and the bus noise has reduced. I thought that the integrated IDE and Floppy were on the VL Bus but I guess that I am incorrect in that thought. I have been trying to tighten memory timings so I pulled the memory out of the system. One of the sticks has HY514400A J-70 on it and the other says RAMS DS4740 AG-60 on it. I can confirm that the Hyundai ram is Fast Page Mode ram, but I have not been able to find anything regarding the DS4740 labeled ram. Strangely enough, I pulled the Hyundai ram out of the system, and the system still reports that there is 16MB installed, that means the Hyundai stick is bad, I think? Or could it be that the DS4740 is EDO and as such the sticks wouldn't run together? Anyway, I have tried to tighten memory and the system tends to lock up and crash when I try to set things low. The Cache is set to 3-1-1-1 I think? It has a setting for 2-1-1-1 or something but I cannot set that. I also tried to reduce the wait states and I can set the RAM wait states to 1 with no issues, but the Cache write wait states can't be moved from 2. I am wondering that if I replace the Cache with different, faster cache chips if that would allow me to tighten timings there. The speeds that are being reported within Speedsys for memory is as follows:

Cache Level 1 Read - 94.29MB/s Write - 31.83MB/s Move - 42.26MB/s Average - 56.12MB/s
Cache Level 2 Read - 51.05MB/s Write - 31.77MB/s Move - 30.04MB/s Average - 37.62MB/s
Memory Read - 17.04MB/s Write 32.02MB/s Move - 13.81MB/s Average - 20.96MB/s

I don't know if these are expected speeds or if they are low or high. Our 486 computer when I was younger was a store bought one and the first PC I built after that was a Pentium 100, so I didn't get to experience configuring these aspects back in the day. I look up the speed information online for FPM and EDO and I see that their max rates are significantly higher than what I am reporting, but I don't even know if I should be getting even remotely near those speeds. I have 128KB of Cache installed and it's UM61256K-15. I recognize that means that they are 15 nanoseconds and 256 Kilobit chips,. But that's about it. I looked up the datasheet to see if I could glean any more information that would be important for me, and I can understand the min/max ns wait times for read and write activity, but aside from that, I am just not educated enough in the other material provided. I did find that they are labeled as 32K x 8 Bit High Speed CMOS RAM. I have 128K more here somewhere, should I install that and increase my cache to 256? Will that improve anything? The machine runs okay for the most part, but there are games that recommend a DX4-100 and when they test the CPU it comes back as a DX-83 or a DX2-66. And this motherboard has jumpers for settings that I am not certain what they pertain to. jumpers for LDEV, which I guess could be Logical Device? But all I can truly ascertain about it, is that it seems to relate to the VL-Bus items. I know I have asked for a ton of help and I consistently keep asking, but any information or suggestions would be very much appreciated.

Thank you,

Nathan

nathanieltolbert wrote on 2021-10-24, 21:07:

One of the sticks has HY514400A J-70 on it and the other says RAMS DS4740 AG-60 on it. I can confirm that the Hyundai ram is Fast Page Mode ram, but I have not been able to find anything regarding the DS4740 labeled ram. Strangely enough, I pulled the Hyundai ram out of the system, and the system still reports that there is 16MB installed, that means the Hyundai stick is bad, I think? Or could it be that the DS4740 is EDO and as such the sticks wouldn't run together?

I don't recommend using noname RAMs.
What I recommend is using interleave, with two or 4 banks.
Does the Hyundai stick work if you pull out the noname stick?

retardware wrote on 2021-10-24, 22:34:

I don't recommend using noname RAMs.
What I recommend is using interleave, with two or 4 banks.
Does the Hyundai stick work if you pull out the noname stick?

The Hyundai stick has been tested. It does not work. I get a repeating low tone on a cyclical pattern of 2 seconds tone, 2 second pause.

I searched through various other sticks of RAM that I have. Sadly I have only 1 stick that is labelled as EDO and it's only 4MB. I checked my stash and I had a single 8MB stick, and two 4MB sticks, but one of them was FPM and the other was EDO. Then the 16MB stick. Is interleave something that is enabled in the BIOS? I can find no setting for it anywhere.

Okay, so since I had such a limited supply of 72 pin RAM, I went ahead and ordered some more online. I got a couple sets of 2x8MB EDO 72 Pin sticks. They should be here by the end of the week. I also managed to snag some 32Kx8 at 12ns Cache chips. Hopefully they will work fine in the board. No tag chip that I know of, so I guess I can't get them to go faster than 15ns. So going back to the question of interleaving on the system RAM, is this done automatically by the system without a setting, or is there a setting in the BIOS? I seem to recall seeing RAM interleave in later systems, like Pentiums and K6 systems, but I haven't seen anything similar in the BIOS for this interestingly strange board. Maybe it's under a different name? I am also curious about the fact that I can install 1MB of RAM to the board using SOJ chips. I have some on the way that are supposed to be EDO 512K chips at 50ns. I wonder if they will work with it? Hopefully the parts will be here by the end of this week and I will be feeling better. (I injured myself in a stupid way this last friday and I am slowly recovering.) Thank you again for all of the advice and tips. My system speed is setting really low right now and I just can't seem to figure out what I need to change to make it work faster.

Your L2 cache cannot be enabled without a tag chip. I hope you get more than 8 cache chips.

Disruptor wrote on 2021-10-26, 12:17:

Your L2 cache cannot be enabled without a tag chip. I hope you get more than 8 cache chips.

The board already has a tag chip installed. I know it's different than the cache chips that I bought, but I thought it would work in a pinch. I also got 15 chips. They are supposed to be NOS. I will know more when they arrive next monday.

Another thing of note. I have a feeling that something is set wrong, so I went and looked through some videos with benchmarks regarding an AMD 486DX4-100. The video I found the user had a chip with 16KB of L1 cache, but in System information he was getting a CPU rating of 216. I get 132. Which, isn't that closer to what you should get with a DX2-66? I'm at a loss. I wouldn't think that the extra 8KB of L1 would improve speed that much, but I am not fully familiar with this CPU. I will go over the jumpers again tomorrow and see if I still have something set wrong. Very strange.

Okay I have gotten some of the parts in. I took out the 15ns cache and replaced it with 256K of 12ns cache. I also got in the 16MB kit of EDO but I haven't installed it yet. I made a few more changes in the BIOS to match what I have seen online for other machines that appear to be similar to my board, and I have managed to get the CPU rating in System Information from a 132 to a 197. Quite a significant jump. It's not 216, but since I only have the DX4-100 with 8KB of L1 cache, I am happy I have gotten it to there. I also tried to reset the jumpers back to the 486DX2-66 and when I put my AMD 486DX2-66 back in, for whatever reason it sees it as a AM486DX4-Plus running at 100MHz. I checked and double checked the jumpers. I don't see anything that I may have missed. And it says that regardless of if I set the jumpers to AM486 or 80486. So progress is progress but I am a little confused why I can't get the board back to recognizing the DX2-66. I'm trying to see if I can get my hands on a 486DX4-100 16BGC from AMD because I am curious if this board will show any difference between the 8KB and 16KB L1 cache.

Can you tell me exactly which DX2-66 you have, and whether your DX4 is an AMD SV8B or NV8T or so?
Please run benchmarks like Speed Sys 4.78 with both processors. The result can store the screenshot in PCX pictures after running the benchmark. Perhaps you can convert them to PNG and upload them here.

I cannot bench the 486DX2-66 as I cannot get it to run as a DX2-66. I tried all different jumper settings from the three different boards which are supposed to have the same jumper settings and all of them identify the DX2-66 as a DX4-100. But the two chips are as follows -
AM486DX2-66 V16BGC and the AM486DX4-100NV8T

The DX2-66 if I understand fully is a 3.3V 16KB L1 cache chip. It ran fine at 66MHz, but I was too concerned to run it for even a minute with it overclocked to 100MHz
The DX4-100 only has 8KB of cache. I has no CPUID string if I try to get that set up. It just doesn't show anything.

All I can guess from the jumpers is that there is a jumper setting that I am missing that is not listed on jumper settings that I have available to me. One thing I noticed is two of the different revision jumper settings indicate a J35 that needs to be opened or closed. I have scoured my board and I cannot find a J35. That doesn't mean it doesn't exist, just that I cannot find it. I will try and get a screenshot of the NV8T in SpeedSys as soon as I have the time.

it isn't being overclocked to 100 (unless you somehow set the FSB to 50MHz)!

that's why people keep telling you to benchmark it, to demonstrate the fact it's still running at 66

the DX2 has a fixed multiplier of 2x

never trust the BIOS to correctly identify CPU speed

nathanieltolbert wrote on 2021-11-01, 13:43:

But the two chips are as follows - AM486DX2-66 V16BGC and the AM486DX4-100NV8T […]
Show full quote

But the two chips are as follows -
AM486DX2-66 V16BGC and the AM486DX4-100NV8T

The DX2-66 if I understand fully is a 3.3V 16KB L1 cache chip. It ran fine at 66MHz, but I was too concerned to run it for even a minute with it overclocked to 100MHz
The DX4-100 only has 8KB of cache. I has no CPUID string if I try to get that set up. It just doesn't show anything.

Here we go.
Your DX2-66 behaves like a DX4 and you may try the jumper settings for DX4. Reason: AM486DX2-66 V16BGC is a downlabelled Enhanced AMD 486 DX4
Your DX4-100 behaves like a DX2 and you may try the jumper settings for DX2. Reason: AM486DX4-100NV8T is basically a 3.3V DX2 with the core from old AMD 486 DX2. However, it may be tricky to find the 2X/3X jumper which may be the Write Through/Write Back jumper from the new 486 DX4 jumper settings.
You may give your DX2 a chance to run at 80 or 100 MHz. Since AMD did not produce DX2's in the die-shrinked process when moving to DX4 opposed to the markets continuing demands for a DX2 they relabelled some Enhanced DX4's to your special edition of a DX2. I'm not certain but I think this just was a marketing measure at this time.

Also, because DX2-80 was a speed grade introduced kinda late, setting a DX2-66 to 2x40 might get it guessed as a DX4-100 on some boards.

Disruptor wrote on 2021-11-01, 18:21:

Here we go. Your DX2-66 behaves like a DX4 and you may try the jumper settings for DX4. Reason: AM486DX2-66 V16BGC is a downlabe […]
Show full quote

Here we go.
Your DX2-66 behaves like a DX4 and you may try the jumper settings for DX4. Reason: AM486DX2-66 V16BGC is a downlabelled Enhanced AMD 486 DX4
Your DX4-100 behaves like a DX2 and you may try the jumper settings for DX2. Reason: AM486DX4-100NV8T is basically a 3.3V DX2 with the core from old AMD 486 DX2. However, it may be tricky to find the 2X/3X jumper which may be the Write Through/Write Back jumper from the new 486 DX4 jumper settings.
You may give your DX2 a chance to run at 80 or 100 MHz. Since AMD did not produce DX2's in the die-shrinked process when moving to DX4 opposed to the markets continuing demands for a DX2 they relabelled some Enhanced DX4's to your special edition of a DX2. I'm not certain but I think this just was a marketing measure at this time.

The reason I believe the DX2-66 to be OC'ed to 100MHz with the jumpers is because checking it with System Information the DX2-66 reports itself as running at 100MHz. It also shows as running at 100MHz with the CPU Check tool in the DOSBenchmark suite. The chip, which ran surprisingly cool when running at a reported speed of 66MHz was extremely hot to the touch when running at the reported speed of 100MHz. That is with a Heatsink and Fan attached to the CPU. Due to the extreme temperature on the chip, I haven't felt comfortable running it for the length of the time to test the CPU. I would rather not destroy the CPU if somehow it is indeed running improperly.

My wife said to go ahead and test it since I can find another one on ebay. I went ahead and did the speedsys test, and I ran CHKCPU. I checked, doubled checked and triple checked my jumpers and they match what the jumper manual says, but it still detects the DX2-66 as a DX4-100. I even checked with System Information 8.0 and it gets the same result as the actual DX4-100 AMD chip, which has half the L1 cache. I have the TXT and PCX files. And I will post them after I get them onto my desktop. I also took a picture showing that CHKCPU shows the DX2-66 as a DX4-100. You can see both the screen and the DX2-66 in the picture. The only thing I can think of is that the DX2-66 is binned that way because it didn't fully pass for DX4-100? I don't know.