The L2 cache size will determine the maximum "cacheable area" but will have NO influence on how much RAM is being detected by BIOS or the OS.

The SIS chipset is capable of addressing 512MB RAM I suppose?

SiS 5511 should support 512MB. Manual of the board (Asus still has it online) says:

"Uses 72-pin SIMM DRAM modules of 1MB to 128MB in multiple configurations up to 512MB"

Now, it's not explicitly mentioned, but to get 128MB you need double-sided SIMMs with 16 64Mb chips. If your chips are bigger, the memory controller will only be able to address the first 64Mb. It would help if you could indicate which 128MB SIMMs you are trying to use - in particular which chips are on it and how many of them there are.

I have the same board and I was changing asynchronous SRAM for synchronous COAST - all I did was shorted pins 2-3 at JP36 and JP37 to disconnect socketed SRAM. I have a COAST size of 256kB, so I left the jumpers JP10 and JP11 the same, you'll have to set them to a size of 512kB if you're going from 256kB.

Unrecognized RAM seems to be the problem here rather than unrecognized L2 cache - as far as I understand the opening post.

Ydee wrote on 2022-07-17, 14:09:

I have the same board and I was changing asynchronous SRAM for synchronous COAST - all I did was shorted pins 2-3 at JP36 and JP37 to disconnect socketed SRAM. I have a COAST size of 256kB, so I left the jumpers JP10 and JP11 the same, you'll have to set them to a size of 512kB if you're going from 256kB.

Thanks for the info. I was mostly concerned about possibly damaging the COASt module, but finding info on specs for voltage or anything really substantial beyond what it is supposed to 'basically' do is nearly non-existent. I have about a dozen project builds I work on, but this board/project has proven utterly frustrating at times. Generally I have little issue with ASUS motherboards and manuals are usually pretty straight-forward and understandable. I bought the board because it had most everything I was looking for (Pentium 200 support, 2032 CMOS battery, PCI/ISA slots). I was hoping to kill 2 birds with one stone with this upgrade, but sounds like this will do nothing for my memory issue. I am still hopeful this resolves the 64MB bottleneck. From what I have seen from some videos, it will also raise my available conventional memory from 64 to 96KB, which will be useful for some of the DOS games I have installed that require 560KB free (such as Duke Nukem 2).

I've written multiple options through config.sys & autoexec.bat, but if I could free enough conventional memory to eliminate the need to drop down into DOS, I could effectively eliminate memory management that locks my computer when CTCM tries to load (I created the multiple options bypass CTCM for the most part), and just go with a simpler config/autoexec loadup at start.

dionb wrote on 2022-07-17, 12:36:

SiS 5511 should support 512MB. Manual of the board (Asus still has it online) says:

"Uses 72-pin SIMM DRAM modules of 1MB to 128MB in multiple configurations up to 512MB"

Now, it's not explicitly mentioned, but to get 128MB you need double-sided SIMMs with 16 64Mb chips. If your chips are bigger, the memory controller will only be able to address the first 64Mb. It would help if you could indicate which 128MB SIMMs you are trying to use - in particular which chips are on it and how many of them there are.

The memory currently on the board: SEC KM44C16100-AK6 (8 each side, 16 total) & SEC KM41C1600BK-6 (4 each side, 8 total). This memory does not lock the computer at startup, just doesn't recognize anything above 256MB.

The other memory tested is: MT4LC16M4H9-5 (8 each side, 16 total). This memory locks the computer during POST as soon as it attempts to read above 256MB.

I have tested each stick with MemTest minimum 8 passes. All sticks have passed. when used individually, or paired in Bank 0, all memory detects properly (128MB with 1 and 256MB paired). As initially stated, I've tried this on a second identical board and the results are the same. I am thinking this is either memory that it isn't 100% compatible for some reason, or the manual is in error and 256MB is the limit (and all other subsequent claims on other sites is solely referenced from the manual). A 486 board I have claims it can read 32MB RAM sticks up to a maximum of 64MB, and perhaps this is also applicable to this board

Has anyone else attempted above 256MB on this board?

Unfortunately I can't help more, I personally have a 2x16mb edo dram with P133 on the board and use it mainly for MS-DOS and W3.1, so this size is quite enough for me. I don't have bigger modules available, so I can't test.

My guess is this is a hardware problem.

First check ALL contacts of the SIMM slots very carefully for dirt, corrosion and bent pins. They all need to have sufficient tension also.
Then check all traces around the memory controller, SIMM-slots and the cache-sockets on the backside very carefully for scratches and shorts.
Probe every single pin of the memory controller (red box) for loose contacts with a toothpick / or needle.
If there´s no issue to be found check all inputs and outputs of the buffer chips (blue box) with an Ohmmeter.
Edit: The buffers seem to belong to the IDE-ports, so forget this point.

During all that remove all RAM sticks and cache-chips / COAST stick.

falcon2051@msn.com wrote on 2022-07-17, 22:42:

[...]

The memory currently on the board: SEC KM44C16100-AK6 (8 each side, 16 total) & SEC KM41C1600BK-6 (4 each side, 8 total). This memory does not lock the computer at startup, just doesn't recognize anything above 256MB.

The other memory tested is: MT4LC16M4H9-5 (8 each side, 16 total). This memory locks the computer during POST as soon as it attempts to read above 256MB.

That's 64Mb EDO DRAM, so density isn't an issue - but according to datasheet it's 3.3V+-0.3V, so not 5V tolerant. 72p SIMMs in PC systems are driven at 5V. No wonder those SIMMs aren't happy...

You need 5V SIMMs.

Ydee wrote on 2022-07-18, 07:42:

Unfortunately I can't help more, I personally have a 2x16mb edo dram with P133 on the board and use it mainly for MS-DOS and W3.1, so this size is quite enough for me. I don't have bigger modules available, so I can't test.

Appreciate the response. I'm running a 486DX/4-100 for my DOS/Win 3.11 setup (originally ran 64MB RAM - because I could - but for DOS it was kinda overkill). This board is my Win 95 setup and I wanted to see if I could push the board to its limit RAM-wise (would probably be fine with 128MB - but I wanted to see how far I could push it)... and perhaps I already am.

I had to fight with Pentium II rig running Win 98SE when I installed 1GB RAM in it, but I was able to win that battle. But this is an altogether different matter as the issue begins at POST. I know my TUSL2-C board on my Pentium III is limited by the chipset to 512MB. Like I've stated before, I have several retro build projects, and I've always been able to figure out the issue before now. This is actually my first time posting a question online because I am simply stumped. The manual COULD be in error, but I would have thought someone would have run into an issue and found some kinda solution in the past 26 years these boards have existed, 🤣.

dionb wrote on 2022-07-18, 08:41:

falcon2051@msn.com wrote on 2022-07-17, 22:42:

[...]

The memory currently on the board: SEC KM44C16100-AK6 (8 each side, 16 total) & SEC KM41C1600BK-6 (4 each side, 8 total). This memory does not lock the computer at startup, just doesn't recognize anything above 256MB.

The other memory tested is: MT4LC16M4H9-5 (8 each side, 16 total). This memory locks the computer during POST as soon as it attempts to read above 256MB.

That's 64Mb EDO DRAM, so density isn't an issue - but according to datasheet it's 3.3V+-0.3V, so not 5V tolerant. 72p SIMMs in PC systems are driven at 5V. No wonder those SIMMs aren't happy...

You need 5V SIMMs.

That's new to me. Okay, thanks for the info. Got the memory from an online seller claiming to have RAM specific to the board. Since ASUS website doesn't possess any recommendations for RAM, can you make any recommendations of the RAM chips I should be seeking out in order to get better compatibility?

majestyk wrote on 2022-07-18, 08:36:

My guess is this is a hardware problem. […]
Show full quote

My guess is this is a hardware problem.

First check ALL contacts of the SIMM slots very carefully for dirt, corrosion and bent pins. They all need to have sufficient tension also.
Then check all traces around the memory controller, SIMM-slots and the cache-sockets on the backside very carefully for scratches and shorts.
Probe every single pin of the memory controller (red box) for loose contacts with a toothpick / or needle.
If there´s no issue to be found check all inputs and outputs of the buffer chips (blue box) with an Ohmmeter.
Edit: The buffers seem to belong to the IDE-ports, so forget this point.

p55yy.jpg

During all that remove all RAM sticks and cache-chips / COAST stick.

I will definitely check those out (though I have looked at the RAM slots like you suggest, I have not considered checking the controller chip). It would be odd (though not impossible) for both my test board and main board to suffer from some form of defect (they are identical, including Rev 1.4 & BIOS version 0312). Thank you for these suggestions.

falcon2051@msn.com wrote on 2022-07-18, 08:58:

[...]

That's new to me. Okay, thanks for the info. Got the memory from an online seller claiming to have RAM specific to the board. Since ASUS website doesn't possess any recommendations for RAM, can you make any recommendations of the RAM chips I should be seeking out in order to get better compatibility?

Asus' incredibly verbose manual is lacking on technical details. But even though the SiS 5511 supports both 5V and 3.3V, it's safe to assume a 72p SIMM slot in a regular PC motherboard is 5V only. In case of doubt you could measure the relevant pins (10 and 30 are VCC iirc) in the SIMM slot.

So what you need is 5V, or at least 5V-tolerant chips on your 128MB SIMMs. Recommendation is simple: don't trust a seller, look at the datasheet of chips on SIMMs you are considering, or the SIMM datasheet - although I'd generally trust chip datasheet more.

Interestingly, the chips on the SIMMs you already have are 5V ("TTL Compatible Inputs and Outputs") but also FP, not EDO. The SiS 5511 supports both (SiS was the first vendor to support EDO, with their previous 501 chipset), but mixing them can cause problems. So don't test with one pair of FP and one pair of EDO.