VOGONS


Reply 20 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I decided to try the 4DPS board first because it has specific mention of the Cx486S. The board boots with Cx486S solo, however DOOM hangs upon loading. Cx486DX runs DOOM fine in the same board. Then tried Cx486S+Cx487S, Landmark doesn't show any FPU value, indicating that the FPU isn't functioning. DOOM also won't load. At least this board booted with the Cx486S, which implies that the CPU is working.

Moving on to another board. I also have revision 1.7 of the DX-6900 (PC Chips M912?) and will try this board now. I also have revision 1.4 of this board.

Plan your life wisely, you'll be dead before you know it.

Reply 21 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I have the original printed manual for this motherboard, however, I do not see jumper settings for the Cx486S (or U5S). Which jumper settings did you use for Cx486S and Cx486S+Cx487S?

EDIT: Looks like those settings are only silkscreened onto the motherboard.

Plan your life wisely, you'll be dead before you know it.

Reply 22 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I tried the Cx486S + Cx487S on, both, the Cx486DX and Cx486S settings, however the system just sits at the start screen, the one saying press Delete to enter BIOS. Pressing Delete or Esc does nothing.

If I do not use the Cx487S adapter, the screen stays blank after power-on...

EDIT: OK, looks like the Dec 1995X BIOS for this board doesn't like the Cx486S chips. If I put in the original BIOS chip into my v1.4 board, dated 7/8/1994, I am able to boot with the Cx486S and run DOOM. This is more than any other motherboard I've tried so far. Unfortunately, this BIOS version puts the L1 cache into write-through mode.

So I tried putting the BIOS from my v1.7 board into my v1.4 board, however I am unable to boot with the Cx486S. I guess I will try my v1.7 board with the v1.7 BIOS. From your screen shot, looks like you have BIOS 7/25/1994. Have you tried the 12/1995X BIOS? I don't remember where I obtained it.

Plan your life wisely, you'll be dead before you know it.

Reply 23 of 60, by Nvm1

User metadata
Rank Member
Rank
Member

I am at work with a terrible headache today so I will try it later and let you know asap feipoa.
I did not try any other bios, I mainly was testing what hardware was in working condition and which not. The Cyrix 486 cpus somehow had issues with making good contact with the sockets. I had to clean and correct/bed pins like a 1000 times before it worked reliable... And still then some motherboards refused to work with them. The Cx486DX2-40 with write-back cache was even worse then the Cx486s to get it working.
So far 3 dead motherboards, 4 dead cpu's en 4 dead graphic cards, and I still have to test more then half of it...

Reply 24 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I can't imagine a CPU being more problematic to get working than the Cx486S. In all my testing, this has been the worst chip. However, once I get the Cx486S and Cx487S working, I'm going to attempt to get the SXL2 working on the socket 3. That will probably be the most difficult.

Why is the Cx486DX2-40 so difficult to get working? On all the boards I've tested, the Cx486DX-40 and Cx486DX2-66 have worked fine. Fo the DX2-40, don't you just need to set the FSB for 20 MHz?

Plan your life wisely, you'll be dead before you know it.

Reply 25 of 60, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Cyrix DX2-40? Are you sure you don't mean DX-40? I know intel had a DX2-40, but it was mostly found in portables.

Hmm...there is also the ST486 DX2-40. That should be Cyrix based.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 26 of 60, by Nvm1

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote:

Cyrix DX2-40? Are you sure you don't mean DX-40? I know intel had a DX2-40, but it was mostly found in portables.

Hmm...there is also the ST486 DX2-40. That should be Cyrix based.

I meant the DX2-50 with Write-Back (and this printed on the heatsink). It seemed a combination of incompabilities and bad contact of the pins with the socket.. Two boards Always worked with it, others intermittent and most not - atleast not with jumper settings I tried.

Reply 27 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

NV1, is this a decently good motherboard in your opinion? So far, I've only compared the results with the HOT-433 and the M912 tests faster. Its the only board I've tested to work properly with the Cx486S. Do you have a system based around the M912?

Plan your life wisely, you'll be dead before you know it.

Reply 28 of 60, by Nvm1

User metadata
Rank Member
Rank
Member
feipoa wrote:

NV1, is this a decently good motherboard in your opinion? So far, I've only compared the results with the HOT-433 and the M912 tests faster. Its the only board I've tested to work properly with the Cx486S. Do you have a system based around the M912?

I don't have a working system around at the moment but it was the board that did work with all cpu's I threw at it (and correctly identifies them) and also worked with all VLB and ISA cards I tested. Despite the PCChips name it was easy to set up and well, just worked 🤣 I never had the time to benchmark them in the process.
Only drawback I found is that it is a bit thin, so you need to mount it with enough support otherwise it bends.

Reply 29 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

That is a good point. This board is abnormally thin in comparison to all others I have.

Success! I was able to use the Cx487S adapter with the Cx486S CPU at 40 MHz. I am using the M912 v1.4 motherboard (I'll try v1.7 later). Landmark shows an FPU benchmark speed! Interestingly, I was able to use the Cx486S + Cx487S with, both, the Cx486S (M6) and Cx486DX (M7) jumper settings.

Plan your life wisely, you'll be dead before you know it.

Reply 30 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Nvm1: What is the Release date of the BIOS you are using? On my rev 1.7 board, I have AMI Core from 7/25/1994 (same as you) and Release 12/02/1994. I am using the same CPU jumper settings as you, however I cannot get rev 1.7 to boot with the Cx486S installed. I tried Cx486S with and without Cx487S.

Plan your life wisely, you'll be dead before you know it.

Reply 31 of 60, by Nvm1

User metadata
Rank Member
Rank
Member
feipoa wrote:

Nvm1: What is the Release date of the BIOS you are using? On my rev 1.7 board, I have AMI Core from 7/25/1994 (same as you) and Release 12/02/1994. I am using the same CPU jumper settings as you, however I cannot get rev 1.7 to boot with the Cx486S installed. I tried Cx486S with and without Cx487S.

I will dig the board up and let you know when I have 😀

Reply 32 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Great, thanks!

I think I have isolated the issue. In the BIOS for the rev 1.7 motherboard, there is an option to set the L1 cache in write-back or write-through mode. The system doesn't like to be set in write-back mode when I am using a SCSI controller. If I set to write-through mode, I can boot fine. Alternately, if I use an IDE controller, I can set L1 to write-back mode. I ran some benchmarks and having the L1 cache set to write-back mode did not alter the benchmarks at all. So I wonder if this motherboard doesn't fully support L1 in write-back mode. Have you tried running an Intel DX4-100, Am5x86, or Cyrix 5x86 with L1 in write-back mode?

On my Asus VL/I-486SV2GX4 motherboard, a BIOS update was needed to get L1 to work properly with write-back mode with Am5x86 chips. And even then, I must use a SCSI controller which has a special jumper to allow the CPU to work in write-back mode.

Although I have not tested it as such, I wonder if the newer 12/1995X BIOS for the M912 allows for using L1: Write-back. If so, this may be why I wasn't able to get the Cx486S or Cx486DX working.

Plan your life wisely, you'll be dead before you know it.

Reply 33 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I have begun mapping out the SXL2-relevant pins on the M912's PGA socket, in particular, S4, B16, C15, A13, A12, R17, and A15. These are the pins that the SXL data book mentions have some difference between the i486SX and the SXL2. I have done this now. I also want to check where these pins go on a board known to work with the SXL2, then try to copy over the working mapping onto the M912 board. Not having the pinouts for the M912's chipset will be an issue, but hopefully there shouldn't be much difference between the 486S and the SXL2.

The purpose of this exercise was to see if my SXL2-66 chips run properly at 80 MHz. So here's the snag... On boards which support the SXL2, the clock input is a 2x clock source, e.g. 80 MHz. On 386 motherboards, the crystal oscillator frequency gets halved by the CPU and the CPU runs at 40 MHz. On the SXL2, when you put it in 2x mode, it is actually taking away the halving. If I were to get the SXL2 running on the M912, it would run at 20 MHz by default, then when put in clock double mode, it would run at only 40 MHz. This pretty much defeats the purpose of getting the SXL2 running on the M912. The only way to get the SXL2 running at 80 MHz on the M912 would be if the PLL clock generator on the M912 contained a 2x output mode. If it did, I would have to rewire the system to use 2X to the CPU, and 1X to the VL bus and everything else. And that is assuming the UM9515 PLL clock generator on the M912 contained a 2X output mode and the signal was clean enough for use. I doubt the traces on this board were designed for anything past 50 MHz.

I assume that hybrid PGA132 / PGA168 boards have some jumper which selects 1X clock or 2X clock, or they always have 1X clock sent to the PGA168 socket and 2X clock sent to the PGA132 socket. This is probably part of the reason that we don't see any 486 boards with SXL2-PGA168 support.

Nonetheless, it would still be modestly fun to get the SXL2 running at 40 MHz on a socket 3 as merely a challenge exercise.

Plan your life wisely, you'll be dead before you know it.

Reply 34 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Mapping finished. With any luck, all I need to do is wire pin A13 (NA#) to a 10K resistor to Vcc, though I'm not hopeful.

I also need to wire S4 to BUSY#, which I don't know what pin on the M912's chipset is BUSY#, or if there is even an analogous pin to BUSY#. This is required for self-test, which I assume the motherboard is wanting to perform.

Plan your life wisely, you'll be dead before you know it.

Reply 35 of 60, by debs3759

User metadata
Rank Oldbie
Rank
Oldbie

According to a comment on http://www.cpu-world.com/CPUs/80486/Texas%20I … 486SXL2-50.html the TI486SXL had a 486SX pinout.

See my graphics card database at www.gpuzoo.com
Constantly being worked on. Feel free to message me with any corrections or details of cards you would like me to research and add.

Reply 36 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

That comment was made by a user here, also known as Anonymous Coward. He made the comment before these tests were done. If interested, read Appendix D in the Texas Instruments 486SXL databook. It details the differences between the SXL and the 486SX.

Thus far, I have been unable to get the SXL working on any board using the i486SX jumper settings.

Plan your life wisely, you'll be dead before you know it.

Reply 37 of 60, by debs3759

User metadata
Rank Oldbie
Rank
Oldbie

I've never found the 486SXL databook. Any chance you can upload a copy, to be added to my library?

See my graphics card database at www.gpuzoo.com
Constantly being worked on. Feel free to message me with any corrections or details of cards you would like me to research and add.

Reply 38 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Unfortunately, the pdf is too large to attach in its entirety. The full PDF is 19 MB and there is a 5 MB upload limit. I have attached just Appendix D from the SXL reference guide.

Attachments

Plan your life wisely, you'll be dead before you know it.

Reply 39 of 60, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Well, I wasn't able to get the TI486SXL2-66 to show any life on my socket 3 after wiring A13 to 10K to Vcc. I tried with, both, the i486SX and Cyrix 486S settings. There simply are too many pin differences that I am not able to solder fix because I do not have the spec sheet for the chipset on the M912. It looks like the following modifications would be necessary:

PGA168 socket
S4 wire to BUSY# on chipset
A13 wire to NA# on chipset (or 100-ohm to GND, or 10 Kohm to Vcc, depending on pipelining support of chipset)
A15 wire to NMI on chipset
R17 wire to PEREQ on chipset
BLAST# on VLB slot wire to 10 Kohm to Vss (BLAST# currently goes to 10 Kohm to Vcc)
B16 wire to MEMW# on ISA slot
A12 wire to ERROR# on chipset

Plan your life wisely, you'll be dead before you know it.