60 MHz
JP15
1-2 = open
3-4 = short
5-6 = open

66 MHz
JP15
1-2 = short
3-4 = open
5-6 = short

Many thanks... will try later today.

Hey I’m subscribed to this, don’t let this die (after 6 years) 😜

I’ve got a Cyrix 5x86-100GP and IBM 5x86C-100GB in storage waiting for a good motherboard to show up on a rainy day. There are definitely still some curious / new 5x86 dreamers out there!

A newcomer to the hobby? There's a lot to learn. If you are trying to run a Cyrix or IBM 5x86 at 2x66 MHz, for 133 MHz, there are only two confirmed motherboards which will handle this - Biostar 8433UUD and Luckystar LS-486 rev.D, but with the later, floppy won't work at 66 Mhz FSB. From my experience, the best chips to run at 2x66 are QFP-based IBM 5x86c-100 chips, but they are hard to find on an interposer. Next best are ceramic IBM 5x86c-100HF chips. In either case, 3.7 V to the CPU is a good starting point. The ceramic chip might need more voltage than the QFP.

It is OK for threads to die if there isn't any new relevant information to supplement.

There may be a third option. There was a small quantity of Cyrix 5x86 made in 1996 with 2x/3x multipliers.

Anonymous Coward wrote on 2021-11-15, 08:03:

There may be a third option. There was a small quantity of Cyrix 5x86 made in 1996 with 2x/3x multipliers.

I think you meant 3x/4x multipliers. But with today's cost for a Cyrix 5x86-133/4x, you just as well leave the hobby before you get in too deep.

feipoa wrote on 2021-11-15, 06:43:

A newcomer to the hobby? There's a lot to learn. If you are trying to run a Cyrix or IBM 5x86 at 2x66 MHz, for 133 MHz, there are only two confirmed motherboards which will handle this - Biostar 8433UUD and Luckystar LS-486 rev.D, but with the later, floppy won't work at 66 Mhz FSB. From my experience, the best chips to run at 2x66 are QFP-based IBM 5x86c-100 chips, but they are hard to find on an interposer. Next best are ceramic IBM 5x86c-100HF chips. In either case, 3.7 V to the CPU is a good starting point. The ceramic chip might need more voltage than the QFP.

It is OK for threads to die if there isn't any new relevant information to supplement.

Let’s say newcomer to the 5x86 world. When I was about 15 this was the holy grail whispered only in hushed tones. My dad was never able to get his hands on one so we went straight from a DX2-66 to a Pentium II 233.

I missed out on a 8433UUD a few months back but was too busy on my dual Deschutes build anyway. There is a Belarusian LS-486 rev.D on the ‘bay right now but a) it’s expensive (though comes with an AMD X5-133 ADW thus saving me from sourcing one later on) and b) it has a fat globule of what looks like glue on one of the ICs.

I notice the two steppings I have aren’t mentioned much in this or the other 5x86 threads. Are they not really known for their overclocking abilities?

PS. Was only in mock-outrage about leaving the 6 year old thread to die 😉

pixelatedscraps wrote on 2021-11-15, 09:42:

There is a Belarusian LS-486 rev.D on the ‘bay right now but a) it’s expensive (though comes with an AMD X5-133 ADW thus saving me from sourcing one later on) and b) it has a fat globule of what looks like glue on one of the ICs.

I would avoid that one. It doesn't look to me like a standard LS-486 Rev.D. All the Rev.D's I've seen have had double-banked SOJ cache chips, while this Rev.D has single-banked DIP cache and in a location on the PCB that wasn't common for these boards. I'm not sure if it is a counterfeit design or if LuckyStar had two official Rev.D's, but in either case, I have not tested that Rev.D configuration and cannot confirm if the 66 Mhz FSB is stable.

feipoa wrote on 2021-11-15, 09:31:

Anonymous Coward wrote on 2021-11-15, 08:03:

There may be a third option. There was a small quantity of Cyrix 5x86 made in 1996 with 2x/3x multipliers.

I think you meant 3x/4x multipliers. But with today's cost for a Cyrix 5x86-133/4x, you just as well leave the hobby before you get in too deep.

No, it wasn't a typo. I really did mean 2x/3x.
Cyrix didn't make 2x/3x 5x86-133s though. They were only 100/120. I have a -120GP made in 1996 that is 2x/3x enabled. It is probably pretty uncommon, but it seems to handle 133 okay. I only tested with a 45MHz FSB though. I don't have any boards stable at 66.

I kind of remember talking about that now: 3x45 = 135 MHz. What voltage was required for stability and did you run it through a few Windows benchmarks or stress tests to ensure it didn't crash?

Anonymous Coward wrote on 2021-11-15, 13:27:

No, it wasn't a typo. I really did mean 2x/3x.
Cyrix didn't make 2x/3x 5x86-133s though. They were only 100/120. I have a -120GP made in 1996 that is 2x/3x enabled. It is probably pretty uncommon, but it seems to handle 133 okay. I only tested with a 45MHz FSB though. I don't have any boards stable at 66.

What do you mean? One can jumper any ordinary Cyrix 5x85-133 to 2x or 3x. Did that many times, and I only have those ordinary 133 Models.

Edit: Sorry I was thinking of the 5x86-100 not the 133. I only have 100MHz ones. The most ordinary ones.

I just left it running quake timedemo all afternoon. Pretty sure it was not more than 3.7V.

Gerwin, you are correct that 2x/3x chips are common, but as far as I can tell, pretty much all the Cyrix branded chips made in 1996 were 3x/4x enabled.
Most of the 133s were produced in 1996, so I would think that the -100s and -120s from that batch should have a better chance of overclocking to 133 than the ones made in 1995.

Put together a Cyrix 5x86-100 machine yesterday:

-QDI V4S471 with Sis471 chipset and 256KB cache
-Cyrix 5x86-100
-2x16MB 60ns 72pin memory
-Cirrus 5434 VLB with 2MB
-VIA VLB IDE -controller
-Yamaha OPL3-SA (YMF 701)
-Gravis Ultrasound Classic 1MB
-SMC Ethercard

Motherboard takes 50MHz with tightest cache and memory settings(!) so now using 2x multiplier.
CPU settings are BTB_EN on, LSSER off, FP_FAST on. Runs everything stable, even Windows 95.

Quake gives ~15fps. 3DBench1.0c gives ~87fps.

This CPU just won't take 120MHz with default voltage, not stable at all and I'm thinking of making voltage mod...

Before that, maybe I'll try Am5x86 with 3x multiplier and see how fast that is. That 50MHz bus speed really makes difference vs 33MHz.

If I remember CACHECHK numbers right it gave 5 for the L1, 13 for the L2 and 14-15 for the memory.

Edit: With 3,8V got it stable in dos but not in windows. 🙁 Very bad overclocker... But, it really hurts when FSB is dropped from 50MHz to 40MHz so I'm happy with 2x50MHz.

Edit2: needed 3,65V to get BTB stable in Windows. Stepping 1, Revision 3 CPU. Memory setting needs to be faster in BIOS, fastest setting is almost stable not quite. Maybe if I lift +5V rail little bit... Well, it's 14,8fps vs 14,3fps in Quake.

I've got:
a) Biostar MB-8433UUD-A Ver3 and Asus VL-I/486SV2GX4 boards
b) Cyrix 5x86-100GP and IBM 5x86C-100 (unsure of the stepping)
c) Cyrix 5x86-120GP

Once I get back from this business trip, I've got a date with a bench table and a couple of motherboards to fix up a few things before getting started on my 5x86 build. I managed to source a case for it that needs a bit of tidying up but has a very good (read: working and silent) EVER Pro Power PPS-230 230W PSU. Unfortunately, the case LED jumpers only go to 99Mhz but hoping I can find some kind of workaround to replace it with a 3-digit LED - suggestions welcome if any!

pixelatedscraps wrote on 2023-03-19, 14:00:

Unfortunately, the case LED jumpers only go to 99Mhz but hoping I can find some kind of workaround to replace it with a 3-digit LED - suggestions welcome if any!

My suspicion is that triple digit displays of the [1 through 199] type also fit in cases that originally came with a double digit display. For example:

You are lucky - they don't always fit, either the LCD square is in the wrong spot, is the wrong size, or the mounting screw is in the wrong position.

feipoa wrote on 2023-03-20, 17:39:

You are lucky - they don't always fit, either the LCD square is in the wrong spot, is the wrong size, or the mounting screw is in the wrong position.

My experience is limited. But the dimensions of the display in the second photo look like a 20x25 mm type could be made to fit ?

PARKE wrote on 2023-03-21, 12:12:

feipoa wrote on 2023-03-20, 17:39:

You are lucky - they don't always fit, either the LCD square is in the wrong spot, is the wrong size, or the mounting screw is in the wrong position.

My experience is limited. But the dimensions of the display in the second photo look like a 20x25 mm type could be made to fit ?

Can you link me to a specific part you think might work?

pixelatedscraps wrote on 2023-03-26, 15:23:

PARKE wrote on 2023-03-21, 12:12:

feipoa wrote on 2023-03-20, 17:39:

You are lucky - they don't always fit, either the LCD square is in the wrong spot, is the wrong size, or the mounting screw is in the wrong position.

My experience is limited. But the dimensions of the display in the second photo look like a 20x25 mm type could be made to fit ?

Can you link me to a specific part you think might work?

No. But as far as I understand the dimensions of the 2 digit x7 type are practically the same as the 2.5 digit x7 type. With bit of luck any 2.5 digit x7 display will fit in the window of your case.

PS
The one that I have in the case from the above photos is marked K-568 but I doubt that you can find them today using that type number.