Retro Review: Lucky Star 5I-TX2A Motherboard - Part 2
20th April 2025
In Part 1 we explored this Intel 430TX-based motherboard from 1997 and got it powered up. In this part 2, we'll run some performance benchmarks using the IBM 6x86MX PR233 CPU that came with it.
Here's a picture of the motherboard again:
Lucky Star 5I-TX2A "Lyra" ATX motherboard (1997)
The Test Setup
For initial benchmark testing, I'll be using the IBM (Cyrix) 6x86MX PR233 which is pretty much the top end of CPUs this motherboard can support. The user manual actually doesn't include a configuration for the 6x86MX beyond 200 MHz, so this is somewhat undocumented. I will play around with the jumpers to see what other configurations will work. The 6x86MX actually runs at a clock speed of 187 MHz (2.5 x FSB of 75 MHz) but was given a PR (Pentium Rating) value of 233 as it more closely matches an Intel Pentium running at 233 MHz on 'instructions per cycle'.
I flashed the BIOS to the latest unofficial patched version from the WIMS BIOS site, which used BIOS version O2-006 as its base. This worked just fine, though I'm using a very small CF card as a hard disk so cannot confirm the 32 GB bug has gone. I installed a single stick of 64 MB PC100 SDRAM installed in bank 0, a CF-to-IDE card interface with several FAT16-formatted cards, and a number of graphics cards.
According to the motherboard's user manual, the jumpers to configure the CPU and FSB are as follows:
| CPU Speed | FSB Speed | Mult. | JP2 | JP3 | JP4 | JP7 |
|---|---|---|---|---|---|---|
| 166 MHz | 60 MHz | 2.5x | 2-3 | 1-2 | 1-2 | 1-2 and 3-4 |
| 166 MHz | 66 MHz | 2x | 1-2 | 1-2 | 2-3 | 1-2 |
| 200 MHz | 75 MHz | 2x | 2-3 | 2-3 | 1-2 | 1-2 |
The manual's 'CPU Speed' column is actually the 'PR' rating of the CPU, not the actual internal clock in MHz, so don't set this by the actual clock speed or you'll have a much slower system on your hands. I spent some time working out what each of the many missing jumper configurations are, along with some synthetic benchmark results:
| CPU | Actual Clock Spd | FSB | Mult. | JP2 | JP3 | JP4 | JP7 | Dhrystones | Whetstones | Speedsys |
|---|---|---|---|---|---|---|---|---|---|---|
| Cx686MX PR133 | 100 MHz | 50 MHz | 2x | 1-2 | 1-2 | 1-2 | 1-2 | 93,056 | 23,424 | 72.02 |
| Cx686MX PR133 | 110 MHz | 55 MHz | 2x | 1-2 | 2-3 | 1-2 | 1-2 | 101,834 | 25,410 | 79.11 |
| Cx686MX PR150 | 120 MHz | 60 MHz | 2x | 2-3 | 1-2 | 1-2 | 1-2 | 111,064 | 27,502 | 86.29 |
| Cx686MX PR166 | 133 MHz | 66 MHz | 2x | 1-2 | 1-2 | 2-3 | 1-2 | 122,670 | 30,201 | 95.70 |
| Cx686MX PR200 | 150 MHz | 75 MHz | 2x | 2-3 | 2-3 | 1-2 | 1-2 | 137,237 | 33,654 | 107.58 |
| Cx686MX PR233 | 166 MHz | 83 MHz | 2x | 1-2 | 2-3 | 2-3 | 1-2 | 152,642 | 37,218 | 119.73 |
| (no POST) | - | - | - | 2-3 | 2-3 | 2-3 | 1-2 | - | - | - |
| (no POST) | - | - | - | 2-3 | 1-2 | 2-3 | 1-2 | - | - | - |
| Cx686MX PR150 | 125 MHz | 50 MHz | 2.5x | 1-2 | 1-2 | 1-2 | 1-2 and 3-4 | 115,646 | 28,587 | 90.03 |
| Cx686MX PR150 | 138 MHz | 55 MHz | 2.5x | 1-2 | 2-3 | 1-2 | 1-2 and 3-4 | 126,680 | 31,126 | 98.89 |
| Cx686MX PR166 | 150 MHz | 60 MHz | 2.5x | 2-3 | 1-2 | 1-2 | 1-2 and 3-4 | 137,672 | 33,716 | 107.87 |
| Cx686MX PR200 | 166 MHz | 66 MHz | 2.5x | 1-2 | 1-2 | 2-3 | 1-2 and 3-4 | 152,642 | 37,141 | 119.62 |
| *Cx686MX PR233 | 188 MHz | 75 MHz | 2.5x | 2-3 | 2-3 | 1-2 | 1-2 and 3-4 | 171,079 | 41,352 | 134.47 |
| Cx686MX PR266 | 208 MHz | 83 MHz | 2.5x | 1-2 | 2-3 | 2-3 | 1-2 and 3-4 | 190,378 | 45,837 | 149.70 |
| (no POST) | - | - | - | 2-3 | 1-2 | 2-3 | 1-2 and 3-4 | - | - | - |
| (no POST) | - | - | - | 2-3 | 2-3 | 2-3 | 1-2 and 3-4 | - | - | - |
| Cx686MX PR166 | 150 MHz | 50 MHz | 3x | 1-2 | 1-2 | 1-2 | 3-4 | 137,923 | 33,582 | 108.04 |
| Cx686MX PR233 | 200 MHz | 66 MHz | 3x | 1-2 | 1-2 | 2-3 | 3-4 | 182,634 | 43,985 | 143.58 |
| Cx686MX PR300 | 225 MHz | 75 MHz | 3x | 2-3 | 2-3 | 1-2 | 3-4 | 204,996 | 48,927 | 161.33 |
| Cx686MX PR266 | 210 MHz | 60 MHz | 3.5x | 2-3 | 1-2 | 1-2 | none | 191,750 | 46,923 | 151.04 |
| Cx686MX PR233 | 175 MHz | 50 MHz | 3.5x | 1-2 | 1-2 | 1-2 | none | 160,444 | 38,962 | 126.11 |
| Cx686MX PR300 | 233 MHz | 66 MHz | 3.5x | 1-2 | 1-2 | 2-3 | none | (boot fail) | (boot fail) | (boot fail) |
| Cx686MX PR233 | 192 MHz | 55 MHz | 3.5x | 1-2 | 2-3 | 1-2 | none | 176,479 | 42,471 | 138.45 |
| (no POST) | - | - | - | 1-2 | 2-3 | 2-3 | none | - | - | - |
| (no POST) | - | - | - | 2-3 | 1-2 | 2-3 | none | - | - | - |
I stopped testing here, as it's evident my IBM 6x86MX PR233 won't cope beyond 233 MHz - trying 75 MHz and 83 MHz FSBs would potentially damage it using the 3.5x multiplier. It coped well with a 20% overclock to 225 MHz though, which is equivalent to the Cx686MX PR300! Perhaps there is some truth that IBM-branded chips were deliberately given more 'headroom' than specced compared to Cyrix' own brand.
So we've learned a few things here:
1) This motherboard does allow configurations that use the non-standard 83 MHz FSB. How stable it is is anyone's guess, but in theory it means you could run an appropriate 6x86MX at 290 MHz.
2) There is BIOS support well beyond the 233 MHz in the user manual (at least in BIOS version O2-006).
3) JP2 , JP3 and JP4 set the FSB clock.
4)
JP7 sets the clock multiplier (1-2 only = 2x, 1-2 and 3-4 = 2.5x, 3-4 only= 3x, open = 3.5x).
Graphics Weirdness
Recall in part 1 where I was getting a load of graphical issues, even on startup ('unsupported mode', etc)? Well, it turns out text modes seem to work ok with the UMC card after the PC has got to the boot stage, which uses the UM85C408 graphics chip with 512 KB of video memory. The RAMDAC is an ICS5341 GENDAC (24-bit colour depth). The Diamond Stealth VRAM was similar, but produced white vertical bands on initial startup. Both these cards failed to ever work in a graphics mode, even with UNIVBE running.
It then dawned on me that it could be the ISA and PCI buses being driven too fast. My CPU jumper settings were configured for a 75 MHz FSB. I have no details on how the clock dividers work on this motherboard (and the BIOS has no config settings for bus speeds at all), but I assume it's probably CLK/5 or CLK/6 for the ISA bus and CLK/2 for the PCI bus. This would mean the ISA bus is being driven at 12.5-15 MHz and the PCI bus at 37.5 MHz (both can be considered overclocks). A lot of cards really don't like working out of spec, but my Tseng Labs ET4000 that worked in Part 1 is known to me as a very reliable card when run on a fast overclocked ISA bus.
I dropped the FSB down to 66 MHz and put in an ET4000/32p PCI card that had failed to run some graphical benchmarks at 75 MHz FSB, just hanging as soon as they started. Reducing the FSB allowed it to run these successfully, so I think I'm onto something with the bus speed.
I will check the ISA and PCI bus clock lines with an oscilloscope and report back soon...
Summary
All in all, there's not too much to get excited about with this motherboard. When we have much better options such as the Gigabyte GA-5AX which also caters to Cyrix/IBM and AMD processors alongside Intel ones but adds solid bus configuration options and many further tweaks. Furthermore, the Intel 430TX's cache limit to the first 64 MB of memory could be very restrictive.
If you have any information that would help me diagnose the graphics card issues on this motherboard, get in touch!