Reply 20 of 20, by kool kitty89
CzasowAnton wrote on 2026-06-30, 23:40:Hello. […]
Hello.
Did someone have experience to use this board with 25 MHz CPU (Harris 286)?
My mobo starts only with 48 MHz oscillator as maximum (CPU freq = 24MHz), trying to start with faster osc leads to board stuck at post code 08.
Looking at theretroweb, i found a board with the same chipset which supports 25 MHz, it has additional 32MHz oscillator: https://theretroweb.com/motherboards/s/biosta … -1216ve-ver.-01
What is this this oscillator for? It has somethig to do with support highest frequencies? Or how do they realize this?
Sorry for my poor English, i'm foreginer...
Many thanks for help.
Not all examples of a given chipset will run at the same maximum speeds, and it's not up to chispet design and support, but just up to whether the specific chips used on any given board will actually tolerate running at a given speed (many boards officially printed or specified for a lower speed will overclock quite a lot, but others may not). On top of that, the speed of the BIOS rom chip will also limit things at POST, more so depending whether a given chipset implements BIOS wait states by default (or what wait states are used).
The last issue can be worked around by burning the BIOS to a much faster ROM chip, but the other issue of the speed the actual logic chips (the chipset) will run at is limited by a case by case basis.
Additionally, if you're trying to overclock a slower rated 286 (ie many 20 MHz 286s will run at 25 MHz), some boards or chipsets are more temperamental than others about tolerances there. I have a PCChips M205 with the same Citygate D-60 chipset as a HEDAKA HED-988 (also used on PCchips/ILION M209), but the HED-988 won't POST or won't boot to DOS when running various 286-20 CPUs I have faster than 24 MHz when the same chips on the M205 will do 25 and usually 27 MHz (though real 286-25s tend to be more stable in that case), and swapping the BIOS chips between them didn't help much. However, with the real (not remarked and not overclocked) 286-25s I've tried, the HED-988 runs perfectly fine at 25 MHz, though I never got around to trying to get beyond that. (as it was, I was using fake/remarked 60 ns DRAM to get 27 MHz in the M205, so it was hard to be sure what was RAM limited and what wasn't; I mostly stuck to using SIMMs via slots added into the SIPP sockets on the HED-988 and mostly used 70 ns as I didn't have faster ones to try and also didn't experiment too much with the DIP sockets on that board)
But that's just one example, and there's lots of factors involved in getting a board to run at specific speeds with specific CPU and RAM (RAM wait state settings in the BIOS, if present, is another one, so would be disabling parity check, especially if the parity DRAM chips are a different make or model from the main DRAM chips as the cycling tolerances can be different, both in official spec and just how prone they are to tolerating out of spec parameters: like RAS time relative to precharge time, some using much shorter precharge than others, some just using equal precharge to RAS pulse time, or varying those ratios a lot more at different wait-state settings, if they have wait-state settings at all).
More generally on that last issue (but not relevant to your specific problem),
Not all zero wait state settings are the same either, and wait state test programs also don't always give the full picture (you can have memory timing that's zero wait-state for access time, but not fast enough to actually cycle in 2 CPU clock cycles, so a handful of very fast instructions will force wait states to be added, but for anything using 3 clock cycles or more to execute, it would be zero wait state; that and chipsets that refresh RAM more often and less smartly will also add wait states more often). Some BIOS settings that say 1 wait state will show up as 0ws in tests, too, probably because it's the same sort of o-wait access but 3-clock cycle time arrangement. (which is a faster arrangement than other 1 wait state examples that add a wait to the access time, while the 0-wait access + forced minimum cycle time between accesses is somewhere in between 1 wait and 0 wait; that D-60 chipset seems to work that way and has no BIOS settings for wait states, just fixed DRAM timing)
Also, on the topic of the PCChips M218, I wonder if the chipset used on the M219 is a clone of the VLSI SCAMP as well, or if bechmarks show similar performance results. If driver compatibility is just as poor, I doubt using SCAMP drivers would have better results there (or help prove whether it's a clone chip or not). Plus some chipsets require EMS configuration within the BIOS to be done for the drivers to work (and if BIOS support is missing, they won't work, or you need a special low-level utility or program that can tweak the chipset registers in a manner similar to the BIOS ... which I think you might be able to do to get hardware EMS working on some 386SX boards, which is nice given how performance sapping EMM386 is vs real mode on a 386SX)