I don't think there are any Pentium II chipsets that allow for lower than 3x which was used for 300MHz Klamath.

appiah4 wrote:

I don't think there are any Pentium II chipsets that allow for lower than 3x which was used for 300MHz Klamath.

Both i440LX and i440BX and probably every other Pentium II and Pentium III chipset can do both 2x and 2.5x multiplier (well if the CPUs can, the chipset dosn't really care). Not every motherboard have jumpers/BIOS options for these settings though and many of those with a 2x setting use it as a fail safe setting disabling the L2 cache. Some motherboards have double 2x jumper settings, one with L2 enabled and one wth L2 disabled.

If 50 MHz FSB is also a must then I would look for a i440LX motherboard, many of those support 50 MHz FSB (there are also some i440BX boards that do). There are threads here on the forum about downclocking Pentium II CPUs. Jumper settings for some boards that can do both 50 MHz FSB and 2x multiplier with and without L2 cache are discussed and documented.

Thank you guys!!!
I will have a search for downclocking Pentium II. In fact the minimum multi setting in my mobo is 3x that gives me 200Mhz... I will search for an LX chipset in flea market.
In sysinfo 4.75 there is no performance difference having L2 cache on or off.. But with both L1and L2 off I think that I have a XT class machine...

Skyscraper wrote:

Both i440LX and i440BX and probably every other Pentium II and Pentium III chipset can do both 2x and 2.5x multiplier (well if the CPUs can, the chipset dosn't really care). Not every motherboard have jumpers/BIOS options for these settings though and many of those with a 2x setting use it as a fail safe setting disabling the L2 cache. Some motherboards have double 2x jumper settings, one with L2 enabled and one wth L2 disabled.

This is generally a very good answer. But I found that the L2 cache behaviour at 2.0x and 2.5x is not related to the motherboard or BIOS, at least not in that way. It is merely that Deschutes core only has the 2.0x/2.5x settings with L2 cache disabled, whereas Klamath core has two different 2.0x settings: one with L2 and one without. On the Klamath 2.5x is always with L2 enabled.
One needs a JUMPERED motherboard to properly access ALL the possible multiplier settings of the CPU.
More info: 50 to 133MHz FSB on a BX Mainboard - Klamath Test

gerwin wrote:

This is generally a very good answer. But I found that the L2 cache behaviour at 2.0x and 2.5x is not related to the motherboard […]
Show full quote

Skyscraper wrote:

Both i440LX and i440BX and probably every other Pentium II and Pentium III chipset can do both 2x and 2.5x multiplier (well if the CPUs can, the chipset dosn't really care). Not every motherboard have jumpers/BIOS options for these settings though and many of those with a 2x setting use it as a fail safe setting disabling the L2 cache. Some motherboards have double 2x jumper settings, one with L2 enabled and one wth L2 disabled.

This is generally a very good answer. But I found that the L2 cache behaviour at 2.0x and 2.5x is not related to the motherboard or BIOS, at least not in that way. It is merely that Deschutes core only has the 2.0x/2.5x settings with L2 cache disabled, whereas Klamath core has two different 2.0x settings: one with L2 and one without. On the Klamath 2.5x is always with L2 enabled.
One needs a JUMPERED motherboard to properly access ALL the possible multiplier settings of the CPU.
More info: 50 to 133MHz FSB on a BX Mainboard - Klamath Test

My answer was perhaps a bit simplified but if you have a motherboard with a jumper free multiplier selection with the option to choose 2x multiplier with Klamath that one is often the 2x with L2 cache disabled and sometimes also the setting the motherboard use as a "fail safe" the next reset after a failure to POST.

I have more then one jumperless motherboard (or in one case an Asus HP OEM motherboard with both jumpers and jumperless mode crossflashed with Asus BIOS) that just refuse to run Klamath at 2x with L2 enabled as the only 2x option is the "wrong" one while 2.5x as you write always have L2 enabled. With a motherboard with all three multiplier selection jumpers present it should as you write always be possible to find the right setting by trial and error as long as there isn't some kind of stupid logic present between said jumpers and the CPUs multiplier selection pins. If the motherboard dosn't have an option for jumperless mode such logic should not be present and I guess there are also jumperless motherboards with all options available.

Right, I suppose we are talking about the same thing, just in our own words and expierence: Jumperfree and Jumpered.

There are 4 multiplier request signals BF0 tot BF3, which are actually physical fingers or pins on the CPU slot cartridge. But they are only probed as such by the CPU at reset. This makes for a max. of 16 possible multiplier request signals. The actual meaning of these signals differ from CPU to CPU. Like the Tualatin core has entirely different mapping compared to a Deschutes. I do not expect any jumperless motherboard out there to support all the possible exotic multipliers on both Klamath/Deschutes AND Engineering samples of the PIII. Also I would not expect a jumperless board to have two variants of 2.0x selectable.
But in case a jumperless board has the settings you need, and the board is working well in all respects, it should not matter that much. Also it would not hurt to know beforehand exactly how this matter works.

gerwin wrote:

Right, I suppose we are talking about the same thing, just in our own words and expierence: Jumperfree and Jumpered.

There are 4 multiplier request signals BF0 tot BF3, which are actually physical fingers or pins on the CPU slot cartridge. But they are only probed as such by the CPU at reset. This makes for a max. of 16 possible multiplier request signals. The actual meaning of these signals differ from CPU to CPU. Like the Tualatin core has entirely different mapping compared to a Deschutes. I do not expect any jumperless motherboard out there to support all the possible exotic multipliers on both Klamath/Deschutes AND Engineering samples of the PIII. Also I would not expect a jumperless board to have two variants of 2.0x selectable.
But in case a jumperless board has the settings you need, and the board is working well in all respects, it should not matter that much. Also it would not hurt to know beforehand exactly how this matter works.

Yea I think we are on the same page! 😀

If I remember right only 3 of the 4 "multiplier request signals" matters for the Klamath and old boards often only have 3 jumpers. I think it's the same with Deutsches only that one of the two of Klamaths 2x settings is used for the unused 5.5x setting (works with a few fully unlocked Deutsches CPUs). Newer P6 CPUs used all 4 "multiplier request pins".

Good points, I agree.

On my Chaintech 6BTA2 I can set in bios FSB (66 / 100) and multiplier (starting in 2x), with any of the unlocked multiplier PII , so I can use a PII 300 @ 133mhz up to 300Mhz. With setmul disabling L1 / L2 this makes for a very flexible system for DOS.

Does it means that setting the multiplier to 2x in bios i'm also disabling L2 cache?