To what degree could a Pentium MMX 200 or 233 be underclocked via multiplier modifications and changes in bus speed? I know the lower speed chips were had higher multipliers locked out, but surely the lower multipliers weren't locked out of the faster chips.
Could a 233 be run from 120 to 233?
Sure. They have 2x multipliers. Pentium "Classic" has a 1.5x multi for the 75MHz version.
Sorry for reviving.
But it is exactly the topic I want to ask about and it allows me to correct the above.
Have a 'plastic' SL27J pentium 200 MMX here, stepping xB1, 2,8V. On an Asus P5A board it does either 2.5x or 3.0x, nothing else. Wonder what multiplier options there are for similar Pentium CPUs. The 133MHz Challenge topic has one running on 133MHz (sspec?). But unfortunately SL27J cannot do 133/66 MHz, neither can it do 233/66 MHz.
There is this article but it only covers the overclocking part.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
The 200 MMX on my AN430TX can underclock to 90mhz [x1.5 @ 60]. But after moving up to a 233 MMX the lowest now *seems* to be 120 [x2 @ 60]. The 233 interprets the published x1.5 jumper settings as x3.5 so the bottom end is lost. But there may be an undocumented jumper setting for x1.5, I just haven't found it yet.
Well at least on my Super Socket 7 boards I can configure the 233 MMX with the following ratios:
2x, 2.5x, 3x, 3.5x
The standard pentium 200 is similar, however interprets the 3.5 ratio as 1.5x. So this allows it clocking lower than the MMX and on a 66MHz FSB board gives you a clock speed of 100 MHz.
Thanks,
Do you guys know the sspec number of these pentiums (like SL27J)?
I think it is the CPU that is the limitation, not the mainboard. With Cyrix and AMD CPU's I have at least double the amount of available multipliers on the Asus P5A. It is just this pentium MMX that is rather boring.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
P200 MMX = SL26J and does X1.5, x2, x2.5, x3 [AN430TX board]
P233 MMX is currently installed under a heatsink and I can't check just now.
For the 686 benchmarks, I ran a P233-MMX SL27S at 133, 166, 200, 233, 250, 262, and 300 MHz. As others have mentioned, 2.0x, 2.5x, 3.0x, and 3.5x.
Plan your life wisely, you'll be dead before you know it.
wrote:For the 686 benchmarks, I ran a P233-MMX SL27S at 133, 166, 200, 233, 250, 262, and 300 MHz. As others have mentioned, 2.0x, 2.5x, 3.0x, and 3.5x.
how stable is your mmx233 at 300? and how much voltage do you need?
wrote:wrote:For the 686 benchmarks, I ran a P233-MMX SL27S at 133, 166, 200, 233, 250, 262, and 300 MHz. As others have mentioned, 2.0x, 2.5x, 3.0x, and 3.5x.
how stable is your mmx233 at 300? and how much voltage do you need?
I have a 166 MMX and in my case I can run it without any problems at 83x3.5 with just 3.1v or 3.2v
wrote:how stable is your mmx233 at 300? and how much voltage do you need?
To get the P233MMX-300 to run all the 686 benchmarks, it took 3.4 V. I beleive some DOS benchmarks finsihed at less voltage (3.0-3.3V). Quake 1 was the most sensitive.
Plan your life wisely, you'll be dead before you know it.
About this SL27J, it was tested again, and really just two multipliers are selectable. Now I think I figured out what is up with this: to block the multipliers on the pentium that are over the specified speed, intel disabled one of the BF signals (BF1), or jumpers if you like. This leaves only one BF signal (BF0) / jumper to select the multiplier, hence just 2 speeds instead of 4. 🙁
For completeness: on super socket 7 Cyrix and AMD CPUs also read a BF2 signal allowing for maximum 8 choices.
On super socket 7, did anyone run a Pentium MMX on 200MHz at 2x 100MHz bus for prolonged time?
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
wrote:On super socket 7, did anyone run a Pentium MMX on 200MHz at 2x 100MHz bus for prolonged time?
I ran about 3 continuous hours of benchmarks at 2.5 x 100 MHz and it did not show any signs of distress.
Plan your life wisely, you'll be dead before you know it.
The early production samples you should get a minimum of 1.5 but they all top out at 3.5 except for the very last gen mobiles however that is another topic. After that point it all eventually comes down to fsb clocks, 2x 66mhz end of story for the usual 133 ect but the fsb wall is around 100mhz on all the samples that I had tested.
On a far away planet reading your posts in the year 10,191.
wrote:P200 MMX = SL26J and does X1.5, x2, x2.5, x3 (AN430TX board)
I obtained two SL26J, in plastic packaging:
-one is dead on arrival.
-one does 2.0x, 2.5x, 2.5x, 3.0x. Speedsys scores for these settings are 100.73, 125.92, 125.92, 151.10. Meaning 2 out of 4 settings have the same result, there is no 1.5x setting on this one 🙁. Tested with an Asus P5A mainboard.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
I seem to recall the Pentium MMX chips were intel's first attempt at locking down the CPU to prevent overclocking.
"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
Don't forget the motherboard underclocking possibility! FSB was more important than the multiplier back in the day, and most socket 7 boards can work at 50, 60, 66, 75, 83 Mhz FSB. So you can underclock a Pentium MMX to 50x2=100Mhz, which is much slower than 66x1.5=100Mhz!
I ordered a P54cs PPGA (oem only) P1 classic that is packaged in the same way as the MMX model so will test it out when it comes in to see what range that it has. The PPGA model has the same core as the ceramics but hits higher clocks with lower temps. Min voltage for it is 3.15v.
On a far away planet reading your posts in the year 10,191.
wrote:Don't forget the motherboard underclocking possibility!
That is not forgotten. For that same reason I have four i440BX motherboards with 50..133MHz FSB support. Yet the Asus P5A, being super socket 7, starts out at 60MHz FSB.
The AMD K6-2+ here will probably be a more desirable processor for SS7, if not the Cyrix MIIv. But it is nice to add the final model Socket 7 Intel CPU with a good range of multipliers.
wrote:I ordered a P54cs PPGA (oem only) P1 classic that is packaged in the same way as the MMX model
Interesting, looking forward to your test result.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
Had the opportunity to test three other Pentium MMX processors. the total is now six:
Pentium MMX 233 SL27S 5.4.3 - 2.0x, 2.5x, 3.0x, 3.5x
Pentium MMX 200 SL26J 5.4.4 - 2.0x, 2.5x, 3.0x
Pentium MMX 200 SL27J 5.4.3 - 2.5x, 3.0x
Pentium MMX 166 SL27H 5.4.3 - 2.0x, 2.5x
Pentium MMX 166 SY059 5.4.4 - 2.5x, 3.0x
Pentium MMX 166 SL239 5.4.4 - 2.0x, 2.5x, 3.0x (ceramic one)
I cease to believe 1.5x exists on a retail MMX. Unless someone has better prove of it, like a Speedsys screenshot.
Strangely 5.4.3 indicates the later xB1 stepping, whereas 5.4.4 indicates xA3?
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
