In 486 systems the motherboards don't have multiplier settings. The multipier is built into the CPU and fixed

stamasd wrote:

In 486 systems the motherboards don't have multiplier settings. The multipier is built into the CPU and fixed

Ouch I didn't know that! Too bad!!!

I thought we could mess around with jumpers and stuff...

EDIT:
Riiiiiight! Now I remember a diagram of some CPUs I saw loooong time ago, in which the clock multiplier was inside the CPU!

never heard of pre-Pentium boards having such a multiplier (non-integer).

I think overclocking is misunderstood.
When Intel or others manufacture CPUs, they set themselves a target clock speed (for a given supply voltage and desired TDP), and when a CPU does not pass their stringent tests, they give it a lower rated clock speed assuming it is not completely defective.
If you get a CPU that is stable at the rated speed, this does not mean it can't be clocked higher without damage. It just means that most likely it was designed around a target that it safely exceeds.

eg: I have plenty of CPUs that can happily run 20% even 40% higher than their rated speed, and this without raising the core voltage.

Don't be fearful of overclocking.

stamasd wrote:

In 486 systems the motherboards don't have multiplier settings. The multipier is built into the CPU and fixed

This is not always the case.
My BIOSTAR and PCChips M919 had 2x/3x multiplier jumpers.

They are not always advertised as such, and often are just part of the baker's dozen of jumpers that need to be set for each CPU type.

tpowell.ca wrote:

never heard of pre-Pentium boards having such a multiplier (non-integer). […]
Show full quote

never heard of pre-Pentium boards having such a multiplier (non-integer).

I think overclocking is misunderstood.
When Intel or others manufacture CPUs, they set themselves a target clock speed (for a given supply voltage and desired TDP), and when a CPU does not pass their stringent tests, they give it a lower rated clock speed assuming it is not completely defective.
If you get a CPU that is stable at the rated speed, this does not mean it can't be clocked higher without damage. It just means that most likely it was designed around a target that it safely exceeds.

eg: I have plenty of CPUs that can happily run 20% even 40% higher than their rated speed, and this without raising the core voltage.

Don't be fearful of overclocking.

This is interesting to know, thanks so much.
Mine is not fear (if done within certain boundaries). I don't know what it is! 🤣 Maybe I don't like that the CPU has a number printed on it and it runs at a different frequency 🤣 I guess it's perfectionism 😁

Those multipliers are still built into the CPU, DX4 and onwards. The same jumper switches one type of CPU between 2x and 3x and another between 3x and 4x.

firage wrote:

Those multipliers are still built into the CPU, DX4 and onwards. The same jumper switches one type of CPU between 2x and 3x and another between 3x and 4x.

Thanks!

On the PCChips M919 v3.4

CPU SPEED CONFIGURATION (DX4/5X86/X5 ONLY)

1 Speed       JP6
2~~~~~~~~~~~~~~~~~
3  2x       Closed
4  3x        Open 
5  4x       Closed

I checked and it worked on my AMD DX5-133.

On the BIOSTAR MB-8433UUD v2.x

1 Speed      JP45
2~~~~~~~~~~~~~~~~~~~~~~~~~
3  1x       1-2 or 2-3
4  2x       1-2, 4-5
5  3x       1-2 
6  4x       1-2, 4-5

What firage said. Closing JP6 can be either 2x or 4x, depending on the CPU, since the multipliers are built into the CPU.

derSammler wrote:

What firage said. Closing JP6 can be either 2x or 4x, depending on the CPU, since the multipliers are built into the CPU.

You mean, 2x/4x OR 3x depending on jumper setting..
I recall getting the DX5-133 to behave like a DX4-100 with the jumper removed, and this while keeping the 33MHz bus speed.
I regret not trying to see if I could go to a clock tripled 50 MHz bus. I wonder how a DX4-150 would run.

That's because the X5 supports 3x and 4x, just like the enhanced 486 by AMD (undocumented).

tpowell.ca wrote:

I regret not trying to see if I could go to a clock tripled 50 MHz bus. I wonder how a DX4-150 would run.

sorry bro! 😎

derSammler wrote:

That's because the X5 supports 3x and 4x, just like the enhanced 486 by AMD (undocumented).

well, well, well. I learned something new, again.

Thanks derSammler

The Intel DX4 was supposed to have support for a 2.5X multiplier, because you can read about it in the DX4 datasheets. Even some motherboards and VRM interposers show the jumper settings for 2.5X, but somehow at the last minute this feature was removed.
I believe the theory was that the 2.5X setting was related to the cancelled DX3 CPU, which would have run at 83MHz like the POD. I guess the DX3 likely would have used the same dies as the DX4.

In addition to what tpowell said, I think CPU manufacturers sometimes deliberately did not want to rate CPU speeds higher or to qualify them at higher speeds so as not to compete with their higher-end CPU lines. An example of this might be the AMD Am5x86-133. I think all, or nearly all, will run stable at 160 MHz and stock voltage. Imagine if they marketed the Am5x86-160, sales of K5, and later K6 CPUs would fall and those CPUs have a higher price tag. Same for the Pentium Overdrive POD83. Some run at 100 MHz at stock voltage. All the ones I've modded to run at 4 V run at 100 MHz. Similarly, the K5-PR200 wasn't really marketed. Cyrix conveniently didn't sell their 5x86-133/4x as a stand-alone CPU. Who would want to buy a higher priced Cyrix 6x86 at 80 or 100 MHz when you can have a Cyrix 5x86 at 133 MHz with marginally less performance. A lot of consumers at the time also just looked at the frequency - 100 MHz vs. 133 MHz. 133 MHz is cheaper and has a higher frequency, buy that one! Because of this marketing ploy, I don't always view some CPU's as being "overclocked", but rather, being run optimally. This does not include cases whereby you need to get fancy with cooling and what not to overclock. The 1.6 V, K6-III+ 400ATZ is another good example of a CPU which nearly always clocks to at least 500 MHz without issue. I don't think I've seen a case whereby it didn't.

Interesting points, thanks!

There's room for countless possibilities and combinations 🤣!

Similarly the first CPU in the Mendocino series, the Celeron 300A. That one was marketed way below the speed that the technology allowed it to reach - as the same process resulted in 500+MHz not too much later. The 300A could effortlessly made to run at 450MHz and above even without any extra voltage (I have one that does 504MHz at stock voltage). Intel didn't want it to cut into the sales of the older Deschutes PII-450 thus only sold it at the beginning as a 300MHz part. Later they released the exact some core at higher "official" speeds.

well, for sure all this would have made my days if I knew it when I was a kid. At that time, the use was for real use (of course mainly gaming, although not only), not like now for the sake of old times. So, you had games running slow and no money to buy better hardware... knowing these things would have been great, not talking about the thrill of boosting performance like that as a kid!