You may have read too much into my comment. Your results are certainly of interest for benchmarking and charting purposes, which for some people, is all they care about. Other members are interested in building both a fast and solid system. I am interested in both, but ultimately, I like my systems to be stable and give most credit to benchmarks for stable systems. The AMD X5-200 charted in the Ultimate 486 Benchmark Comparison was surprisingly stable.

I only have an AMD X5-200 benched at 50x4. Your 3x66 and 3x60 systems have surprisingly fast memory throughput. For charting purposes, I would need to have the results from all the programs listed on the Ultimate 486 Benchmark Comparison to add your values to the chart. It sounds like you may be able to do this at 3x60 and I would be interested to see the results. It also sounds like you may be able to get a system stable at 60x3 and may be something you want to pursue. Could you also show the Cachechk -v7 results? cachechk.exe -d -t4 would be sufficient.

Your Speedsys averaged memory throughput is as fast as my L2 cache throughput at 2x66, making your system not really need L2 cache. This is why it would be of interest to demonstrate your system stable at 3x60.

Anonymous Coward wrote:

I never understood why no company ever released a 486 to Pentium upgrade adapter (the shitty one from Intel doesn't count). We had these things for 8088s, 286s, and 386s called "dinguses". Was there something so different about Pentium that made a socket adapter too difficult to produce?
Ideally what I'd want though is a newly fabricated 486 CPU that had really high multipliers but bus speeds of 33 and 40MHz.

I have seen 2rd party 486s with added float point operations.

Could you provide any details on these "2nd party" 486s?

haha 🤣 "2rd" woops. Unfortunately I'm just a crazy guy talking out of his arse as i do not have a source I just remember stumbling on to a web page with list of the 3rd party 486s made through the years, and then seeing how expensive they were on eBay.

I just had a thought, since it's looking like running beyond 200MHz is slim... has anyone had any success running at a 66MHz bus w/ 1/2 PCI divider? Even 66MHz x 2? Which RAM and cache are you using to achieve a stable 66MHz bus?

AMD x5-133's only have 3x and 4x operation, so personally I cannot test a 66MHz x 2, but for these chips what about 66MHz x 3?

A 200MHz AMD x5, 66MHz bus, 33MHz PCI clock sounds mighty promising. Finding the right combination of RAM and cache seems to be the tough part.

It is too difficult to find an AMD X5 that is stable at 200 MHz. I've stopped looking for these magic chips.

The only 486 chip I've seen relatively stable at 66 Mhz x 2 was an IBM 5x86c. Maybe there is some pin modification you can do to get the X2 into 2x mode? You can also try using an AM486DX4-120 to run at 2x66. You loose out on 8 KB of L1 cache though. The Intel DX4-100 will also run at 2x66, but it is not very stable.

pushing over 200mhz seems crazy for me, as 486 components are getting rare today, its better to run them within a safe limit rather than risk blowing up.

numeriK wrote:

While I can respect where you're coming from, I disagree. A benchmark is exactly that. Many of the best 3DMark systems in the […]
Show full quote

For my interests, the benchmark results are not of interest until the system is demonstrated stable.

While I can respect where you're coming from, I disagree. A benchmark is exactly that. Many of the best 3DMark systems in the world are only stable long enough to run 3DMark, usually due to thermal issues/lack of enough liquid nitrogen. So, according to you my results are not valid?

What I'm testing here is the limits of a 486. Benchmark results prove these limits can be achieved, for a short period of time or not. I'm not looking for a 200Mhz+ 486 that's 100% 24/7 stable - to be quite honest that's most likely impossible - but who truely knows? Thus the point of this thread.

I'm playing with 60-66Mhz bus speeds and ~205Mhz CPU speeds, and I'm getting complete runs through various benchmarks, yet they're irrelevant?

Like I said, I can appreciate where you're coming from but that simply isn't the point of this thread. If I want a 100% stable system I'll go back to factory defaults... 😵

The point that is being made that the system might run through your benchmarks OK, but may fail as soon as you try booting a game or other app. It may also fail to boot Windows or do undesirable things when writing information to memory or to your hard drive leading to data corruption. A system that is only good for running benchmarks isn't worth the effort that went into building it because you'll never be able to use it for anything else.

I just came across with this. Sorry for necroposting, but i think i have answer for the question.

https://www.youtube.com/watch?v=04EavG0bLX0

Am5x86-133(ADW) OC to 211MHz (52.9x4) 15ns cache, 60ns memory - slow settings. Some benchmarks done, then unfortunately DOOM failed, maybe it can run without problems with better cooling and more power to CPU.

Is this the record for a VOGONS post with the longest duration necro?

In reply to your thread, if that user at 211 MHz failed DOOM, it certainly will not hold up in Windows 9x. I would write it off.

Anonymous Coward wrote on 2012-10-23, 01:06:

I never understood why no company ever released a 486 to Pentium upgrade adapter (the shitty one from Intel doesn't count). We had these things for 8088s, 286s, and 386s called "dinguses". Was there something so different about Pentium that made a socket adapter too difficult to produce?
Ideally what I'd want though is a newly fabricated 486 CPU that had really high multipliers but bus speeds of 33 and 40MHz.

From what I was told 1990+ all the dinguses were solely developed because of the small but dedicated Japanese x86 market and the fact some sold here was an afterthought.

As we entered the Pentium era the amount of proprietary Japanese 486dx class x86 systems was small and the Bulwarks of Transputer style adapter sockets and Buffalo memco cpu upgrades seemed to give up.

The large installed base of proprietary 286/386 Japanese systems didn’t seem to attract development of a pentium solution for those either. (Maybe buffalo figured 486 and 486 like upgrades was good enough)

In so far as technical reasons I never could find any, a pentium had a 32bit mode, hybrid motherboards took advantage of it offering both 486 and pentium compatibility, one would think it’s not a bridge that far, especially for socket 3 with 3.3 volt support.

My guess is without Japanese demand the market dried up for socket adapters to upgrade to next gen chips.

iPonRMA wrote on 2026-01-30, 22:05:

I just came across with this. Sorry for necroposting, but i think i have answer for the question.

https://www.youtube.com/watch?v=04EavG0bLX0

Am5x86-133(ADW) OC to 211MHz (52.9x4) 15ns cache, 60ns memory - slow settings. Some benchmarks done, then unfortunately DOOM failed, maybe it can run without problems with better cooling and more power to CPU.

There is a rather colorful account of drunken overclocking while immersing a 486 system in liquid nitrogen to achieve well past 200mhz but it may be apocryphal.

https://forums.anandtech.com/threads/whoa-ove … -247mhz.519421/

feipoa wrote on 2026-01-30, 23:25:

Is this the record for a VOGONS post with the longest duration necro?

In reply to your thread, if that user at 211 MHz failed DOOM, it certainly will not hold up in Windows 9x. I would write it off.

Maybe. I used 4.3V, so i think i still have possibilities, the cooler also far from optimal. I think i can check the system with liquid nitrogen on the end of May.

14 years ago is like last week for me. At least somebody finally answered the damn question.

Since it has been many years, i would like to know if anyone got it stable within 4volts. Of course ram and cache should be relaxed, i think 3-2-2-2 cache and 5-3-3-3 dram should work.

3x66, 200mhz with full gear and tightest wait states - yes.
240mhz - never stable so far.

pshipkov wrote on 2026-02-07, 05:41:

3x66, 200mhz with full gear and tightest wait states - yes.
240mhz - never stable so far.

with a peltier cooler.

Yes.
Many Am5x86-133 are 200mhz capable, but very very few can be long term stable on air cooling.
So far only Chatdi99 and gonzo (if not mistaken) demonstrated that.
It is easier to strap a weak peltier on a promising cpu to get it 200 long term stable than searching for the super rare unicorn (expensive given the number of chips one has to go through).

rmay635703 wrote on 2026-01-31, 00:45:

From what I was told 1990+ all the dinguses were solely developed because of the small but dedicated Japanese x86 market and the […]
Show full quote

Anonymous Coward wrote on 2012-10-23, 01:06:

I never understood why no company ever released a 486 to Pentium upgrade adapter (the shitty one from Intel doesn't count). We had these things for 8088s, 286s, and 386s called "dinguses". Was there something so different about Pentium that made a socket adapter too difficult to produce?
Ideally what I'd want though is a newly fabricated 486 CPU that had really high multipliers but bus speeds of 33 and 40MHz.

From what I was told 1990+ all the dinguses were solely developed because of the small but dedicated Japanese x86 market and the fact some sold here was an afterthought.

As we entered the Pentium era the amount of proprietary Japanese 486dx class x86 systems was small and the Bulwarks of Transputer style adapter sockets and Buffalo memco cpu upgrades seemed to give up.

The large installed base of proprietary 286/386 Japanese systems didn’t seem to attract development of a pentium solution for those either. (Maybe buffalo figured 486 and 486 like upgrades was good enough)

In so far as technical reasons I never could find any, a pentium had a 32bit mode, hybrid motherboards took advantage of it offering both 486 and pentium compatibility, one would think it’s not a bridge that far, especially for socket 3 with 3.3 volt support.

My guess is without Japanese demand the market dried up for socket adapters to upgrade to next gen chips.

So far as I know, the differences between a 486 and Pentium chip are too great for a manufacturer to simply create an upgrade option similar to a 386-to-486 upgrade. The Pentium used a different electrical control signal type than the 486 and a 64-bit path to memory. An upgrade manufacturer would have to essentially reverse engineer the Pentium, risking litigation from Intel's private IP, or license the technology from Intel. The end result would be a very expensive CPU that is slowed down by the rest of the components on the motherboard. The upgrade chip would have to reduce the Pentium's 64-bit data bus down to the 486's 32-bit bus design, add control signal logic, possibly increase the L1 cache to hide performance losses, and the manufacturer would have to work with BIOS writers to ensure the chip can be recognized and initialized properly.

All of that is theoretically possible, but it would be very expensive to produce. Someone else mentioned the law of diminishing returns. It was more cost effective to just move to a Pentium-class system if you needed Pentium performance.

I'm also not quite sure why Intel's offering was referred to as "shitty" either. For what it was, I think it is an excellent product. The Am5x86 chip was far more cost effective, but the 486-to-Pentium OverDrive was a good option if you needed Pentium compatibility but weren't ready to buy a new computer yet.

I have both the Am5x86 133MHz CPU and a 486-to-Pentium OverDrive CPU (in different systems), and I like both for different reasons.

OzzFan wrote on 2026-02-10, 17:33:

I'm also not quite sure why Intel's offering was referred to as "shitty" either. For what it was, I think it is an excellent product.

Probably because very few socket 3 motherboards worked with the POD when L1 was in write-back mode. Has anyone made a list? I'd be surprised if it topped more than a dozen motherboards. I can only think of 3.

OzzFan wrote on 2026-02-10, 17:33:

So far as I know, the differences between a 486 and Pentium chip are too great for a manufacturer to simply create an upgrade op […]
Show full quote

rmay635703 wrote on 2026-01-31, 00:45:

From what I was told 1990+ all the dinguses were solely developed because of the small but dedicated Japanese x86 market and the […]
Show full quote

Anonymous Coward wrote on 2012-10-23, 01:06:

I never understood why no company ever released a 486 to Pentium upgrade adapter (the shitty one from Intel doesn't count). We had these things for 8088s, 286s, and 386s called "dinguses". Was there something so different about Pentium that made a socket adapter too difficult to produce?
Ideally what I'd want though is a newly fabricated 486 CPU that had really high multipliers but bus speeds of 33 and 40MHz.

From what I was told 1990+ all the dinguses were solely developed because of the small but dedicated Japanese x86 market and the fact some sold here was an afterthought.

In so far as technical reasons I never could find any, a pentium had a 32bit mode, hybrid motherboards took advantage of it offering both 486 and pentium compatibility, one would think it’s not a bridge that far, especially for socket 3 with 3.3 volt support

My guess is without Japanese demand the market dried up for socket adapters to upgrade to next gen chips.

So far as I know, the differences between a 486 and Pentium chip are too great for a manufacturer to simply create an upgrade option similar to a 386-to-486 upgrade. The Pentium used a different electrical control signal type than the 486 and a 64-bit path to memory. An upgrade manufacturer would have to essentially reverse engineer the Pentium, risking litigation from Intel's private IP, or license the technology from Intel. The end result would be a very expensive CPU that is slowed down by the rest of the components on the motherboard. The upgrade chip would have to reduce the Pentium's 64-bit data bus down to the 486's 32-bit bus design, add control signal logic, possibly increase the L1 cache to hide performance losses, and the manufacturer would have to work with BIOS writers to ensure the chip can be recognized and initialized properly.

All of that is theoretically possible, but it would be very expensive to produce. Someone else mentioned the law of diminishing returns. It was more cost effective to just move to a Pentium-class system if you needed Pentium performance.

I'm also not quite sure why Intel's offering was referred to as "shitty" either. For what it was, I think it is an excellent product. The Am5x86 chip was far more cost effective, but the 486-to-Pentium OverDrive was a good option if you needed Pentium compatibility but weren't ready to buy a new computer yet.

I have both the Am5x86 133MHz CPU and a 486-to-Pentium OverDrive CPU (in different systems), and I like both for different reasons.

The proprietary bus is why nobody cloned into ppro, the “pentium” bus was nothing special besides being pipelined more deeply with a longer burst cycle. The next address and burst was what was different, whether those could be handled by a simple hold and release is unknown to me.

Certain old 3rd party pentium motherboards used chipsets that stuck to 32bit memory transfers, OPTi 82C571 , so besides speed the chip was already being interfaced to 32 bit boards at retail