AMD 5x86 or Cyrix 5x86?

I suppose you're referring to the AMD part, which has always been much more common and probably easier to overclock.
But keep in mind that not all 5x86 parts are created equal. Some parts may overclock better then others and the AMD 5x86 had a long production run. More recent ones are a bit more likely to result in a succesful overclock.
With overclocking, there's always the risk of damaging parts and there's hardly ever any guarantee.

A lot of people run Amd 5x86 at 160 MHz so you should be fine. As an extra precaution, look for a later revision (AFAIK, BGC/ADZ tolerate higher temperatures better than earlier makes) and active cooling.

Can't say anything about Cyrix though.

Tetrium wrote:

AMD 5x86 or Cyrix 5x86? […]
Show full quote

AMD 5x86 or Cyrix 5x86?

I suppose you're referring to the AMD part, which has always been much more common and probably easier to overclock.
But keep in mind that not all 5x86 parts are created equal. Some parts may overclock better then others and the AMD 5x86 had a long production run. More recent ones are a bit more likely to result in a succesful overclock.
With overclocking, there's always the risk of damaging parts and there's hardly ever any guarantee.

I first mean the AMD 5x86 chips, and I am aware of the overclock risk, I just want to know has it been done, what happened?

My SV8B DX4-100 clocked up perfectly, and these are early PC chips, i think that cooling isn't much of an issue given a good copper heatsink and fan.

Perhaps an ADW will be close enough? Those are the ones for sale on Ebay.

It should run fine at 40MHz FSB, which should then give you 160 MHz (4x 40 MHz). This is based on my own personal tests and numerous reviews and feedback I've read. Overclocking past that speed could cause instability issues.
However, I've tested this on a motherboard that officially supported this CPU.
You can also refer to this thread - The Ultimate 486 Benchmark Comparison. Interesting to note how the Intel 486DX4 120 MHz performed.

Since your motherboard's BIOS doesn't officially support the Am5x86, it will probably not recognise the CPU correctly and it's also possible that it will not utilise the L1 write back cache of the CPU (it supports both write back and the older write thru L1 cache mode, the latter being what most 486 CPU's supported).
Just remember that at 40 MHz FSB, you will probably need to apply wait states to your VLB to ensure a more stable system.
Also remember to set the CPU voltage to 3.3V.

I've also attached the motherboard's manual, in case you don't have it - you require a PCX viewer to view these files.

jesolo wrote:

It should run fine at 40MHz FSB, which should then give you 160 MHz (4x 40 MHz). This is based on my own personal tests and num […]
Show full quote

It should run fine at 40MHz FSB, which should then give you 160 MHz (4x 40 MHz). This is based on my own personal tests and numerous reviews and feedback I've read. Overclocking past that speed could cause instability issues.
However, I've tested this on a motherboard that officially supported this CPU.
You can also refer to this thread - The Ultimate 486 Benchmark Comparison. Interesting to note how the Intel 486DX4 120 MHz performed.

Since your motherboard's BIOS doesn't officially support the Am5x86, it will probably not recognise the CPU correctly and it's also possible that it will not utilise the L1 write back cache of the CPU (it supports both write back and the older write thru L1 cache mode, the latter being what most 486 CPU's supported).
Just remember that at 40 MHz FSB, you will probably need to apply wait states to your VLB to ensure a more stable system.
Also remember to set the CPU voltage to 3.3V.

I've also attached the motherboard's manual, in case you don't have it - you require a PCX viewer to view these files.

Alright, well I have the manual downloaded, printed, and used even before I had the motherboard.

And this board should support most of the features even if official support isn't there. It fully supports WB cache, and VLB is capable of handling 40Mhz FSB just fine as it is fully in spec (And I am running my system, at 40Mhz anyways)

And the voltage is already 3.3v

But I will get one and try it out.

Perfect - you might have to play around with the jumper settings, perhaps using the setting for an AMD486DX2 (if the 486DX4 setting doesn't work).

jesolo wrote:

Perfect - you might have to play around with the jumper settings, perhaps using the setting for an AMD486DX2 (if the 486DX4 setting doesn't work).

It has an internal multiplier, so I assume I would just have to set it for a DX4 with 40mhz base clock and no motherboard clock multiplier.

160 MHz can be a little bit more of a challenge than just flipping bus jumpers; it requires half decent cooling and possibly a small voltage bump. You should be able to jumper it for 3x40MHz, alternatively. The same jumper that switches a DX4 from 2x to 3x respectively switches the Am5x86 between 4x and 3x.

My acer ap43 pci board will not take it. This board is known here to run it just fine but mine has a slightly older bios and as soon as i jumper it to enable the 4x multi it acts really strange and extremely slow (cache not working).

Running it with slower timings and 3x50 is no problem and almost equal to 160mhz .40fsb gives a bit better timings ofcourse but 3x40 is still slower than 3x50.

So if it is not running properly try the dx4 setting with 3x40 or 3x50. The 16kb cache alone will make it faster than the dx4-100@120.

meljor wrote:

My acer ap43 pci board will not take it. This board is known here to run it just fine but mine has a slightly older bios and as soon as i jumper it to enable the 4x multi it acts really strange and extremely slow (cache not working).

Running it with slower timings and 3x50 is no problem and almost equal to 160mhz .40fsb gives a bit better timings ofcourse but 3x40 is still slower than 3x50.

So if it is not running properly try the dx4 setting with 3x40 or 3x50. The 16kb cache alone will make it faster than the dx4-100@120.

My board has 0 support for 50 mhz FSB for obvious reasons.

I've had okay luck with ADWs. I believe I own three, and they all overclock to 160. Funnily enough, I have one ADZ and it doesn't overclock at all. I recommend the BGC which was made as late as 2001 or 2002.

meljor wrote:

My acer ap43 pci board will not take it. This board is known here to run it just fine but mine has a slightly older bios and as soon as i jumper it to enable the 4x multi it acts really strange and extremely slow (cache not working).

Running it with slower timings and 3x50 is no problem and almost equal to 160mhz .40fsb gives a bit better timings ofcourse but 3x40 is still slower than 3x50.

So if it is not running properly try the dx4 setting with 3x40 or 3x50. The 16kb cache alone will make it faster than the dx4-100@120.

I very vaguely remember having used a DX4 in my Acer AP43 instead of a 5x86 and apparently it can be a tricky board on which to get the 5x86 running properly...or running at all even.

RacoonRider send me some undocumented jumper settings along with an explanation on how to get a 5x86 running on an AP43 a couple years ago.
I never got around to try these out for myself though, but for the AP43 this might be worth a try and I might as well spread this bit of info in case anybody might ever need it.

_________JP25____JP26
25 MHz___2-3____2-3
33 MHz___1-2____1-2
40 MHz___1-2____2-3 <unlike the manual/th99/PCB prints
50 MHz___2-3____1-2 <not stated in manual

and

The jumper settings are described on page 15-2 of the second part of the manual. However, if you just set it up, nothing will work or even POST! You have to set jumpers for DX4 SL-enhanced WB first, start the system, and only then turn it off and add JP12. I have now idea why it does not work otherwise, I triple-checked everything and still no post. You have to do this only once though.

The EXP-4045 might have undocumented jumper settings as well?

Btw, I recall some people actually managing 200MHz with the AMD 5x86 part, but those might've been some of the more recently manufactured parts.

Tetrium wrote:

I very vaguely remember having used a DX4 in my Acer AP43 instead of a 5x86 and apparently it can be a tricky board on which to […]
Show full quote

meljor wrote:

My acer ap43 pci board will not take it. This board is known here to run it just fine but mine has a slightly older bios and as soon as i jumper it to enable the 4x multi it acts really strange and extremely slow (cache not working).

Running it with slower timings and 3x50 is no problem and almost equal to 160mhz .40fsb gives a bit better timings ofcourse but 3x40 is still slower than 3x50.

So if it is not running properly try the dx4 setting with 3x40 or 3x50. The 16kb cache alone will make it faster than the dx4-100@120.

I very vaguely remember having used a DX4 in my Acer AP43 instead of a 5x86 and apparently it can be a tricky board on which to get the 5x86 running properly...or running at all even.

RacoonRider send me some undocumented jumper settings along with an explanation on how to get a 5x86 running on an AP43 a couple years ago.
I never got around to try these out for myself though, but for the AP43 this might be worth a try and I might as well spread this bit of info in case anybody might ever need it.

_________JP25____JP26
25 MHz___2-3____2-3
33 MHz___1-2____1-2
40 MHz___1-2____2-3 <unlike the manual/th99/PCB prints
50 MHz___2-3____1-2 <not stated in manual

and

The jumper settings are described on page 15-2 of the second part of the manual. However, if you just set it up, nothing will work or even POST! You have to set jumpers for DX4 SL-enhanced WB first, start the system, and only then turn it off and add JP12. I have now idea why it does not work otherwise, I triple-checked everything and still no post. You have to do this only once though.

The EXP-4045 might have undocumented jumper settings as well?

Btw, I recall some people actually managing 200MHz with the AMD 5x86 part, but those might've been some of the more recently manufactured parts.

The EXP-4045 clearly lists the jumpers needed to configure all this stuff, I should have no problem. My issue was not how to configure it, it was has anybody OC'd a 5x86 and not fried it.

Ampera wrote:

The EXP-4045 clearly lists the jumpers needed to configure all this stuff, I should have no problem. My issue was not how to configure it, it was has anybody OC'd a 5x86 and not fried it.

Yes, many people have overclocked the AMD 5x86 and a link to a thread where even a 200MHz clocked AMD 5x86 is benched was already given to you by Jesolo (this one The Ultimate 486 Benchmark Comparison) so your question was basically already answered.
And about the undocumented jumper settings for your particular board, I was just wondering and chances are your board actually does have undocumented jumper settings that might actually be interesting to you.

Tetrium wrote:

Ampera wrote:

The EXP-4045 clearly lists the jumpers needed to configure all this stuff, I should have no problem. My issue was not how to configure it, it was has anybody OC'd a 5x86 and not fried it.

Yes, many people have overclocked the AMD 5x86 and a link to a thread where even a 200MHz clocked AMD 5x86 is benched was already given to you by Jesolo (this one The Ultimate 486 Benchmark Comparison) so your question was basically already answered.
And about the undocumented jumper settings for your particular board, I was just wondering and chances are your board actually does have undocumented jumper settings that might actually be interesting to you.

I can't even find a whole lot of information about the company that made the board, which annoys me. I did find DTC's old headquarters on google streetview though, and that was interesting.

Ever hear of someone who killed their AMD Am5x86 by running it at 160 MHz and less than 3.6 V? I haven't. It would be rare to find one which does not clock to 160 MHz at 3.6 V. People even run them at 3.3 V and 160 MHz without issue, however I recommend 3.45 V to start and if issues present itself, your board might have a jumper setting for 3.6 V, but that is less common.

There are reports of 180 MHz and 200 MHz. I have tried almost a dozen Am5x86 chips, even those with post-millennium datecodes, but could not get 180 MHz running reliably. Good enough for a few DOS benchmarks at 180 MHz and 4 V, but Windows 95 was a no go.

feipoa wrote:

Ever hear of someone who killed their AMD Am5x86 by running it at 160 MHz and less than 3.6 V? I haven't. It would be rare to find one which does not clock to 160 MHz at 3.6 V. People even run them at 3.3 V and 160 MHz without issue, however I recommend 3.45 V to start and if issues present itself, your board might have a jumper setting for 3.6 V, but that is less common.

There are reports of 180 MHz and 200 MHz. I have tried almost a dozen Am5x86 chips, even those with post-millennium datecodes, but could not get 180 MHz running reliably. Good enough for a few DOS benchmarks at 180 MHz and 4 V, but Windows 95 was a no go.

My board is strict 3.3v or 5v, but who knows, I should probably take a quick probe of the VRM to see what it's actually outputting.

And to be fair, I haven't heard anything. I've just started getting serious about retro computing, it's always been a would be a nice thing to do.

From my experience, "3.3V" has meant anything from 3.26 V to 3.42 V.