AFAIK the K6's are not as rugged as the Intel offerings.
So I wouldn't try 2.8v on one.
Besides, if your (OEM) MB didn't have one as option there's a slim to no chance that the BIOS will support it.

So the most I would try is an 233 MMX.

Otherwise you can look at a WinChip 2, but I agree 2.8v is too much for a K6 2/3.

duboisea wrote on 2022-03-21, 03:44:

I have a Compaq Presario 4505 and looking at CPU upgrades. I am currently using a MMX 166 Mhz. The chipset is a 82430VX. […]
Show full quote

I have a Compaq Presario 4505 and looking at CPU upgrades. I am currently using a MMX 166 Mhz. The chipset is a 82430VX.

Just from looking at others benchmarks it sounds like the K6-3 is the best you can get for a Socket 7. It requires a vcore of 2.2 or 2.4v [0]

On the motherboard it only lets me configure the vcore to 2.8v, 2.9v, 3.0v, 3.1v, 3.2v or 3.3v. So the best I can do is probably a 233MMX?

[0] https://en.wikipedia.org/wiki/AMD_K6-III

Unless you have a voltage adapter (which are expensive to get these days) that supports 2.2v-2.4v, don't attempt to use an AMD K6-III CPU at 2.8v. It will run way hotter with the potential that could be fried.

As bofh.fromhell said the safest upgrade is the Pentium 233MMX.

bofh.fromhell wrote on 2022-03-21, 04:12:

Besides, if your (OEM) MB didn't have one as option there's a slim to no chance that the BIOS will support it.

It is possible to run an unsupported CPU provided that the BIOS could perform POST without any issues. In my case an unsupported K6 CPU was identified as a 486DX2-66 during POST without BIOS support but it worked fine and everything else (except the CPU multiplier unless it's K6-2+ or K6-III+) can be configured by software afterwards such as the write allocation feature.

All jumpers off might give you 2.5V, it's a known trick from other motherboards. Measure though.

Is it possible? Perhaps. But I would not recommend it.
K6-3 was already known to run notoriously hot, especially the 2.4v version (with the 2.2v version running a little cooler).
2.8v might simply fry the K6-3. And on top of that K6-3 CPUs seem to not be common anyway.

Do consider that running a CPU out of spec by this much may fry the CPU, but it may fry your motherboard as well.

Definitely wouldn't recommend it.

Yes I'd go with the 233MHz MMX

Hello,

I believe I can help you.

I need to see a good pic of your motherboatd though. High res, so I can read the markings, and uninstructed so I can see model numbers, and parts installed (mosfets, inductors, etc)

If all looks well, you can probably even run a k6-3+ cpu

A note on your chipset: 430vx supports a maximum of 128mb of installed ram. But at least it supports SD ram. So that is a nice consolation prize. Two cl2 pc133 64mb sticks are probably your best bet.

The chipset supports 64mb of cached memory, but we won’t worry about that because k6-3/2+/3+ makes this basically a non issue.

A k6-3 might survive 2.8v but I would advise against doing so. Personally, I wouldn’t Run a 3 at anything above 2.5-2.6

Or a 2+/3+ at anything above 2.3-2.4.

Some reading:

Huge number of K6-2E+/570ACZ CPUs for sale on eBay
Cheap cpus

SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
Make cheap cpus even faster

DIY Bios Modding guide Jan Steunebrink k6-2+/3+ 128gb update your bios to work with these cpus

vrm module project thread roundup, share ideas, make new designs in case you want/need better cpu power

AMD K6 3DMARK, aiming for stars.
Good thread where they push the limits of these cpus and show which hardware to pair with it

Re: Bf0, Bf1, Bf2 Multiplier Mod
I don’t think you’ll need any bf mods though as you’ll probably just use the 2x=6x setting

One final note on the 233 mmx recommendation,

Some 166 mmx chips were unlocked. Sometimes you can simply set the 166 to run at 233 and call it a day. All my unlocked 166 and 200 chips run at 233 no problem. Max clock varies cpu to cpu but most will run at 266mhz so 233 isn’t asking too much of a 166

Which pentium 1 cpus were multi locked?

It should be possible to run K6-2 AHX which were rated for 2.4V (2.3-2.5 to be precise)), there are tons of old overclocking reports of people running them even higher than 2.8V. It should be noted, however, that cooling and power requirements skyrocket at this voltage and the voltage regulator on the mobo might not be up to it.

chiveicrook wrote on 2022-03-21, 14:38:

It should be possible to run K6-2 AHX which were rated for 2.4V (2.3-2.5 to be precise)), there are tons of old overclocking reports of people running them even higher than 2.8V. It should be noted, however, that cooling and power requirements skyrocket at this voltage and the voltage regulator on the mobo might not be up to it.

Depending on how your boards vio/v core is supplied sometimes a resistor over the right pins or some desoldering can lower the 2.8 volt setting to anything you want, understand that many PMMX boards are too weak to supply a k6-3’s higher power demands and will crash due to the power section overheating.

As mentioned earlier, the board *probably* has undocumented voltage settings down to around 2v trying to figure that out through pics.

Typically somewhere there is a voltage block

Jumper 1 will be +0.1v
Jumper 1 will be +0.2v
Jumper 1 will be +0.4v
Jumper 1 will be +0.8v
And all jumpers off will be base 2.0v

But this may not be the case, the settings table will give clues.

Gotta see the board, see what kinda vrm hardware is installed, what the jumper block looks like, see the settings table, etc

Another thing, it would be handy to see the clock gen, may be undocumented settings there also.

Another thing to keep in mind is that K6-3 chips have really gone up in price - current listings on Ebay start at $85.00 and go up from there. It wouldn't be prudent to spend nearly $100.00 for a chip that you aren't sure will work, or risk destroying by overvolting.

@Sphere478 I uploaded the pictures here https://imgur.com/a/F6sk5qC

Thanks for the feedback everyone! I will probably just upgrade to a 233. I am able to play all the games I really care about (Dark Forces 2, Age of Empires 2 and Red Alert 2) just fine. Everything DOS works great on it also (Duke 3D, Blood, Doom, Heretic and Hexen) so I should probably stop fiddling with it.

This is a childhood computer. It is fun now that I have money 30 years later doing stuff that I couldn't afford then 😀

Looks like the SC1150 chip has 4 inputs to control the output voltage, and JP3 connects to VID2,1,0 (pins 14,15,16). Looks like those inputs are normally pulled high and when a jumper is fitted then it pulls the input low. VID3 (pin 13) must be is probably always pulled low. From the datasheet that gives possible outputs of 2.8V to 3.5V. It should be possible make a mod to add a jumper to optionally pull pin 13 high, which would allow for outputs from 2.0V to 2.7V.

SC1150 datasheet: https://www.digchip.com/datasheets/parts/data … C1150CS-pdf.php

[edit: I'm thinking RA18 might be a resistor pack that goes to pins 13,14,15,16, and if so, might be a good place to make a mod for pin 13.
I also made it a bit clearer that VID3 is probably pulled low by the motherboard, but there's no reason it can't be modified to be pulled high to get other output voltages. When I said 'must be' I didn't mean that the datasheet says it has to pulled low, just that we can tell from the voltage options given on the motherboard silkscreen that it is currently pulled low]

[edit2: datasheet for the clockgen is here: https://www.datasheets360.com/pdf/-9214174927401046563 ICS9169C-27
looks like the motherboard fixes it at 66MHz, but the actual clockgen can also do 75 and 83 by changing a couple of input resistors, although this might stop USB from working as the faster CPU clocks also change the 48MHz USB output. But this might not be a problem as I think the chipset can use either the 24MHz or 48MHz clocks for USB]

Here is your clock gen, it can tell us if higher fsb is supported if you can get a better pic

Here appears to be your multiplier settings which you probably already know (you’ll want to set 2x for a k6-3/2+/3+ To get 6x)

Here are your voltage settings as you know,

Your board designers weren’t very kind to you.

You appear to only have these options

I’ll decipher their actual meaning here

Your base voltage is 2.8v (all off)
1-2 gives you +0.1v
3-4 gives you +0.2v
5-6 gives you +0.4v

I’m betting that one of those chips by the donut shaped coils is your voltage regulator controller but need better pic

I’m also betting that there is a half way decent chance that the designers simply hard shorted your +0.8v option, so if you can get a better pic of those chips we may be able to look up the data sheet and find out which resistor or leg you need to desolder to get down to 2.0v if my guess is correct

The good news is your power circuitry looks nice, large heatsinks with switching regulators. Very nice.

I don’t see a option for your fsb (a jumper anywhere) do you see one?

Looks like you have some onboard ram. Which you may actually consider desoldering and putting all your ram in the provided slot. Because you should be able to get 128mb in the slot alone and with better chips. Otherwise you’ll be limited to 64mb plus whatever is on the board. (Your chipset supports max 128mb and with onboard plus 128mb in slot you’ll probably get errors)

Edit: I see snufkin beat me to it, and seems to have been able to read your image. (My phone may be compressing the images)

You should be able to do this mod if the data sheet shows the option for it, snufkin has walked me through these mods a few times. If willing, follow snufkin’s instructions and we may see 2.0v and 75mhz/83mhz before you know it 😀

Thanks Evasive!

https://discord.com/channels/7131124000599572 … 724185529888799

chiveicrook wrote on 2022-03-21, 14:38:

It should be possible to run K6-2 AHX which were rated for 2.4V (2.3-2.5 to be precise)), there are tons of old overclocking reports of people running them even higher than 2.8V. It should be noted, however, that cooling and power requirements skyrocket at this voltage and the voltage regulator on the mobo might not be up to it.

The 2.4v rated K6-2 (and K6-3) chips were really just factory overclocked 2.2v parts and have the same maximum voltage tolerances.

Tetrium wrote on 2022-03-22, 14:46:

chiveicrook wrote on 2022-03-21, 14:38:

It should be possible to run K6-2 AHX which were rated for 2.4V (2.3-2.5 to be precise)), there are tons of old overclocking reports of people running them even higher than 2.8V. It should be noted, however, that cooling and power requirements skyrocket at this voltage and the voltage regulator on the mobo might not be up to it.

The 2.4v rated K6-2 (and K6-3) chips were really just factory overclocked 2.2v parts and have the same maximum voltage tolerances.

Took a look at K6-2 datasheet and... you are right 😀

While theoretically parts with AHX, 400AFQ and AFR suffixes that are marked 2.4V are rated for 2.3-2.5V operation and 2.2V parts are for 2.1-2.3V the absolute maximum rating is the same for both: 2.6V
Max operational temperature (Tcase) ratings are 70C for AFR, 65 Cfor AHX and 60C for AFR.

It is interesting, however, that AGR, AFX and 400AFR absolute maximum rating is 2.5V. 450MHz AHX/AFR dissipate almost 30W at maximum while 450MHz AGR/AFX are barely touching 20W at max.

When it comes to K6-3 2.2V parts absolute maximum rating is 2.4V, while 2.4V parts absolute max is 2.6V.

K6-2+'s absolute is also 2.6V/2.4V (for 2.2V/lowerV parts) but Tcase is up to 80C operational.
K6-3+'s are all 2.2V absolute max but also 85C Tcase.

Didn't check embeddded E variants.

K6-2/III will work fine on 2.8v.

I don't think I'd ever have the guts to run one at 2.8V, but I can say that the K6-3 (original) isn't hard to damage.
I had a K6-3 450MHz 2.2V back when new. My cooling wasn't very good and it ran hot, usually between 60-65C. I had a BIOS alarm set at 65C but I'm sure it could have gotten a few degrees above that occasionally.
After a few months it became unstable. Much more recently I checked this out and reproduced it, and found I have to run it at 2.3V now to make it stable at 450MHz. So it would still qualify as a "2.4V" model, but it was originally marked and working as a 2.2V so it definitely has been damaged.

If you ever decide to attempt 2.8V on one of these CPUs, I'd use a good, but not too heavy heatsink with a powerful fan on it.
A heavy heatsink is risky with socket-7 boards because of how fragile the heatsink retaining hooks are, and you can't clamp them down very hard for the same reason.
The motherboard's onboard power circuitry is also at risk due to the high power demands, so it would need the benefit of a fan that's powerful enough to help cool the board and not just the CPU.

If you have any interest in modifying the board, unlocking the lower voltage range of the VRM would be a much better way to go about it. It might turn out to be simple - as was discussed above the "extra" VID pin is probably connected to something somewhere, you'd just have to find where and change it's hardwired connection from GND to 5V (or even install a jumper, if the designers planned for that).
The absolute worst case would be having to snip/lift one pin of the VRM controller chip so it's off the PCB and then solder a little wire to it.

With a 2.0V setting unlocked, you might be able to run a "+" model CPU which are much better - not only can they clock higher but they're a lot easier on the motherboard electrically. They're very stable and cool vs quirky and hot for the original nonplus version.
After I got a K6-3+ running on my old super-7 board, I quickly decided I'll never go back to the original K6-3 again.

shamino wrote on 2022-03-22, 22:37:

I don't think I'd ever have the guts to run one at 2.8V, but I can say that the K6-3 (original) isn't hard to damage. I had a K6 […]
Show full quote

I don't think I'd ever have the guts to run one at 2.8V, but I can say that the K6-3 (original) isn't hard to damage.
I had a K6-3 450MHz 2.2V back when new. My cooling wasn't very good and it ran hot, usually between 60-65C. I had a BIOS alarm set at 65C but I'm sure it could have gotten a few degrees above that occasionally.
After a few months it became unstable. Much more recently I checked this out and reproduced it, and found I have to run it at 2.3V now to make it stable at 450MHz. So it would still qualify as a "2.4V" model, but it was originally marked and working as a 2.2V so it definitely has been damaged.

If you ever decide to attempt 2.8V on one of these CPUs, I'd use a good, but not too heavy heatsink with a powerful fan on it.
A heavy heatsink is risky with socket-7 boards because of how fragile the heatsink retaining hooks are, and you can't clamp them down very hard for the same reason.
The motherboard's onboard power circuitry is also at risk due to the high power demands, so it would need the benefit of a fan that's powerful enough to help cool the board and not just the CPU.

If you have any interest in modifying the board, unlocking the lower voltage range of the VRM would be a much better way to go about it. It might turn out to be simple - as was discussed above the "extra" VID pin is probably connected to something somewhere, you'd just have to find where and change it's hardwired connection from GND to 5V (or even install a jumper, if the designers planned for that).
The absolute worst case would be having to snip/lift one pin of the VRM controller chip so it's off the PCB and then solder a little wire to it.

With a 2.0V setting unlocked, you might be able to run a "+" model CPU which are much better - not only can they clock higher but they're a lot easier on the motherboard electrically. They're very stable and cool vs quirky and hot for the original nonplus version.
After I got a K6-3+ running on my old super-7 board, I quickly decided I'll never go back to the original K6-3 again.

Another possibility for you needing more voltage is supply noise. Your caps may be due to be changed

Doubt this is the case but I swear I’ve had bios updates solve unstable stock clocks. My 3+ couldn’t go past like 350mhz until I updated the bios on my freeway mobo Or my p5a forget which.