Performed other benches at 233 MHz with the M550, I put here the most recent ones, but there would be others, I will only put the values in a summary table, currently I did those at 133 166 and 233, those at 200 are missing.

Of course I won't put those for slow PCs, in short the resulting value is higher than the one it displays, that is, if at most it writes 99.9, it can't write 123.4, maybe we'll only see 23.4, and we'll think it's slower, but it really isn't like that.

When I'm done with the P.MMX tests, I'll take a few days off, to try some other motherboard, always like S.7, I would have a Compaq Camaro chipset MVP4 (Mitac), I think a BIOS rewrite is also needed there, unfortunately it's not easy to do, because the PLCC32 BIOS chip is soldered, to solve the problem you should remove, reprogram and re-solder, but I thought it could be avoided, with a special extension that connects the soldered chip to the USB programmer, this is a project I should carry out, but it's not simple, because you have to build a fake BIOS chip, and Connect the socket to be inserted into the soldered chip, with a cable like that of the floppy, only two of the thirty-four would remain unused

Unfortunately, despite having a couple of COAST modules (L2 cache), it is proving quite difficult 😩, to understand on which MB they are good 😌, yes because from what it seems, the pinout does not follow a standard, this means that any manufacturer can change the order of the pins at will.

Another problem is that on the modules, there are no indications 🗺, on the manufacturer, the difficulty is maximum, except that you have a card with a module already installed, theoretically it could work on similar motherboards of the same manufacturer, but it may not be so, a different chipset may need different pins, in short, the ideal solution is to find an MB equal to the one you have, and find the same COAST module, only then we will be sure that it works, and that we will not risk our MB unnecessarily.

I'm still doing some tests, with the PCCHIPS M550, for now I have an Intel Pentium MMX 233, and the L2 cache is the one present on the 512 KB MB, I think it's also the maximum for the Intel 430TX chipset, at least that I know there are no TX cards with 1 MB of cache, on other subsequent motherboards, like the Tyan and others, I have 1MB cache, but the chipset is VIA, I think it's MVP3, but those cards have full CPU support with FSB 100 or 95, mainly AMD K6-2 or higher, and if I'm not mistaken too Some of the last CYRIX MII, but I don't have any of the latter, if I'm not mistaken they are later than those with PR333, so up to PR433, and with 2.2V VCORE voltage.

I think there may be other CPUs with FSB 100, but I'm not sure, maybe the RISE MP6? However, we are talking about the factory FSB 100, theoretically all CPUs with FSB 66, can go to 100 in overclock, but reducing the multi, yes because not all of them like extra frequencies of over 30%.

I summarize, I used the PCCHIPS M550 motherboard with a 233 MHz Intel Pentium MMX CPU, which can be used by changing the multi, from 133 MHz (2X), 166 MHz (2.5X), 200 MHz (3X), up to 233 MHz (3.5X), of course using an FSB other than 66, you can get other frequencies, even higher than 233 MHz, the minimum should be 100 💯 MHz (50X2), later I will try lower and higher FSBs.

Let's say that for convenience, my and yours I used Phil's DOSBENCH, in reality I already used many of the programs present, plus others that are not present, for the moment I limit myself to reporting the scores of some of Phil's benches, which are the most important, and which later we will compare with those that I will do by changing CPU, just to remember, I will change the Pentium MMX with an AMD K6, and then with a K6-2, I will then be able to reach the same frequencies with the K6 (133-233), while with the K6-2 I will be able to get up to 400, and even more.

I haven't ⚓️ decided yet, but maybe 🤔 just as a curiosity, I could also use a CYRIX MII 300, just to see the scores, and if at least in some bench, it gets interesting scores 🧐

So let's see what 😨 I got, and with which programs, at the frequencies of 133 166 200 and 233:

Let's start with 133 MHz

1) 90.9 2) 90.6 3) 67.3 4) 26.2 5) 33.8 6) 15.7

A) 217.12 b) 61.22 c) 32.6 d) 14.3 e) 12.7 k) 452.0 l) 1068.25 e 2235.69 n) 100.63

At 166 MHz

1) - - 2) 97.8 3) 74.0 4) 29.6 5) 38.4 6) 17.5

A) 233.41 b) 65.23 c) 36.5 d) 15.8 e) 14.1 k) 565.0 l) 1335.31 e 2794.64 n) 125.80

At 200 MHz

1) - - 2) 103.4 3) 79.4 4) 32.0 5) 42.2 6) 19.1

A) 244.89 b) 67.90 c) 39.8 d) 17.0 e) 15.2 k) 678.0 l) 1602.38 e 3353.55 n) 150.96

At 233 MHz

1) - - 2) 107.7 3) 83.4 4) 33.8 5) 45.2 6) 20.2

A) 254.91 b) 69.61 c) 42.1 d) 17.8 e) 16.0 k) 791.0 l) 2430.29 and 3912.52 n) 176.12

That's all for the moment, later we'll add other benches at different frequencies.

I continue with the benches on the M550, today I had little time ⏱, because I am reorganizing the workbench area, I did some bench at the frequencies of 125 MHz (2.5X 50) and 188 MHz (2.5X75), the first turn to FSB 50, it was a mistake in the change of FSB, but now that I was there I detected the scores and wrote them down, ditto for FSB 75, except that the selection was right.

Seeing the scores obtained at 188 MHz, I remain quite perplexed 😮, we are in some cases very close to those at 233 MHz, and I wonder why Intel had left as FSB maximum 66, the others (Cyrix first) went further, even reaching 100 MHz on the S.7, and I don't think the problem was insurmountable, after all it is even possible to go specifically with the PCI, with an update of the frequency generator, probably the S.7 was left to the competition, in fact there are no Slot1 CPUs not based on Intel CPUs, with the 370 instead, VIA released the C3, But in my opinion, that CPU was an unsuccessful attempt to compete with the Celerons, maybe 🤔 they cost less, but from what I got to see with a 600, I would spread a pitiful veil, I was impressed by a negative, I prefer any K6-2 around 400 - 500 MHz.

In the next few days, I'll try to complete the benches with FSB 50 and 75, and then maybe 🤔 I'll also try with 55 and 60, so as to finish those with the Pentium MMX, I'll see if I can run some other program, like QTPRO or others where you get scores, DOS programs of course.

Basically, the problem I solved in the M550, was due to a BIOS maybe poorly updated, unfortunately I couldn't see anything on video, maybe because the monitor I used can't show any image, because the resolution should be changed manually, however I managed to write the BIOS (correct) by copying it into a floppy disk, the rest did it by itself (automatically), in short, luckily you want because now I have taken some MB, I know that it is possible to avoid programming by disconnecting the chip, or with hot reprogramming techniques, the latter is very Simple, but you can easily fail if you don't use the necessary attention, yes because in chips with pins on two sides, you have to avoid inserting the chip in reverse, but there is an alternative technique, which I recently discovered, that can allow us to reprogram the BIOS and resurrect our MB!!!!!

What is the name of this technique?

Technically I don't know, in short it's about shorting a couple of pins, and this causes the Bootblock to be activated at startup, the rest must be done via floppy drive, it is important to use a VGA ISA (and not a PCI or AGP), to display the guidelines.

Everything is very simple, at the end of writing the file, and reprogramming, the PC is working again!!!!!

Also on the PCCHIPS M550, with Intel Pentium MMX 233, I completed Phil's benches with FSB 75, I summarize the results briefly here, unfortunately if you want to compare them, you can do it by displaying those with FSB 66, there are no surprises in short it was predictable that at 75 the results were better, but I noticed that with the same multi, the performance at 75 reaches even exceeds those with FSB 66 and multi +.5X, this could be even more evident, if I try the FSB 83, the frequency at multi 3.5X would rise up to almost 300 MHz, and here I think Which would be very evident, even visually, that the PC is faster and more responsive, of course with operating systems of the period, in practice Windows 9X, the later ones may need a certain amount of memory (greater), currently I only have 32 MB.

The scores are these: FSB 75

150 MHz 2X

1) — 2) 99.8 3) 71.9 4) 29.7 5) 37.3 6) 17.7

A) 237.11 b) 68.09 c) 35.9 d) 15.7 e) 14.3 k) 509.3 l) 1203.71 e 2519.19 n) 113.40

188 MHz seen by the BIOS as 190, 2.5X

1) — 2) 107.6 3) 78.8 4) 33.5 5) 42.2 6) 19.8

A) 252.33 b) 72.09 c) 40.1 d) 17.3 e) 15.9 k) 636.6 l) 1504.65 e 3148.99 n) 141.75

225 MHz seen by the BIOS as 233, 3X

1) — 2) 113.7 3) 84.2 4) 36.3 5) 46.3 6) 21.5

A) 264.86 b) 75.14 c) 43.6 d) 18.6 e) 17.1 k) 763.9 l) 1805.58 and 3778.82 n) 170.10

263 MHz seen by the BIOS as 275, 3.5X

1) — 2) 118.3 3) 88.1 4) 38.4 5) 49.4 6) 22.8

A) 274.60 b) 76.92 c) 46.2 d) 19.5 e) 18.0 k) 891.3 l) 2106.51 and 4408.60 n) 198.45

I would have expected 263 to be displayed as 266, but it is seen as 275, this is a strange frequency, I have no idea if there was a CPU planned with such a frequency, that it never went on sale.

PC@LIVE wrote on 2024-10-15, 21:01:

Basically, the problem I solved in the M550, was due to a BIOS maybe poorly updated, unfortunately I couldn't see anything on vi […]
Show full quote

Basically, the problem I solved in the M550, was due to a BIOS maybe poorly updated, unfortunately I couldn't see anything on video, maybe because the monitor I used can't show any image, because the resolution should be changed manually, however I managed to write the BIOS (correct) by copying it into a floppy disk, the rest did it by itself (automatically), in short, luckily you want because now I have taken some MB, I know that it is possible to avoid programming by disconnecting the chip, or with hot reprogramming techniques, the latter is very Simple, but you can easily fail if you don't use the necessary attention, yes because in chips with pins on two sides, you have to avoid inserting the chip in reverse, but there is an alternative technique, which I recently discovered, that can allow us to reprogram the BIOS and resurrect our MB!!!!!

What is the name of this technique?

Technically I don't know, in short it's about shorting a couple of pins, and this causes the Bootblock to be activated at startup, the rest must be done via floppy drive, it is important to use a VGA ISA (and not a PCI or AGP), to display the guidelines.

Everything is very simple, at the end of writing the file, and reprogramming, the PC is working again!!!!!

That's a neat trick 😀 You're shorting address pins A15 & A16 which makes it only able to read about half of the BIOS data. There must be some routine that only activates the bootblock recovery BIOS if it finds a bad checksum on the BIOS contents it reads out, rather than using that recovery mode by any kind of default.

Yes Thank you very much, I discovered it recently, and it's a trick I saw here:
https://m.youtube.com/watch?v=45UQ8Y8rl4s&t=5 … AGPBJACAQ%3D%3D
I didn't know exactly what it did, connecting those two pins, I think it's like you suggest, the trick is not mine, but probably in the future, I'll give it a try on some motherboards, I imagine it can also be done with PLCC32 chips, but the pins could be different, and it's not necessarily close to each other.

Dedicating more time ⏱ to the M550, I temporarily suspended the repair of other motherboards, but I watched some YouTube videos that show some repairs on apparently failed MB, in one in particular, a 486 PCI, shows setting errors that bring the PCI frequency to 66 MHz (instead of the usual 33), contrary to what I thought, the video card is not damaged, but it does not work at such a high frequency.

Another problem shown, chipset PINs, apparently ok, but in reality some move, a sign that the welds need to be redone, this operation for me at the moment is impossible, I would only do damage and ruin the motherboard, you don't need an air welder, but the welding of the PINs is revived, with a welder with a concave tip, different is the case of having to replace it (the chipset or similar), there the air one would be the right tool 🥁.

Maybe 🤔 taking inspiration from these videos, I could try some 486 MB, I would really like to fix one, it's one with a PCI SKT3 slot, I keep cards especially for similar motherboards, VGA and Audio for the most time, in short, finding a working one these days, it can be very expensive.

Since I have written about it before, in addition to that 486 PCI, I have other more classic 486, ISA and VLB, some work, but they have problems to solve, one the keyboard one the L2 cache, these are solvable problems, but they are probably due to corrosion (keyboard) of the tracks, and to something broken (cache), it is necessary to investigate well, and follow the traces, maybe renew some welding, while for the cache problem, I would have to do some tests with other working chips, just to exclude that they are not them, which I remember having done in the past, it could be something else So, maybe a look with the microscope could be a good idea.

Going back to the 486 PCI, aesthetically it is not like new, I notice that the welds of the chipset chips are quite dark, they are usually silver mirror type, I wonder if you can do a polishing treatment, something that is not abrasive, apart from that, I also see some residue near the electrolytic capacitors, maybe they changed them, thinking that they were the problem, which in reality I don't know what it is, could it work?

Purchased as not working, this does not mean that the problem is not solvable, assuming that there is one, yes because sometimes as I have said other times, an incorrect jumper setting or anything else, like a low battery, can make you believe that it is broken, the only thing to do is a control test, for any short, it is a subsequent boot test, to see if the post card shows some code, in that case you can understand what problem there may be.

The M550 chapter always remains open, but a lot depends on the time ⏱ that I can dedicate to it, as I would also like to be able to work on other PCs, like the 486 or as soon as possible, on Slot1, in particular there is the OLI-BX waiting for the BIOS chip (new), but I will come back to write about this, when the chips to be reprogrammed arrive.

For the moment I have finished the benches at 66 and 75 MHz (FSB), I have put the jumpers back at 66 and multi 1.5/3.5X, which correspond to 233 MHz, and I have downloaded the ASTRA DOS program, so that I can try it and see what info it shows, and if they correspond to reality.

I didn't examine item by item, but I saw that the most important info is correct, there is no performance index, but for that there are more specific programs.

Hello!
I have a question for you and your expertise.
Are you able to put another cache chip on this motherboard?

Attached:

Mamba wrote on 2024-10-20, 07:32:

Hello! I have a question for you and your expertise. Are you able to put another cache chip on this motherboard? […]
Show full quote

Hello!
I have a question for you and your expertise.
Are you able to put another cache chip on this motherboard?

Attached:

Hello and welcome back, so re-welding is much easier than unsolding, lately I have seen some YouTube videos, in which they disconnect and reattach similar chips, visually it is not particularly complicated, but they use professional equipment, which makes my work easier, with those tools you can change various similar chips in a short time, and therefore it seems quite easy but it is not.

At least speaking for me, my equipment is not at that level, and I have never tried to do these micro welds, you need a soldering iron with a suitable tip, and a magnifier (microscope) that focuses the image at a certain distance.

In conclusion at the moment, I can't do these welds, but in the future I think I'll equip myself better, because I'm currently quite limited, that is, I can't do some jobs, not because I wouldn't be able to, but because I don't have the right tools, and trying without, you risk a lot, maybe for the chip welding the right solder could be enough, and a certain eye.

I downloaded the manual of your Soltek, the motherboard is very interesting 🧐, and I understood why it has only one cache chip, it exists in three different versions, with cache of 512 KB 1 MB 2 MB, in theory you should add a second chip of 512 KB, to get a cache of 1 MB, I see a writing of 128K under the chip that will be soldered, this initially can be deceiving, theoretically welding another chip equal, you get the doubling of the L2 cache, but I don't know if it is necessary to set jumpers for recognition, in short it should be quite simple to do, But I don't know how much it can affect the performance, the gain could be very low, other than the case where there was no cache, there the difference would be visible (enough).

I can't say how much better it does, a card with 1 MB of cache 🫡 compared to one of 512 KB, but if you use Frequencies over 100 FSB, or CPU K6+ these cache chips on the MB, can in some particular cases, be deactivated, because at high frequencies they create stability problems, and in the final performance an MB with FSB 133 without cache, is faster than one with more than 100 with cache, that's if the L2 cache is integrated in the CPU, you can go beyond the limit frequency of the cache chips present on the MB.

PC@LIVE wrote on 2024-10-20, 13:15:

I downloaded the manual of your Soltek, the motherboard is very interesting 🧐, and I understood why it has only one cache chip, it exists in three different versions, with cache of 512 KB 1 MB 2 MB, in theory you should add a second chip of 512 KB, to get a cache of 1 MB, I see a writing of 128K under the chip that will be soldered, this initially can be deceiving, theoretically welding another chip equal, you get the doubling of the L2 cache, but I don't know if it is necessary to set jumpers for recognition, in short it should be quite simple to do, But I don't know how much it can affect the performance, the gain could be very low, other than the case where there was no cache, there the difference would be visible (enough).

I can't say how much better it does, a card with 1 MB of cache 🫡 compared to one of 512 KB, but if you use Frequencies over 100 FSB, or CPU K6+ these cache chips on the MB, can in some particular cases, be deactivated, because at high frequencies they create stability problems, and in the final performance an MB with FSB 133 without cache, is faster than one with more than 100 with cache, that's if the L2 cache is integrated in the CPU, you can go beyond the limit frequency of the cache chips present on the MB.

Yes, correct.
Basically the only thing that differentiates the three models is the amount of cache.
That us why it is a perfect test bed.
But thank you for your answers.

I saw several YouTube videos on the repair of ASUS P2B MB, which give interesting ideas for my Olidata OLI-BX, regarding the MB, I realized that in addition to the BIOS chip, the battery holder is also missing, it's not a big problem, you can find it in scrap motherboards, and it's not even difficult to weld it, but you can also buy new ones, in case we don't have one.

Basically, this ASUS P2B is a SLOT1 that depending on the revisions, you can also mount the Coppermine, more than anything else depends on the chip used, to adjust the VCORE, there are two types, one with 1.8V minimum voltage and the other 1.3V, the 1.8V one does not support Coppermine, even if they can be slightly overwhelt, to make them work anyway.

If you have the necessary skills and the appropriate tools 🛠, you can replace the 1.8V chip with the 1.3V one, and if I'm not mistaken you don't need to modify anything else.

A little while ago I changed the CPU on the 430TX M550, I put the 3.2V K6-233 that it originally had, I had replaced it to avoid problems, with a 166 MMX definitely working, later changed with a 233, to try all the possible multi and the FSB 75, but having concluded the tests at various frequencies, I decided that it was time to put the K6-233.

This CPU, honestly did not give me the impression of being better than Intel's 233, the results of the benches however show scores in some benches clearly higher, while some are at the level of a 166, these low scores, make me think that there is a greater use of the FPU, and that compensated by greater efficiency (at the same frequency on integers), a 1X of multi is lost overall, perhaps at the end of the 90s, the use was almost totally with integers, and AMD did not bother to make improvements on the FPU, but after all neither the Competition did better.

The first thing I notice is that ASTRA doesn't work, I think it could be a problem with the program, while with Phil's benches, the program of the letter k, doesn't work and the PC crashes, the others I've tried work, let's see the scores briefly:

2) 130.0 3) 84.5 4) 33.8 5.8 51.0 6) 22.0

A) 254.05 b) 67.35 c) 36.5 d) 15.5 e) 13.9

L) 3039.40 and 3683.11 n) 266.78

From bench l) you can see that the FPU does not score high, while that of the i233 does about double, if it were so here too we would have to get about 6000.00, instead of over 3600.00.

I tried to overclock at 263 with FSB 75, yes at a first start it works, but then it doesn't load the OS, then it crashes when starting the BIOS, maybe you could try at 250 MHz with FSB 83, but I chose not to use this FSB, to sum up, the i233 CPU is definitely better for any overclocks, while the K6-233 doesn't, but if used at standard frequency, for Office Home use, it could be a valid alternative.

Continuing with the K6-233, I had the time to try the QTPRO, the integrated bench provided scores, in my opinion a little lower than what you would expect, perhaps for use with Game programs, these scores are more realistic, and provide an idea of how competitive it really is, in short, apart from the home Office use, for the rest it seems that the Pentium MMX 233 is slightly better.

The reason is that the AMD CPU was probably developed a little quickly, and even in subsequent developments, see K6-2 and 3, the part of the FPU was not improved, perhaps because it was thought that the 3DNow! Could make up for it, that is, it would have improved performance quite a bit with decimals.

Unfortunately, as said several times, these additional instructions were only partially supported, and therefore the theoretical advantage was lost, and the overall performance remained lower than the same-frequency Pentium MMX.

In the next few days I should get the BIOS chips, which I can reprogram, one would go on the Slot1 Olidata OLI-BX, which for those who don't know, would be the same as ASUS P2B, in practice one of the first with BX chipset, as the name suggests, is a motherboard designed for Pentium II, but with an adapter the P3 S.370 could also work, but if it works it depends on the VRM chip, if it is possible to select less than 1.8V, any Coppermine can be put, and it should work, in short if so, there is even the possibility of putting a modified P3S, that is, that has a mod ai Pin, I have one like this on an ABIT, today buying another one costs a lot, but you can find Celeron 1200 to modify, of those you can find the 1300 as an alternative, and to make a long story short, a 1200 can easily go to 1600 (it takes luck), while the 1300 could reach over 1700, but the difficulty is much greater.

So the next motherboard I put on the bench, will be this OLI-BX, and I hope I don't find insurmountable difficulties, I would really like it to work again.

Among the other cards I have to repair, there is an ASROCK for P4 478 chipset VIA, of this card I should have a couple working, plus others (one or two?) To be repaired, one is very dirty, it seems that it has been in a workshop for a long time, or in an environment with a lot of smoke, brushing it I think is not a good idea, but seeing the YouTube videos, I see that the MBs are often washed, even with only water, or maybe adding a degreaser, maybe you could use detergent, but I think I would use hot water, and initially only in one part, just to see the difference, but in my case, given the dirt, the board should be completely immersed in water, then maybe finished washing, yes It should dry as soon as possible, I don't know if I should use a hair dryer.

If the washing of the MB, is an operation that must be done, only in special cases, or in the presence of a lot of dirt, or to return a very dusty MB as new, here suggestions or experiences in this regard, would be necessary.