MattRocks wrote on 2026-01-16, 13:30:

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know. […]
Show full quote

DaveDDS wrote on 2026-01-16, 08:31:

nali wrote on 2026-01-16, 06:37:

... It seems almost nobody knew thermal paste yet ...
At least in the computers world.

That is really weird - I've been working in the computing industry for almost 50 years (since 1978) - lots of various types of hardware/processors - lots of PC hardware and *many* other types before and after the PC - I've seen/used LOTS of thermal paste over the years (since the beginning) ...

If you are using a computing shop that doesn't know what thermal paste is - might want to look for another shop!

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know.

I replaced the CPU and cooler at the same time - I don't know what the cooler was but it had a pad built into it. I remember pressing my thumb nail into it and leaving a mark, then realising I wasn't supposed to have done that.

Only after that I read about overclocking and thermal paste. The magazine was not commenting on stock clocks, so it was implied that thermal paste was necessary for overclocking.

So I bought some thermal paste. I then disassembled and reassembled, with a surprise: That pad had behaved like a soft metal (maybe lead?) and when the CPU got hot the metal melted and stuck to the CPU. When it cooled solid and hard to detach. The thumb nail cut had melted out.

That strange cooler worked really well, mechanically and thermally - the paste never did match that. I spent years looking for coolers like it and never found one again.

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential at the turn of the millennium. It was kinda "breaking out" in late 98 into 99 for overclocking celerons, but those of us who had a clue from electronics hobbies were using it as soon as CPUs needed heatsinks.

However that didn't mean your HSF from the Pentium era always didn't have a TIM of some sort, they came with the rubbery pad stuck on, or they had that metal which is an Indium thermal pad, sometimes an Indium/Gallium alloy, but usually known as indium thermal pads, you can find them if you search. The pre-installed TIM was more of a thing with MMX class upwards, rather than when P54 classics were top end.

My CPUs from DX2 up first saw RS Components white thermal paste, but I only scrounged up the dregs of a tube, and by the time I was onto Pentium class it was Tandy/RadioShack white paste. Otherwise I guess I would be still using RS components tube if I got it new, because they came in the size of a small toothpaste back then, none of these little 3g tubes or syringes like the buddy who did that GPU regarded as single serve, but you should be getting more like 10 applications out of.

Something I knew from the start and has been hard to drum into people, thermal paste has absolutely terrible thermal conductivity... but it's miles better than air. You only want to replace the air gap, not float a heatsink on it. But this is also why some people say a particular paste is awesome, and some say it's terrible. The ones that just squish out of the way work fine for n00bs and large contact areas, whereas the ones which are pastier pastes may often work better in small contact areas, if sparingly applied.

I remember it being a bit amusing to me when the hardware and overclocking pages "discovered" thermal grease in 98ish, I was just trying to find some pages on archive dot org, but they are just out of range for good capture, or had server side scripted pages that didn't preserve. So will be a bit in that memory hole between post 2000 when most web got captured and when everyone faded out of newsgroups towards web BBS and messageboards. Though I did see a slight clue, that the "cyrix mailing list" was big into pastes early, so if an archive for that exists one might see the process of PC enthusiasts "catching on" to thermal pastes back in the day. Not sure if there might be something on Tom's Hardware still in archives, but that got to be a PITA to navigate.

BitWrangler wrote on 2026-01-16, 15:07:

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential a […]
Show full quote

MattRocks wrote on 2026-01-16, 13:30:

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know. […]
Show full quote

DaveDDS wrote on 2026-01-16, 08:31:

That is really weird - I've been working in the computing industry for almost 50 years (since 1978) - lots of various types of hardware/processors - lots of PC hardware and *many* other types before and after the PC - I've seen/used LOTS of thermal paste over the years (since the beginning) ...

If you are using a computing shop that doesn't know what thermal paste is - might want to look for another shop!

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know.

I replaced the CPU and cooler at the same time - I don't know what the cooler was but it had a pad built into it. I remember pressing my thumb nail into it and leaving a mark, then realising I wasn't supposed to have done that.

Only after that I read about overclocking and thermal paste. The magazine was not commenting on stock clocks, so it was implied that thermal paste was necessary for overclocking.

So I bought some thermal paste. I then disassembled and reassembled, with a surprise: That pad had behaved like a soft metal (maybe lead?) and when the CPU got hot the metal melted and stuck to the CPU. When it cooled solid and hard to detach. The thumb nail cut had melted out.

That strange cooler worked really well, mechanically and thermally - the paste never did match that. I spent years looking for coolers like it and never found one again.

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential at the turn of the millennium. It was kinda "breaking out" in late 98 into 99 for overclocking celerons, but those of us who had a clue from electronics hobbies were using it as soon as CPUs needed heatsinks.

However that didn't mean your HSF from the Pentium era always didn't have a TIM of some sort, they came with the rubbery pad stuck on, or they had that metal which is an Indium thermal pad, sometimes an Indium/Gallium alloy, but usually known as indium thermal pads, you can find them if you search. The pre-installed TIM was more of a thing with MMX class upwards, rather than when P54 classics were top end.

My CPUs from DX2 up first saw RS Components white thermal paste, but I only scrounged up the dregs of a tube, and by the time I was onto Pentium class it was Tandy/RadioShack white paste. Otherwise I guess I would be still using RS components tube if I got it new, because they came in the size of a small toothpaste back then, none of these little 3g tubes or syringes like the buddy who did that GPU regarded as single serve, but you should be getting more like 10 applications out of.

Something I knew from the start and has been hard to drum into people, thermal paste has absolutely terrible thermal conductivity... but it's miles better than air. You only want to replace the air gap, not float a heatsink on it. But this is also why some people say a particular paste is awesome, and some say it's terrible. The ones that just squish out of the way work fine for n00bs and large contact areas, whereas the ones which are pastier pastes may often work better in small contact areas, if sparingly applied.

I remember it being a bit amusing to me when the hardware and overclocking pages "discovered" thermal grease in 98ish, I was just trying to find some pages on archive dot org, but they are just out of range for good capture, or had server side scripted pages that didn't preserve. So will be a bit in that memory hole between post 2000 when most web got captured and when everyone faded out of newsgroups towards web BBS and messageboards. Though I did see a slight clue, that the "cyrix mailing list" was big into pastes early, so if an archive for that exists one might see the process of PC enthusiasts "catching on" to thermal pastes back in the day. Not sure if there might be something on Tom's Hardware still in archives, but that got to be a PITA to navigate.

The whole point of thermal paste was to fill microscopic imperfections in the metal surface (on heatsink, and on CPU).

So why did the CPU makers put great big markings on top of their CPUs? Intel had engraved specifications, others had painted logos, and even if they anodised the surface - that surface would then be thicker than un-anodised surfaces.

But to prove thermal paste really was optional, I remember a magazine article where a guy ranked CPUs by how fast they fried an egg in his pan 😁

Got a working floppy drive through an Adaptec EISA SCSI controller!

Bit of a chicken and the egg story. One has to configure the EISA board and cards first, so an ISA I/O with FDD is needed. Then the other FDD controller on the EISA card can be recognized by the system...

Couldn't get it to work on an Asus PCI/E-P54NP4 so I'm happy. 😀

MattRocks wrote on 2026-01-16, 17:12:

The whole point of thermal paste was to fill microscopic imperfections in the metal surface (on heatsink, and on CPU). […]
Show full quote

BitWrangler wrote on 2026-01-16, 15:07:

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential a […]
Show full quote

MattRocks wrote on 2026-01-16, 13:30:

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know. […]
Show full quote

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know.

I replaced the CPU and cooler at the same time - I don't know what the cooler was but it had a pad built into it. I remember pressing my thumb nail into it and leaving a mark, then realising I wasn't supposed to have done that.

Only after that I read about overclocking and thermal paste. The magazine was not commenting on stock clocks, so it was implied that thermal paste was necessary for overclocking.

So I bought some thermal paste. I then disassembled and reassembled, with a surprise: That pad had behaved like a soft metal (maybe lead?) and when the CPU got hot the metal melted and stuck to the CPU. When it cooled solid and hard to detach. The thumb nail cut had melted out.

That strange cooler worked really well, mechanically and thermally - the paste never did match that. I spent years looking for coolers like it and never found one again.

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential at the turn of the millennium. It was kinda "breaking out" in late 98 into 99 for overclocking celerons, but those of us who had a clue from electronics hobbies were using it as soon as CPUs needed heatsinks.

However that didn't mean your HSF from the Pentium era always didn't have a TIM of some sort, they came with the rubbery pad stuck on, or they had that metal which is an Indium thermal pad, sometimes an Indium/Gallium alloy, but usually known as indium thermal pads, you can find them if you search. The pre-installed TIM was more of a thing with MMX class upwards, rather than when P54 classics were top end.

My CPUs from DX2 up first saw RS Components white thermal paste, but I only scrounged up the dregs of a tube, and by the time I was onto Pentium class it was Tandy/RadioShack white paste. Otherwise I guess I would be still using RS components tube if I got it new, because they came in the size of a small toothpaste back then, none of these little 3g tubes or syringes like the buddy who did that GPU regarded as single serve, but you should be getting more like 10 applications out of.

Something I knew from the start and has been hard to drum into people, thermal paste has absolutely terrible thermal conductivity... but it's miles better than air. You only want to replace the air gap, not float a heatsink on it. But this is also why some people say a particular paste is awesome, and some say it's terrible. The ones that just squish out of the way work fine for n00bs and large contact areas, whereas the ones which are pastier pastes may often work better in small contact areas, if sparingly applied.

I remember it being a bit amusing to me when the hardware and overclocking pages "discovered" thermal grease in 98ish, I was just trying to find some pages on archive dot org, but they are just out of range for good capture, or had server side scripted pages that didn't preserve. So will be a bit in that memory hole between post 2000 when most web got captured and when everyone faded out of newsgroups towards web BBS and messageboards. Though I did see a slight clue, that the "cyrix mailing list" was big into pastes early, so if an archive for that exists one might see the process of PC enthusiasts "catching on" to thermal pastes back in the day. Not sure if there might be something on Tom's Hardware still in archives, but that got to be a PITA to navigate.

The whole point of thermal paste was to fill microscopic imperfections in the metal surface (on heatsink, and on CPU).

So why did the CPU makers put great big markings on top of their CPUs? Intel had engraved specifications, others had painted logos, and even if they anodised the surface - that surface would then be thicker than un-anodised surfaces.

But to prove thermal paste really was optional, I remember a magazine article where a guy ranked CPUs by how fast they fried an egg in his pan 😁

Yeah it was kinda stupid the way they did the stampings. What is in those craters and valleys or bridges between high points you see when you zoom into the surface is air though, which is why I said air. I got a real good interface on a Cyrix PR-166 rated, 133 x2 clock and multi stock, by lapping the CPU and the heatsink, that one went to 2x83 and was very snappy in everything but quake.... took a K6-2-350 before it felt improved on in Win95.

edit: didn't put that particularly clearly, it ran at 133 with a 2x multiplier on a 66mhz bus as stock setting.

As the MOSFETS seem fine and the southbridge/VRM controller don's show any sign of shorts (by checking temperatures with the ATX power simply plugged in), I removed all the caps of my MSI KT7 Pro 2 which refuses to start despite having 3.3V on the PS_ON pins. Some of the capacitors show clear signe of bulges, and measuring them, they are indeed pretty dead.

Caps are weird, you get some looking like Mr Staypuft and still doing the business well enough for the machine to run, and some look pretty good, or only mild signs on a couple, but they are dead as a doornail. IMO MSI are about middle of the road on that, don't hammer their caps to death to soon, and aren't real finicky when the values slide a little. Some boards are like "aaargh this capacitor is not in exact specification, it is 2% out, I will freeze up and sulk." and some boards are like "cap is a cap, Cap."

I have bought an Asus P55T2P4 v.3.10 Baby AT only because I always wanted to have an Intel HX board.
Will be paired either with a P166MMX or K6-233 it will be DOS 6.22 games only.
I have read that whenever you disable the 2nd level or even the 1st level cache of those CPU you can slow them down to 486 DX2 levels.
Maybe ideal for Prince of Persia 1, Turrican 1+2, Wing Commander 1+2, U6, U7 part 1+2, U8, Populos 1, Lemmings 1, Monkey Island 1, Cadaver, Heimdall 1, Commander Keen, Gods, Jazz Jack Rabbit, One Must Fall and other speed sensitive games.

Working on finishing off my Pentium MMX 233 build with the Silverstone FLP01 case.

Minutemanqvs wrote on 2026-01-16, 18:33:

As the MOSFETS seem fine and the southbridge/VRM controller don's show any sign of shorts (by checking temperatures with the ATX power simply plugged in), I removed all the caps of my MSI KT7 Pro 2 which refuses to start despite having 3.3V on the PS_ON pins. Some of the capacitors show clear signe of bulges, and measuring them, they are indeed pretty dead.

On mine, the big 4700uF and the 2700 uF near the AGP were correct, but all the small 1000uF/6.3v were suspicious. I have removed and measured all.
I changed them with some 560 uF polymer I got from a dead modern Asus mainboard for Intel i3. Just because I had them and didn't have to wait for an order.
It works with no problem.

What is your cpu ?
I would like to upgrade my Duron 600, but I find conflicting informations about the cpu it can use.
I read as well XP 1800 as 2400 ...

Ozzuneoj wrote on 2026-01-16, 07:22:

Oof... yeah, that's probably not good.

In high school a friend gave me a PNY Geforce 2 MX200 PCI card he said was toast. When he gave it to me it had so much arctic silver squishing out around the edges of the HEATSINK (!) that it looked like it was melting. I asked him what the heck that was about and he said he tried to make it cooler by using arctic silver... apparently he thought that you used one tube per application, so he used the entire thing.

Funny thing, I cleaned it up really well and it worked fine after that. I still have it over 20 years later. I honestly never tried powering it on with the paste gushing out, so I don't know if it was shorted or if there was just so much paste that the card overheated quickly. In either case, it's hard to believe that it didn't suffer permanent damage.

"I paid for the whole tube of thermal paste, I'm going to use the whole tube of the thermal paste"

sunkindly wrote on 2026-01-13, 01:19:

Began the task of replacing a faulty power switch on a Commodore 1084S-P monitor. The switch is out and I was hoping to use an A […]
Show full quote

Began the task of replacing a faulty power switch on a Commodore 1084S-P monitor. The switch is out and I was hoping to use an AT power switch I had lying around, but looks like the prongs are too long.

Off to ordering one!

The attachment IMG_9074.JPG is no longer available

The attachment IMG_9076.JPG is no longer available

New switch is in and everything is functioning as it should, woohoo!

Shponglefan wrote on 2026-01-16, 19:03:

Working on finishing off my Pentium MMX 233 build with the Silverstone FLP01 case.

That case looks super nice, I like it better than the new tower version but I think I'm biased towards desktop cases.

PcBytes wrote on 2026-01-16, 22:09:

Ozzuneoj wrote on 2026-01-16, 07:22:

Oof... yeah, that's probably not good.

In high school a friend gave me a PNY Geforce 2 MX200 PCI card he said was toast. When he gave it to me it had so much arctic silver squishing out around the edges of the HEATSINK (!) that it looked like it was melting. I asked him what the heck that was about and he said he tried to make it cooler by using arctic silver... apparently he thought that you used one tube per application, so he used the entire thing.

Funny thing, I cleaned it up really well and it worked fine after that. I still have it over 20 years later. I honestly never tried powering it on with the paste gushing out, so I don't know if it was shorted or if there was just so much paste that the card overheated quickly. In either case, it's hard to believe that it didn't suffer permanent damage.

"I paid for the whole tube of thermal paste, I'm going to use the whole tube of the thermal paste"

I really wish this site had reactions to mark posts as helpful, particularly when they help us live longer with a 🤣.

nali wrote on 2026-01-16, 20:53:

On mine, the big 4700uF and the 2700 uF near the AGP were correct, but all the small 1000uF/6.3v were suspicious. I have remove […]
Show full quote

Minutemanqvs wrote on 2026-01-16, 18:33:

As the MOSFETS seem fine and the southbridge/VRM controller don's show any sign of shorts (by checking temperatures with the ATX power simply plugged in), I removed all the caps of my MSI KT7 Pro 2 which refuses to start despite having 3.3V on the PS_ON pins. Some of the capacitors show clear signe of bulges, and measuring them, they are indeed pretty dead.

On mine, the big 4700uF and the 2700 uF near the AGP were correct, but all the small 1000uF/6.3v were suspicious. I have removed and measured all.
I changed them with some 560 uF polymer I got from a dead modern Asus mainboard for Intel i3. Just because I had them and didn't have to wait for an order.
It works with no problem.

What is your cpu ?
I would like to upgrade my Duron 600, but I find conflicting informations about the cpu it can use.
I read as well XP 1800 as 2400 ...

There is a Duron 700 on mine. But it was in my "for parts" bin forever, I got it as "dead".
The 2700 uF caps were completely bulged and dead on mine, and 2 of the 4700uF also show severe bulging. The 470uF and 1000uF seem physically fine, but they measure horrible values. I'll see if I can revive this board, it's just a fun project at this point.

In the same lot I also have an Asus A7V which powers up but hangs during POST, this will come next...caps also seem bulged near the CPU.

Shponglefan wrote on 2026-01-16, 19:03:

Working on finishing off my Pentium MMX 233 build with the Silverstone FLP01 case.

That’ll be a nice build when finished. I’m guessing the bottom panel has no provision to hide cables (double layer) and not enough room to keep the PSU cable down low. Would look sweet with a glass top!

sunkindly wrote on 2026-01-17, 00:25:

That case looks super nice, I like it better than the new tower version but I think I'm biased towards desktop cases.

Same, I really love the look of classic desktop cases. Really happy they decided to make a modern desktop case.

AndrettiGTO wrote on 2026-01-17, 16:42:

That’ll be a nice build when finished. I’m guessing the bottom panel has no provision to hide cables (double layer) and not enough room to keep the PSU cable down low. Would look sweet with a glass top!

Unfortunately this case doesn't feature any cable management options. But with it all closed up it's all hidden anyway.

I agree though, a tempered glass top would look pretty sweet with this case.

MattRocks wrote on 2026-01-16, 13:30:

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know. […]
Show full quote

My first PC was a Pentium 200 MMX and it was assembled for me without any thermal paste. I didn't know.

I replaced the CPU and cooler at the same time - I don't know what the cooler was but it had a pad built into it. I remember pressing my thumb nail into it and leaving a mark, then realising I wasn't supposed to have done that.

Only after that I read about overclocking and thermal paste. The magazine was not commenting on stock clocks, so it was implied that thermal paste was necessary for overclocking.

So I bought some thermal paste. I then disassembled and reassembled, with a surprise: That pad had behaved like a soft metal (maybe lead?) and when the CPU got hot the metal melted and stuck to the CPU. When it cooled solid and hard to detach. The thumb nail cut had melted out.

That strange cooler worked really well, mechanically and thermally - the paste never did match that. I spent years looking for coolers like it and never found one again.

What you're describing is similar to modern-day phase-change pads. PTM7950 is one of the best, but it's kinda expensive, and not that easy to find. It DOES however, last a ridiculous amount of time, and actually gets even better after 100 thermal cycles. Lenovo uses it in many of their gaming laptops.

BitWrangler wrote on 2026-01-16, 15:07:

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential a […]
Show full quote

Yes it was also my observation that thermal paste was very little known in the world of PCs until it became somewhat essential at the turn of the millennium. It was kinda "breaking out" in late 98 into 99 for overclocking celerons, but those of us who had a clue from electronics hobbies were using it as soon as CPUs needed heatsinks.

However that didn't mean your HSF from the Pentium era always didn't have a TIM of some sort, they came with the rubbery pad stuck on, or they had that metal which is an Indium thermal pad, sometimes an Indium/Gallium alloy, but usually known as indium thermal pads, you can find them if you search. The pre-installed TIM was more of a thing with MMX class upwards, rather than when P54 classics were top end.

My CPUs from DX2 up first saw RS Components white thermal paste, but I only scrounged up the dregs of a tube, and by the time I was onto Pentium class it was Tandy/RadioShack white paste. Otherwise I guess I would be still using RS components tube if I got it new, because they came in the size of a small toothpaste back then, none of these little 3g tubes or syringes like the buddy who did that GPU regarded as single serve, but you should be getting more like 10 applications out of.

Something I knew from the start and has been hard to drum into people, thermal paste has absolutely terrible thermal conductivity... but it's miles better than air. You only want to replace the air gap, not float a heatsink on it. But this is also why some people say a particular paste is awesome, and some say it's terrible. The ones that just squish out of the way work fine for n00bs and large contact areas, whereas the ones which are pastier pastes may often work better in small contact areas, if sparingly applied.

I remember it being a bit amusing to me when the hardware and overclocking pages "discovered" thermal grease in 98ish, I was just trying to find some pages on archive dot org, but they are just out of range for good capture, or had server side scripted pages that didn't preserve. So will be a bit in that memory hole between post 2000 when most web got captured and when everyone faded out of newsgroups towards web BBS and messageboards. Though I did see a slight clue, that the "cyrix mailing list" was big into pastes early, so if an archive for that exists one might see the process of PC enthusiasts "catching on" to thermal pastes back in the day. Not sure if there might be something on Tom's Hardware still in archives, but that got to be a PITA to navigate.

Tell me about it. All of my socket 5/7 stuff (4 complete Pentium 200MMX PCs, 1 Pentium 100) came with NO TIM whatsoever, just the heatsink on top the CPU. Of course, 486 systems didn't even have heatsinks in the first place. Apparently they worked fine??? I don't even get how. Seems like many idiots worked on computers in the 90's to make a quick buck, but then again, stuff like this wasn't common knowledge then. I do however, expect better from a 'pro'. Even toothpaste would be better than that

TechieDude wrote on 2026-01-17, 18:20:

Tell me about it. All of my socket 5/7 stuff (4 complete Pentium 200MMX PCs, 1 Pentium 100) came with NO TIM whatsoever, just the heatsink on top the CPU.

I think I've also never seen a P2/P3 Slot 1 with paste.
In fact I have no idea when paste became a regular thing, for years I stopped other's computers circa 2003 😀

Most of the Slot 1 processors I have seen have had paste on them. Same with AMD K6-2 processors. Before that, it was definitely less common, though I have cleaned paste from a fair amount of all-ceramic Pentiums over the years too.

GigAHerZ wrote on 2026-01-11, 15:06:

Found these products with the help of AI - i asked for ingredients, searched what is locally available, and then described the i […]
Show full quote

GigAHerZ wrote on 2026-01-10, 15:02:

...
I now need to go to find some things to weld back all other plastic clips that have broken off - the back case has nothing to hold on to. Thankfully i have all the broken pieces. Found some local products with the help of AI to actually weld and recreate ABS plastic. I hope i don't have to just glue the backside onto the monitor somehow...
...

Found these products with the help of AI - i asked for ingredients, searched what is locally available, and then described the ingredients of the products to the AI to confirm if these are the right tools for the job.

Just applied these products on some plastic pieces.
The yellow product is like water, but it makes the Compaq's plastic a bit smeary-pastey on the surface it was applied after 10-15 seconds. (I used cotton swab) Applied it to the both sides of the break.
Then i took the red bottle. This is thicker, like some rubber-glue. Applied a thin coat on one piece on the broken surface.
Then pushed the pieces together and as AI predicted, I could even see some plastic slightly bulging out as it was soft. (I hope I still am in the margin of error for the measurements 😀 )
Now i have to wait at least 24h before I could apply a second coat of the red one to build a bit more material on the pieces.
But so far it's promising - I've never had myself any chemistry that could "melt" the original plastics. I hope it works.

It would be amazing, if at last, i have some sort of procedure and approach to repairing plastics once and for all. I already have 2-part epoxy to create "hard rock, concrete" in places where i have room and don't need any flexibility. (Think of laptop's display hinge screw posts) I really-really hope i have found "the way" now for slightly flexible plastics.

In summary:
Yellow Arrow 901 - Used only initially to soften the original plastics.
Red Arrow 1108 - Used initially as well as later to build additional material. After the first steps, some tiny cracks may appear in plastic - these can also be filled with this 1108 after 24+h.

The original usage for those products is to "glue" together PVC piping for hot and cold water.

So... I think it did the trick.

The yellow bottle just softened/melted the plastics so i could position the pieces in their place and only thing that kept them in their place was their own softened plastic itself. The yellow bottle is a solvent of some kind. It doesn't leave anything behind by itself.

Then after every 24 hours, i added a thin covering from the red bottle. This is like some thick syrup and a bit brownish. That is what is most visible on the pictures. I added it both to the broken areas as well as reinforced the areas that hadn't broken yet. I did maybe about 4 coats, 24h for each coat and then waited about 48h until now.

I tried to put the covers back on now, and i think i succeeded. No parts broke of and as much as i can confirm (while the covers are now fully on) these parts stayed solid.

At last, I have some kind of solution to fix broken clips!
Because I've never had a solution for that problem, while the problem has existed for me for over a decade, it feels like magic!

I can now go and put together a full Compaq Deskpro set!