You win the internet for today. Saved for posterity.

Just messing around in the collection, when I discovered that the GigaByte GA-586DX actually had room around the sockets for a couple of GoldenOrbs 😁

Well, then I had to find some other to mill out for the BP6 project.

What I didn't notice was that the sockets are too close to eachother to have room for 2 GOrbs ... 🤣

Thermalwrong wrote on 2024-09-25, 12:43:

Lately I've been getting too many Slot type motherboards which are missing the retention mechanism or the mechanism is broken. So I've been making repair parts, here's the replacement for the top half of a "hometom" branded SECC 2 clip:

Fantastic work! I need to print a few of these. Any chance you could share the 3d print files?

H3nrik V! wrote on 2024-09-26, 18:17:

Just messing around in the collection, when I discovered that the GigaByte GA-586DX actually had room around the sockets for a couple of GoldenOrbs 😁

Well, then I had to find some other to mill out for the BP6 project.

What I didn't notice was that the sockets are too close to eachother to have room for 2 GOrbs ... 🤣

I use these for my 1GHz CPUs https://www.startech.com/en-eu/computer-parts/fan3701u

douglar wrote on 2024-09-26, 17:32:

Thermalwrong wrote on 2024-09-25, 12:43:

Lately I've been getting too many Slot type motherboards which are missing the retention mechanism or the mechanism is broken. So I've been making repair parts, here's the replacement for the top half of a "hometom" branded SECC 2 clip:

The attachment replacement hometom folding secc 2.png is no longer available

That looks great! Is that 3d printed? Can you share a link to your design?

We have a thread here: Retro 3D print topic

zuldan wrote on 2024-09-26, 20:30:

Thermalwrong wrote on 2024-09-25, 12:43:

Lately I've been getting too many Slot type motherboards which are missing the retention mechanism or the mechanism is broken. So I've been making repair parts, here's the replacement for the top half of a "hometom" branded SECC 2 clip:

Fantastic work! I need to print a few of these. Any chance you could share the 3d print files?

Sure, this is the basic version shown in the pictures which honestly I think will be fine for most uses. With a SECC2 bracket the heatsink should lock the slotket / cpu into place vertically and simply pressing in these clips will stop the CPU rocking back and forth.
If it's an older SECC1 bracket then adding an M3 screw into the mounting clip hole will hold it in fully for vertical and side to side motion.

The MS6905 version only really stops it rocking around in the slot not stopping it from going up, but it's a tight fit so it should hold it pretty securely too.

If you do print them out let me know how they fit and if they need any tweaks, they're based off of a mix of measurements from calipers and the Intel SECC documentation so they should be a good fit on any Slot 1 board.
The full SECC Slot bracket I'm still making tweaks to and hopefully can get it able to fold and make a clip version of the SECC2 mounting clip following this design which I think is pretty good:

edit: now v3 of the generic one, realised I put too much material that touches the SECC heatsink

zuldan wrote on 2024-09-26, 20:34:

H3nrik V! wrote on 2024-09-26, 18:17:

Just messing around in the collection, when I discovered that the GigaByte GA-586DX actually had room around the sockets for a couple of GoldenOrbs 😁

Well, then I had to find some other to mill out for the BP6 project.

What I didn't notice was that the sockets are too close to eachother to have room for 2 GOrbs ... 🤣

I use these for my 1GHz CPUs https://www.startech.com/en-eu/computer-parts/fan3701u

The look really cool! Might wanna look into that when slotketing up my P2B-DS 😁

A month ago, my K6-2 system suddenly stopped POSTing. I tried swapping the RAM modules, replacing the CMOS battery, removing expansion cards and adjusting connectors to no avail. I even measured the 5 V-to-3.3 V regulator and found out it was operating within spec.
Without much else to try, I simply assumed the motherboard was dead and left matters there.

Today, I got desperate and thought I'd power the system on one more time. It worked right away!
Once the initial excitement wore off, I got around to testing a silly idea I had cooked up a month ago – to run TempleOS on a processor nowhere near the minimum requirements.
I already knew QEMU could emulate x86-64 on newer processors, allowing Windows 10 to "run" on a Pentium II.
I gave the last Windows XP-compatible build of QEMU a shot, only for it to crash. Running a debugger revealed that the program was trying to execute the now-ubiquitous cmov instruction. I was thinking I could patch it out, but a search in a disassembled listing of the binary came up with over a thousand instances. I wasn't going to spend a quarter of a year on this experiment, so this idea had to go.
I soon switched my focus to Linux. This computer multiboots with Arch Linux 32 – a distribution that happens to provide an i486-compatible build of QEMU. I was golden, or so I thought.
Any attempt to start QEMU with a graphical front-end caused a segmentation fault. The kernel log suggested that the fault took place in the graphics driver, nouveau_vieux_dri.so. The Nouveau website has a section pertaining to this driver, ending with the following message:

Do not file bug reports about this driver.

Sigh. And they say Linux breathes new life into old hardware...
Now that I was back on Windows, I could only hope an older build of QEMU supported the K6-2. I tried one dated November 19, 2011; sure enough, it worked, and I was able to play Varoom at a blistering 12 seconds per frame!

Does flexing my ABIT and MSI mainboards count? 🤣

PcBytes wrote on 2024-09-28, 20:02:

Does flexing my ABIT and MSI mainboards count? lol […]
Show full quote

Does flexing my ABIT and MSI mainboards count? 🤣

It pretty much does! What's the dual one?

H3nrik V! wrote on 2024-09-28, 22:28:

PcBytes wrote on 2024-09-28, 20:02:

Does flexing my ABIT and MSI mainboards count? lol […]
Show full quote

Does flexing my ABIT and MSI mainboards count? 🤣

It pretty much does! What's the dual one?

ABIT's BP6 and MSI is 694D Pro-AR. Former runs 2x Celeron 333, latter runs 2x Coppermine 1000EB SL52R.

Continuing the topic... did my first fun thing with a MMX on my freshly rebuilt (and recapped) Azza/XFX/PineTech PT-5VMD... running a standard MMX@250MHz w/ 100FSB and the Vcore of a K6-3.

Thermalwrong wrote on 2024-09-27, 00:01:

Sure, this is the basic version shown in the pictures which honestly I think will be fine for most uses. With a SECC2 bracket th […]
Show full quote

Sure, this is the basic version shown in the pictures which honestly I think will be fine for most uses. With a SECC2 bracket the heatsink should lock the slotket / cpu into place vertically and simply pressing in these clips will stop the CPU rocking back and forth.
If it's an older SECC1 bracket then adding an M3 screw into the mounting clip hole will hold it in fully for vertical and side to side motion.

The MS6905 version only really stops it rocking around in the slot not stopping it from going up, but it's a tight fit so it should hold it pretty securely too.

If you do print them out let me know how they fit and if they need any tweaks, they're based off of a mix of measurements from calipers and the Intel SECC documentation so they should be a good fit on any Slot 1 board.
The full SECC Slot bracket I'm still making tweaks to and hopefully can get it able to fold and make a clip version of the SECC2 mounting clip following this design which I think is pretty good:

The attachment slot1-secc-bracket-clipversion.webp is no longer available

edit: now v3 of the generic one, realised I put too much material that touches the SECC heatsink

Thanks so much for that.

How is the the full SECC Slot bracket going?

Any chance you could share the file for the replacement folder secc2?

PcBytes wrote on 2024-09-28, 20:02:

Does flexing my ABIT and MSI mainboards count? 🤣

Very nice collection but where is your ABIT Fatal1ty AN8 SLI with Guru Panel 😉

Completed my hardware downsizing exercise so that what I've got left makes more sense and might actually be used one day (or is reasonable spares and alernatives).

[So, off to ebay to buy some more then....]

zuldan wrote on 2024-09-28, 23:24:

PcBytes wrote on 2024-09-28, 20:02:

Does flexing my ABIT and MSI mainboards count? 🤣

Very nice collection but where is your ABIT Fatal1ty AN8 SLI with Guru Panel 😉

Honestly, if I ever hope to get a Fatal1ty board, it's probably going to be an ABIT AA8XE and a ASRock Z68 Professional Gen3. Or a FP-IN9, if I ever find any.

Does this count as a retro activity?

Only if you use tinkertoy computer principles, technic gears, and a large school gym and make a working version. 😜 .... just kidding, it's kinda neat.

PcBytes wrote on 2024-09-28, 20:02:

Does flexing my ABIT and MSI mainboards count? lol […]
Show full quote

Does flexing my ABIT and MSI mainboards count? 🤣

Heyy it's the MS-6168 again 😀 I'm still blown away that you were able to get the V3 on there up to 16MB just by changing those resistors: Re: Bought these (retro) hardware today

zuldan wrote on 2024-09-28, 23:20:

Thanks so much for that. […]
Show full quote

Thermalwrong wrote on 2024-09-27, 00:01:

Sure, this is the basic version shown in the pictures which honestly I think will be fine for most uses. With a SECC2 bracket th […]
Show full quote

Sure, this is the basic version shown in the pictures which honestly I think will be fine for most uses. With a SECC2 bracket the heatsink should lock the slotket / cpu into place vertically and simply pressing in these clips will stop the CPU rocking back and forth.
If it's an older SECC1 bracket then adding an M3 screw into the mounting clip hole will hold it in fully for vertical and side to side motion.

The MS6905 version only really stops it rocking around in the slot not stopping it from going up, but it's a tight fit so it should hold it pretty securely too.

If you do print them out let me know how they fit and if they need any tweaks, they're based off of a mix of measurements from calipers and the Intel SECC documentation so they should be a good fit on any Slot 1 board.
The full SECC Slot bracket I'm still making tweaks to and hopefully can get it able to fold and make a clip version of the SECC2 mounting clip following this design which I think is pretty good:

The attachment slot1-secc-bracket-clipversion.webp is no longer available

edit: now v3 of the generic one, realised I put too much material that touches the SECC heatsink

Thanks so much for that.

How is the the full SECC Slot bracket going?

Any chance you could share the file for the replacement folder secc2?

The attachment replacement hometom folding secc 2.png is no longer available

Okay, but please be aware it will only fit the SECC 2 folding clips that have "HOMETOM" as the brand on the side, also I didn't model up the bottom part since it was intact and there's another plastic part that you push down to unlock the CPU, again didn't model because it was intact. To have strength in the orientation it needs to, the part has to be printed on its side with organic supports:

edit: it seems to be quite a common slot mounting mechanism on MSI slot 1 / A motherboards

Yeah, figured I either missed something or MSI pulled a fast one with the RAM chips on my board, especially since the model numbers on the Samsung chips I had matched a Matrox G400, of all things. It's been my go-to Slot1 3dfx mobo. I'm not fully sure if I should replace the current passive heatsink for an active cooler saved off a Mac Rage128 Pro, since the card seems to hold stable with its OC to v3 3k levels, and it scores fairly well in 3dMark 99 - around 5-6k 3DMarks.

Otherwise, it's either Riva TNT2 Pro for most of the pre-nF2 era, though I do alternate between it and an actual v3 2000, found in the most mundane retro build I had bought. (with a Fastfame mobo, of all things.)

I wonder if they started out using "the good half" of cheap and defective RAM chips, then when that initial supply dried up, and maybe they got cheap enough just normal fully good ones.

Probably the latter, given the original design was made for Packard Bell.
They may have dried up EtronTech supplies (since the true 8MB boards I've seen are EtronTech/EliteMT mostly) and resorted to halving 16MB boards by not populating R658 and R604.
Those "16MB-crippled-to-8MB" boards are easily identifiable as on those, R686 and R687 are already populated with 0 ohm links. The standard 8MB have R686 and R687 empty, and only R686 is populated.