BitWrangler wrote on 2022-02-23, 14:15:

Y'all are playing it wrong, situations like this are where you snag all the cheap 486 gear while everyone is distracted 😉

But I've not really seen much 486 gear lately that I would be interested in buying.

I am still pining for an 8086 setup for some odd reason though. I know where I can get a motherboard and processor for a decent price but haven't pulled the trigger yet.

0xCats wrote on 2022-02-21, 17:28:

For testing in Windows plain Prime 95 with small FFT is fine for finding unstable L2 cache. No need for large FFT, large FFT fun […]
Show full quote

NostalgicAslinger wrote on 2022-02-21, 16:29:

Nemo1985 wrote on 2022-02-21, 16:27:

I'm curious to see if those cpus are really stable when the cache is enabled, if so this is a game changing discovery!

Prime 95 Blend or large fft test (good for cache) for ~8 hours and 3DMark 2000 in loop

For testing in Windows plain Prime 95 with small FFT is fine for finding unstable L2 cache. No need for large FFT, large FFT functionally adds nothing to a cache test except more time.

Though I generally prefer just the ease of slamming the L2 in rapid succession with memtest using small test address range that kinda breaks memtests anti-caching logic.
From personal testing I know CPU's with bad caches fail this really quickly.

Which version of prime 95 do you recommend and link? My version won’t launch

Thandor wrote on 2022-02-22, 21:14:

Legends say that if you connect all resistors you’ll end up with a magical AMD K6-IV with 384KB L2-cache :P. […]
Show full quote

Am386DX-40 wrote on 2022-02-22, 19:45:

So has anyone tested what happens if instead of moving the 0 ohm resitor it's just taken away? Will it report 0 kb of cache?

Legends say that if you connect all resistors you’ll end up with a magical AMD K6-IV with 384KB L2-cache 😜.

Serious note:
Would be interesting to see if the chip will overclock higher with 0KB L2-cache 😀.

Anyone wanting to try this may consider doing some measurements first, that could be a pull up/pull down config jumper. In which case no jumper would be probably safe but all three would short out

Has anyone found all the grounds on those components yet? .... I has deviouf and diverfe conjurations to plot which require grounds.

cyclone3d wrote on 2022-02-24, 00:18:

I am still pining for an 8086 setup for some odd reason though. I know where I can get a motherboard and processor for a decent price but haven't pulled the trigger yet.

If you get a gold cap 8080 maybe you can pry off the lid and...

.

... idk solder another 8080 die in there so 8+8 makes 16 🤣

Does anyone know which K6-2+ might be soldered and which ones just have thermal paste + silicone attaching the heatspreader? I'm trying to remove the lid on a K6-2+ 550 ACZ, datecode 0039BPMW. It's from a Sony Vaio laptop that had failed and I'm pretty sure it still works (worked?)
My olfa cutter blade made short work of the silicone, but the heatspreader is still attached. Could it be indium attaching it still?
There's enough resistance that I really won't want to pry it...

None of them are going to b soldered. The IHS is aluminum.

My guess is that you don't have the glue cut all the way.

Yeah, verified. Thanks for confirming it probably isn't indium solder. It was just the silicon all the way on top of the core. I took the heatgun to it thinking it was indium, then it just unceremoniously popped off - it still has a core! That's wonderful! 😁
I confirmed that the edge silicon was cleared with some dental floss.

Something like 260C made it pop off with a little bit of pressure from the craft knife.

Thanks for the screenshots and applying a little heat sounds like a great idea, thanks for sharing. I'm also interested in seeing some numbers for your 550ACZ case if you're thinking about posting them. 😀

Thermalwrong wrote on 2022-02-24, 23:31:

Does anyone know which K6-2+ might be soldered and which ones just have thermal paste + silicone attaching the heatspreader? I'm trying to remove the lid on a K6-2+ 550 ACZ, datecode 0039BPMW. It's from a Sony Vaio laptop that had failed and I'm pretty sure it still works (worked?)
My olfa cutter blade made short work of the silicone, but the heatspreader is still attached. Could it be indium attaching it still?
There's enough resistance that I really won't want to pry it...

Use the shaving razors I showed in previous post. Also, others have said acetone can help.

Be sure to reinstall the spreader. Don’t want to have a accident with the die.

maestro wrote on 2022-02-25, 03:10:

Thanks for the screenshots and applying a little heat sounds like a great idea, thanks for sharing. I'm also interested in seeing some numbers for your 550ACZ case if you're thinking about posting them. 😀

I’m going to eventually make a difinitive guide here showing how to take the lid off with no risk, and reinstall it in a way that looks factory.

I’m still working out the details,

But so far what I have is this:

I think acetone, and a shaving razor is the way to go, no heat. I think the heat may actually hurt at the levels required

The material used for the adhesive is still a mystery to me, but it may be urethane.

I’d like to recommend a thermal adhesive for the die but I can’t seem to find artic silver thermal adhesive these days.

Also some sort of 3d printed guide needs to be made for installing the spreader in one motion to avoid the adhesive smearing.

Can someone whip something like this up?

On the die, it's probably just 20 year old thermal paste turned to something more like cement. All the K6-2s I've had the caps off, it started off when they were a year or two old, like new AS-2 or 3, but was stiffening up on the ones I had off 5 years later.

I got some plans about maybe coating the discretes with "liquid tape", fabricating a copper heatspreader and then using indium or other liquid metal on the die.

BitWrangler wrote on 2022-02-25, 04:40:

On the die, it's probably just 20 year old thermal paste turned to something more like cement. All the K6-2s I've had the caps off, it started off when they were a year or two old, like new AS-2 or 3, but was stiffening up on the ones I had off 5 years later.

I got some plans about maybe coating the discretes with "liquid tape", fabricating a copper heatspreader and then using indium or other liquid metal on the die.

For the more extreme among us there are of course the athalon copper shimms

proposed modification:

the ones I've seen appear to be using the same rubbery compound that the lid is attached with as the thermal material. I think as modders we can do better. I worry about using compound that dries out though.

Thermalwrong wrote on 2022-02-25, 00:19:

Something like 260C made it pop off with a little bit of pressure from the craft knife.

Pretty hot. A CPU with organic substrate would be scrap now.

My favorite way to remove the heatspreader is rotation after cutting the sides with a razor blade or stove top scraper.

Hi there, for me it seems that it doesnt work. I removed the smd completely. Now my GA-5AX beeps. One long and two short beebs followed by something thats sound like dööödööö. The CPU still gets warm but doenst work even at 300 MHz. Removing the RAM gives another beep code. So I guess on my K6-2+ 350 AUZ there is maybe only 128 KB Cache in working condition and now it runs into an error when checking 256KB cache. But yeah... thats only a guess. Now I would need a new resistor because my got broken during removal. Does anybody know which specs it has?

3faltigkeit wrote on 2022-02-25, 16:52:

Hi there, for me it seems that it doesnt work. I removed the smd completely. Now my GA-5AX beeps. One long and two short beebs followed by something thats sound like dööödööö. The CPU still gets warm but doenst work even at 300 MHz. Removing the RAM gives another beep code. So I guess on my K6-2+ 350 AUZ there is maybe only 128 KB Cache in working condition and now it runs into an error when checking 256KB cache. But yeah... thats only a guess. Now I would need a new resistor because my got broken during removal. Does anybody know which specs it has?

Thats quite the pickle, but just a quick question . .the board is running a K63+ compatible bios right ?

Because this mod effectively changes the K62+ into a K63+ and it'll identify to the board as such, if the bios doesnt support it it'll throw errors.

As for the resistor that a more difficult question but perhaps one of the others who has some measurements can tell you, you might need to consider buying a dead K62+ and pinching the resistor from that.

It's not real special

0xCats wrote on 2022-02-21, 08:34:

I've just gone and measured the part and it is indeed a resistor (0 Ohm link)

Just solder copper wire on there.

Yeah, you can replace it with a solder link if you need to.

Try a lower frequency like 66x2.5

I think you guys should really consider not doing this to your existing k6 2+ chips, but rather order 570s off ebay. This is yet another less than ideal result from a non 570.

And a reminder that that resistor(0ohm link) is not easy to transplant.

Hi again, yes it is alive again. Connecting the two pads with solder does the trick.
I also tried 66x 3 but doesnt help. So no luck with that one.

Well good that it's back to normal and shame mod didn't get anything, though I'd have suspected a 350 to be bottom of the barrel scrapings in the first place.

BitWrangler wrote on 2022-02-25, 19:01:

Well good that it's back to normal and shame mod didn't get anything, though I'd have suspected a 350 to be bottom of the barrel scrapings in the first place.

Indeed, there was probably a reason it was turned into such a low specimen.

Are you willing to do some risky experiments on this chip for us? (Preferably in a disposable motherboard)

If you can find a way, it would be nice to know if you can find ground and voltage on the left and right cache pads.

Also what happens if you change that upper link?

We now know that the processor glitches out with no link so that helps. (To confirm no link is different than link in 256k position right?)

You may be able to check around for continuity from those pads to various pins. Vcc, vss, vcc2, etc… basically what pins do those pads show connection to.

I suspect pad in middle goes into the core so probably nothing to see there so prob avoid touching it

TrashPanda wrote on 2022-02-25, 17:26:

Thats quite the pickle, but just a quick question . .the board is running a K63+ compatible bios right ? […]
Show full quote

3faltigkeit wrote on 2022-02-25, 16:52:

Hi there, for me it seems that it doesnt work. I removed the smd completely. Now my GA-5AX beeps. One long and two short beebs followed by something thats sound like dööödööö. The CPU still gets warm but doenst work even at 300 MHz. Removing the RAM gives another beep code. So I guess on my K6-2+ 350 AUZ there is maybe only 128 KB Cache in working condition and now it runs into an error when checking 256KB cache. But yeah... thats only a guess. Now I would need a new resistor because my got broken during removal. Does anybody know which specs it has?

Thats quite the pickle, but just a quick question . .the board is running a K63+ compatible bios right ?

Because this mod effectively changes the K62+ into a K63+ and it'll identify to the board as such, if the bios doesnt support it it'll throw errors.

As for the resistor that a more difficult question but perhaps one of the others who has some measurements can tell you, you might need to consider buying a dead K62+ and pinching the resistor from that.

My understanding is that if the board supports a 2+ it will also fully support a 3+

Be careful while testing. Die is really fragile as has been pointed out. Two microscopic pieces broke off from edge of the die and that's it. Finished. It's no longer stable, programs crashes. Tiny possibility disabling cache helps but most probably it's just gold scrap.
Next one I'll glue heat spreader back right away. This first one was for practice and wrecking was inevitable.

fool wrote on 2022-02-25, 20:13:

Be careful while testing. Die is really fragile as has been pointed out. Two microscopic pieces broke off from edge of the die and that's it. Finished. It's no longer stable, programs crashes. Tiny possibility disabling cache helps but most probably it's just gold scrap.
Next one I'll glue heat spreader back right away. This first one was for practice and wrecking was inevitable.

Yes, I’ve updated the OP with a warning to this regard. I got ahold of some grey rtv and it is different. I will try to get ahold of some grey urethane next to see if that is similar. I think hardscape masonary adhesive may have similar properties.

If anyone can drop a link to some artic silver thermal adhesive that would help.
Best thermal adhesive