I have an 400ATZ that did do 600, but the board got uncomfortable hot at places while doing so. It was @ 2.2 or 2.3v while doing so otherwise it didn't work.

Don't know for sure which board it is, will check it out when I go get everything from my parents place, renovating my own house so I don't have much access at it right now. 😢

Well thats my K6-2+ system:

It scores 1352 in 3DMark2001SE: ftp://78.46.141.148/pictures/bench_results/K6 … _quadro2pro.png

The CPU is a K6-2+ 570.

I have a AMD K6-3+ that do 600mhz, but it was unstable in normal use, just stable enough to run benchmarks. As others have shown you will experience diminishing returns above 500mhz on the K6. There is almost no point with the added heat and instability issues. I run my K6-3's at 500mhz. If the boards supports FSB above 100mhz, I'd rather increase that 😀

PhilsComputerLab wrote:

There is a reason AMD stopped at ~ 570 MHz. Just my 2c 😀

I think too that K6-2 core with or without the internal cache showed its limits (on a power usage/performance result "logic comparison") at 550/570Mhz point and everything above should have been just useless in benchmark numbers, TDP and with the newer solutions (Athlon, PIII) at the door.

Someone posted voltage scaling results with different K6-3+ steppings here on Vogons. Almost all CPUs could do 600 MHz in that test but it depends just as much on the motherboard as on the CPUs.

While almost all K6-X+ CPUs can do ~600 MHz in one of the best boards (in perfect shape) 550 MHz is perhaps the realistic goal for most. I have 4x K6-3+ 400 and two K6-2+ 500, I have tested both K6-2+ and one of the K6-3+ and they can all do 600 MHz at >2.2V with a standard Intel Socket-370 cooler using a Gigabyte GA-5AX rev 5.2 motherboard. I will test the rest of the K6-3+ CPUs... later...

vetz wrote:

I have a AMD K6-3+ that do 600mhz, but it was unstable in normal use, just stable enough to run benchmarks. As others have shown you will experience diminishing returns above 500mhz on the K6. There is almost no point with the added heat and instability issues. I run my K6-3's at 500mhz. If the boards supports FSB above 100mhz, I'd rather increase that 😀

Yeah, In fact I was looking for the official released ones cause overclocking really make no sense (even less with the original K6-2 cores). But certainly one big problem still is memory performance...

A cool project would be to solder together some memory sticks using newer die shrunk memory ics that would allow using the top end settings that were there but non usable from the hardware not being up to snuff.

But being limited to the 100Mhz CL2 what kind of benefit there would be?

I remember higher settings than 100MHz. edit: yep: http://compatlist.k6plus.com/ Soltek SL54U5, 150 fsb.

Nice 150Mhz! PC133 memories would certainly be useful. Any mod to use some DDR on SDR mainboard? 😁

386SX wrote:

Nice 150Mhz! PC133 memories would certainly be useful. Any mod to use some DDR on SDR mainboard? 😁

I don't believe these SS7 boards have a 1/4 PCI divider, meaning the PCI bus will be running extremely high at 50mhz, which will cause trouble.

F2bnp wrote:

I've tried a couple of K6-III+ 400ATZ and both max at 1.9V 550MHz. My OCD is strong though, I really need those 50MHz to round u […]
Show full quote

Tetrium wrote:

Just wondering, which chips have failed to reach 600MHz for you and how many have you tried so far?
Tbh, I wouldn't have been bothered really that much if 550MHz does run stable 😊

I've tried a couple of K6-III+ 400ATZ and both max at 1.9V 550MHz. My OCD is strong though, I really need those 50MHz to round up that number and achieve the insane performance of... a Pentium II 350 🤣 .

Skyscraper wrote:

575 MHz is ~600 MHz in my book! 😉

And you'd be totally right, but I'd really love to hit 600 with one chip. I love these little buggers!

I have a 400Mhz K6-3+ ATZ and it only does 600MHz stable on my Aopen AX59Pro @ 2.1v. It will do 600 stable on my lucky tech P4MVP3-X but only at 2.4v. I tried it on other boards - two MVP3 based AT boads -a soltek and a jetway, then on my Asus P5A and a gigabyte GA-G5AA - neither will do more then 550MHz regardless of voltage.

What I'm trying to say is that the mainboard you use might be more important then the chip.

I actually got the thing running at 560MHz at one point (fsb 112) and 533 (fsb 133) both stable on my Aopen AX59Pro. Both settings are faster then 6x100, but none are practical since the MVP5 chipset will not divide FSB/4 for PCI, just like the intel 440bx...

A K6-3+ at 2.4v sounds quiet a lot.. 😁 And I was afraid to run my K6-2+ at 2.1v, considering they are not so easy to find in a ocean of original K6-2s.

386SX wrote:

A K6-3+ at 2.4v sounds quiet a lot.. 😁 And I was afraid to run my K6-2+ at 2.1v, considering they are not so easy to find in a ocean of original K6-2s.

Have you delided it?

kanecvr wrote:

will not divide FSB/4 for PCI, just like the intel 440bx...

Just to correct you there, that is true for most 440BX boards, but the later ones, like ASUS P3B-F have such a divider and can run at 133mhz without much trouble. AGP bus is another subject though..

It still have the original metal plate cause I can imagine the core maybe it's not designed to directly support the stress of the heatsink on it. I had to use +0,1 volt cause my mainboard seems to not boot it at the default 2.0v. At default 500Mhz it stay enough cool on the usual (big) SS7 heatsink. Definetely cooler than the K6-3 400 and the K6-2 500.

kanecvr wrote:

I have a 400Mhz K6-3+ ATZ and it only does 600MHz stable on my Aopen AX59Pro @ 2.1v. It will do 600 stable on my lucky tech P4MVP3-X but only at 2.4v. I tried it on other boards - two MVP3 based AT boads -a soltek and a jetway, then on my Asus P5A and a gigabyte GA-G5AA - neither will do more then 550MHz regardless of voltage.

What I'm trying to say is that the mainboard you use might be more important then the chip.

I actually got the thing running at 560MHz at one point (fsb 112) and 533 (fsb 133) both stable on my Aopen AX59Pro. Both settings are faster then 6x100, but none are practical since the MVP5 chipset will not divide FSB/4 for PCI, just like the intel 440bx...

I know what you're trying to say. I used to have a Gigabyte GA-5AX and now I own a P5A-B. Neither of them could run my CPUs at 600MHz 🙁.

386SX wrote:

It still have the original metal plate cause I can imagine the core maybe it's not designed to directly support the stress of the heatsink on it. I had to use +0,1 volt cause my mainboard seems to not boot it at the default 2.0v. At default 500Mhz it stay enough cool on the usual (big) SS7 heatsink. Definetely cooler than the K6-3 400 and the K6-2 500.

deliding doesn't mean you can't put something back on, but you replace the tim and the lid with better, like copper.

awgamer wrote:

386SX wrote:

It still have the original metal plate cause I can imagine the core maybe it's not designed to directly support the stress of the heatsink on it. I had to use +0,1 volt cause my mainboard seems to not boot it at the default 2.0v. At default 500Mhz it stay enough cool on the usual (big) SS7 heatsink. Definetely cooler than the K6-3 400 and the K6-2 500.

deliding doesn't mean you can't put something back on, but you replace the tim and the lid with better, like copper.

But the problem could be to find a perfect height-correct replacement of the metal plate cause any differences (even minimal I'd think) could end up breaking up the core or simply not dissipate the heat; I mean sure anything is possible but the benefit seems really not much for less than 30W of TDP. 😉

Just to note for anyone who might now go delid their K6-X's, don't forget to write the part number on the Ceramic somewhere or you might forget which chip was which 😜
And yes, keep the original metal heatspreader, it's not like it takes up lots of room anyway 😜