VOGONS


Reply 60 of 104, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
386SX wrote on 2021-12-09, 12:14:

AMD should have built an official last epic version of the K6-3+ at 600Mhz.

What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, memory interlacing, etc similar to the Ali Aladdin 7 but all the way back in late 1998. Aka an Iron Maiden k6 entry chipset to compliment the k7 high end Iron Gate

Per AMD The reason K6 stayed below 600mhz is because nobody wanted to design a new Super 7 chipset in 1998+ with higher 133mhz FSB and by the time someone did it was a dead platform.
AMD was ready to expand to 900mhz k6+ per their old commentary but board makers weren’t making new designs by that point so it didn’t matter.

Last edited by rmay635703 on 2021-12-10, 18:22. Edited 1 time in total.

Reply 62 of 104, by Repo Man11

User metadata
Rank Oldbie
Rank
Oldbie
rmay635703 wrote on 2021-12-10, 17:45:
What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, me […]
Show full quote
386SX wrote on 2021-12-09, 12:14:

AMD should have built an official last epic version of the K6-3+ at 600Mhz.

What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, memory interlacing, etc similar to the Ali Aladdin 7 but all the way back in late 1998. Aka an Iron Maiden k6 entry chipset to compliment the k7 high end Iron Gate

Per AMD The reason K6 stayed below 600mhz is because nobody wanted to design a new Super 7 chipset in 1998+ with higher 133mhz FSB and by the time someone did it was a dead platform.
AMD was ready to expand to 900mhz k6+ per their old commentary but board makers weren’t making new designs by that point so it didn’t matter.

I've always wondered what the clock speed upper limit for the Socket 7 321 pin interface was. Intel pushed Socket 370 to 1.4 GHz, and AMD started the 423 pin Socket A at 600 MHz and ended at 2.2 Ghz.

"I'd rather be rich than stupid" - Jack Handey

Reply 63 of 104, by NostalgicAslinger

User metadata
Rank Member
Rank
Member

This K6-III+ 500 also runs fine (+8h Prime test) with 550 MHz at 1,8V and 600 MHz with a VCore of 1,95V.

Attachments

Reply 64 of 104, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++
Jasin Natael wrote on 2021-12-10, 18:07:

I gave $25 for my K6-3+ 450.
I know it will do 550 at 2.1v on a decent board.
Currently running at 500 @ 2.2 on a not so decent board.

Both of my K6-III+ 450 CPUs seem to do at least 600Mhz not problem.

This is on an ASUS P5A-B. I have at least two of those boards. One of them will run them at 600Mhz+ and the other won't. Might be time for a recap.

Back in the day, I had a K6-2 550 running at 660Mhz (120x5.5) on an ASUS P5A using only a cheap air cooler but I did have a nice large fan pointed at it because it ran super hot at that speed.

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 65 of 104, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++
rmay635703 wrote on 2021-12-10, 17:45:
What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, me […]
Show full quote
386SX wrote on 2021-12-09, 12:14:

AMD should have built an official last epic version of the K6-3+ at 600Mhz.

What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, memory interlacing, etc similar to the Ali Aladdin 7 but all the way back in late 1998. Aka an Iron Maiden k6 entry chipset to compliment the k7 high end Iron Gate

Per AMD The reason K6 stayed below 600mhz is because nobody wanted to design a new Super 7 chipset in 1998+ with higher 133mhz FSB and by the time someone did it was a dead platform.
AMD was ready to expand to 900mhz k6+ per their old commentary but board makers weren’t making new designs by that point so it didn’t matter.

I would imagine that another reason to drop and the one often cited was excellent yields on Duron. ... if Athlon and Duron had ballsed up and they had to have 200W chips to get 1500 Mhz, I bet they'd have done the kind of "tualatinised" K6-3 with 512 or even 1MB of L2 and pushed it over a ghz... then they'd have been trying hard to patch the consumption and speed issues of the Athlon line, and possibly developing an enhanced K6-4M to go head to head with PentiumM . But Duron could sell as cheap as K6-2/3 and kick snot out of celeron, so it was pointless keeping K6 on.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 66 of 104, by Sphere478

User metadata
Rank l33t++
Rank
l33t++

There is no chance of a 462 to 321 adapter is there? Different bus right?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 67 of 104, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
Repo Man11 wrote on 2021-12-10, 18:38:
rmay635703 wrote on 2021-12-10, 17:45:
What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, me […]
Show full quote
386SX wrote on 2021-12-09, 12:14:

AMD should have built an official last epic version of the K6-3+ at 600Mhz.

What they should have done is made their own high end super 7 chipset with official 133mhz FSB , a better agp implementation, memory interlacing, etc similar to the Ali Aladdin 7 but all the way back in late 1998. Aka an Iron Maiden k6 entry chipset to compliment the k7 high end Iron Gate

Per AMD The reason K6 stayed below 600mhz is because nobody wanted to design a new Super 7 chipset in 1998+ with higher 133mhz FSB and by the time someone did it was a dead platform.
AMD was ready to expand to 900mhz k6+ per their old commentary but board makers weren’t making new designs by that point so it didn’t matter.

I've always wondered what the clock speed upper limit for the Socket 7 321 pin interface was. Intel pushed Socket 370 to 1.4 GHz, and AMD started the 423 pin Socket A at 600 MHz and ended at 2.2 Ghz.

If upper limit is overclock ing and we trust the guys who use liquid nitrogen and freeze the whole motherboard quotes of 800-900mhz were made back in the day, the k6+ chips were mostly motherboard limited.

Sphere478 wrote on 2021-12-10, 20:27:

There is no chance of a 462 to 321 adapter is there? Different bus right?

Completely different
DEC made DDR bus with long bursts possible versus the old 64 bit Pentium interface

Would need lots of logic to interface
similar to the old transputer upgrades

Reply 68 of 104, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

I found an 840Mhz LN2 run on HWbot the other day. That's the max 140Mhz of the GA-5AX on 6x multiplier.

edit: there it is https://hwbot.org/submission/3585743_gumanoid … _550mhz_840_mhz

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 69 of 104, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
BitWrangler wrote on 2021-12-10, 21:43:

I found an 840Mhz LN2 run on HWbot the other day. That's the max 140Mhz of the GA-5AX on 6x multiplier.

edit: there it is https://hwbot.org/submission/3585743_gumanoid … _550mhz_840_mhz

Holy crap, I didn’t realize someone finally did 140mhz x6

😳

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 70 of 104, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++
Sphere478 wrote on 2021-12-10, 22:00:
BitWrangler wrote on 2021-12-10, 21:43:

I found an 840Mhz LN2 run on HWbot the other day. That's the max 140Mhz of the GA-5AX on 6x multiplier.

edit: there it is https://hwbot.org/submission/3585743_gumanoid … _550mhz_840_mhz

Holy crap, I didn’t realize someone finally did 140mhz x6

😳

Truly the system worked no longer than 7-10 second on 840MHz, but I was able to do the validation!!!

I figured validation was all they were able to get.

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 71 of 104, by Jasin Natael

User metadata
Rank Oldbie
Rank
Oldbie
cyclone3d wrote on 2021-12-10, 19:17:
Both of my K6-III+ 450 CPUs seem to do at least 600Mhz not problem. […]
Show full quote
Jasin Natael wrote on 2021-12-10, 18:07:

I gave $25 for my K6-3+ 450.
I know it will do 550 at 2.1v on a decent board.
Currently running at 500 @ 2.2 on a not so decent board.

Both of my K6-III+ 450 CPUs seem to do at least 600Mhz not problem.

This is on an ASUS P5A-B. I have at least two of those boards. One of them will run them at 600Mhz+ and the other won't. Might be time for a recap.

Back in the day, I had a K6-2 550 running at 660Mhz (120x5.5) on an ASUS P5A using only a cheap air cooler but I did have a nice large fan pointed at it because it ran super hot at that speed.

Yeah, on a good board they will usually do very well.
I'm currently running mine on a bottom feeder PC Chips M571 that doesn't officially support it.
But quality socket 7 and especially ss7 boards are pricey these days....plus it was mostly about recreating my first Windows 98SE PC when I built mine, and it was also a crappy PC Chips budget build.
So it seems appropriate to be using that board. Even so performance with a Voodoo 3 PCI is satisfactory.

Reply 73 of 104, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
Delerium wrote on 2022-04-27, 09:57:

Sheesh, maybe I should sell mine.. 🤣

At this point, these are purely a collectors item given the 570 2+ chips that are cheap, plentiful an unlockable. (I was right 😁)

SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod

Last edited by Sphere478 on 2022-06-05, 15:30. Edited 1 time in total.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 74 of 104, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Maybe I can always get $500+ for a vintage PC item if I just include "HOLY GRAIL UNICORN!!!!" in the listing.

I've got a 550. Are they so collectable? Is it as rare as a K5-200?

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 75 of 104, by ChrisK

User metadata
Rank Member
Rank
Member

They are rare but not as much as the K5-PR200.
Last one I've seen here on ebay was sold for ~1200 EUR.

Attachments

  • K5PR200_eBay.png
    Filename
    K5PR200_eBay.png
    File size
    259.25 KiB
    Views
    659 views
    File license
    Public domain

Reply 76 of 104, by RaiderOfLostVoodoo

User metadata
Rank Member
Rank
Member
Sphere478 wrote on 2022-04-27, 10:07:

At this point, these are purely a collectors item given the 570 2+ chips that are cheap, plentiful an unlockable.

SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod

Even before the mod was discovered, the 550ACR was a collectors item only.
My 400ATZ (low voltage) can be overclocked to 550MHz if I increase the voltage from 1.6V to 1.7V.

Reply 77 of 104, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
RaiderOfLostVoodoo wrote on 2022-04-27, 15:47:
Sphere478 wrote on 2022-04-27, 10:07:

At this point, these are purely a collectors item given the 570 2+ chips that are cheap, plentiful an unlockable.

SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod

Even before the mod was discovered, the 550ACR was a collectors item only.
My 400ATZ (low voltage) can be overclocked to 550MHz if I increase the voltage from 1.6V to 1.7V.

good point.

speaking of fun finds, there is a tillamook 233 and a cyrix 2.2v 300gp on ebay right now
the 2.2v cyrix is the 433gp core
and the tillamook 233 may be able to overclock to 4-500mhz though I suspect they disabled the 4x multi on it 🙁

Reading on these:
Cyrix/ST/IBM Aiming for the stars!
Tillamook 266MHz and working L2 cache?

anyway, back to topic. I just got another batch of 570s gonna turn em into 3+s :p

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 78 of 104, by Repo Man11

User metadata
Rank Oldbie
Rank
Oldbie
RaiderOfLostVoodoo wrote on 2022-04-27, 15:47:
Sphere478 wrote on 2022-04-27, 10:07:

At this point, these are purely a collectors item given the 570 2+ chips that are cheap, plentiful an unlockable.

SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod

Even before the mod was discovered, the 550ACR was a collectors item only.
My 400ATZ (low voltage) can be overclocked to 550MHz if I increase the voltage from 1.6V to 1.7V.

Mine took 2.4 volts to overclock to 560 (112x5) and wasn't stable at any speed beyond that; my 2.0 volt K6-2+ 550 could overclock to 616 at 2.2 volts. For benchmarking, the K6-3+ @ 560 was slightly faster than the K6-2+ @ 616, so I always wondered what a higher rated K6-3+ could do. Modifying a K6-2+ 570 so that it has the full 256k L2 finally allowed me to find out!

"I'd rather be rich than stupid" - Jack Handey

Reply 79 of 104, by Cosmic

User metadata
Rank Member
Rank
Member

What do you all think about the TriLevel cache feature of the K6-3? Does it make a noticeable difference? I have a K6-3 450 (256K L2) and my board has 1MB cache, but I can't enable it with the K6-3 or a K6-2 500 (no L2), not even at FSB 66. Just wondering how much I might be missing out on if the cache worked...