Socket 3 or 4 to socket 5 or 7 adapter. \ VOGONS

Socket 3 or 4 to socket 5 or 7 adapter.

Topic actions

First post, by Sphere478

Posted on 2022-03-19, 02:21

Sphere478 Offline

Rank l33t++

Rank: l33t++
Posts: 6000
Joined: 2021-01-13, 04:45

It ocurs to me that there are a few boards out there that seem to demonstrate that at the very least a socket 4 to socket 5 adapter is possible.

We have a lot of socket 3 enthusiasts around there and it occurred to me it might be kinda fun to run some of these chips on later platforms maybe even set some new records with the help of this. I believe many if not all pf these later model platforms still have cpu support for earlier cpus even if they didn’t physically fit.

Conversely, some old socket 3 systems may be able to, (with a bios update) Support later model chips from socket 5/7 as a upgrade path for specialized motherboards that are hard to upgrade.

Obviously, voltage will be an issue here but with a interposer it isn’t hard to add a regulator.

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 1 of 6, by Anonymous Coward

Posted on 2022-03-19, 03:37

Anonymous Coward Offline

Rank l33t++

Rank: l33t++
Posts: 5047
Joined: 2008-03-20, 05:37
Location: Shandong, China

POD5V83 exists, so obviously an adapter is possible, but it certainly won't be passive.
Aside from the voltage regulation and glue logic, you'll to need to design something similar to the Evergreen and Powerleap adapters that feeds the module extra current from the power supply. You will need to supply some external clock multipliers, as the target systems will only be 33 or 40MHz and at best a Pentium can only do 3.5x.

"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 2 of 6, by rmay635703

Posted on 2022-03-19, 04:01

rmay635703 Offline

Rank Oldbie

Rank: Oldbie
Posts: 1974
Joined: 2019-01-19, 19:32

The 386DX had a rarely mentioned 16bit mode to interface to 16 bit motherboards

I do not believe the Pentium had a hidden 32bit bus mode.

Lots of glue logic would be needed to handle the bus and pipeline differences

Reply 3 of 6, by BitWrangler

Posted on 2022-03-19, 04:07

BitWrangler Offline

Rank l33t++

Rank: l33t++
Posts: 8540
Joined: 2017-10-11, 00:55
Location: Ontario

I favor a Heath-Robinson/Rube-Goldberg affair by perching it on extra long wirewrap pins, with glue logic and discretes deadbugged between them along with a D cell NiMh for extra current perched on top, (Which recharges every time you can sneak in a HALT cycle) looking something like a classic era Byte cover parody of Dali's Temptation of St Anthony.

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 4 of 6, by Anonymous Coward

Posted on 2022-03-19, 05:52

Anonymous Coward Offline

Rank l33t++

Rank: l33t++
Posts: 5047
Joined: 2008-03-20, 05:37
Location: Shandong, China

There were lots of crappy motherboards in 1993 that ran the Pentium 60/66 with 32-bit memory and cache, but I suppose the glue must have been in the chipset. I even saw that same crappy OPTi 571/572 chipset used on a socket5 board.

"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