Los Pebos wrote on 2024-03-13, 07:29:

What about the Athlon 1400 Thunderbird, with a 100 MHz FSB (serial A1400AMS3B, not C which is the 133 MHz FSB one)? I’m having a headache finding one and it was an old dream as I wanted to update my motherboard at its best when I was younger (an old Abit KT7, not KT7A).
I know it’s a mass product bit only a few found their way through their client, I believe, as the 133-MHz chipsets and the high-clock Duron arrived at the same time.

I have that exact CPU, bought it off Evilbay in April last year for 30 USD, bought it for a build that I haven't gotten around to doing yet.

Trashbytes wrote on 2024-03-13, 14:12:

Los Pebos wrote on 2024-03-13, 07:29:

What about the Athlon 1400 Thunderbird, with a 100 MHz FSB (serial A1400AMS3B, not C which is the 133 MHz FSB one)? I’m having a headache finding one and it was an old dream as I wanted to update my motherboard at its best when I was younger (an old Abit KT7, not KT7A).
I know it’s a mass product bit only a few found their way through their client, I believe, as the 133-MHz chipsets and the high-clock Duron arrived at the same time.

I have that exact CPU, bought it off Evilbay in April last year for 30 USD, bought it for a build that I haven't gotten around to doing yet.

A thing that may have made them less popular in the day also was that the earlier motherboards, the ones that couldn't run 133 officially, often didn't implement support for multipliers over 12.5x correctly. So they were not an upgrade for every 100Mhz board. Some people were wanting them for the early DDR200 boards though. However the core design and process were bouncing off physical limits at 1500, so pushing them any was not on the cards. I think many might have gone for a 1.1 Tbred and eased their FSB up to 124. ... the Duron Morgan followed in months.... but also had the same problem with earlier boards, if BIOS was updated to support new core, still might not support 13X multi of the 1300 "out of the box". However, due to being based off Palomino and the cache-stration being not as severe on Durons as Celerons, the performance of a Morgan might be about 5% ahead of Tbred at same clock, where full pally was 10%.

But also a rarer one might be the Morgan core Duron at 1400, they never seemed to be for sale very widely, DHD1400AMT1B and you lose sight of them amongst the later 1400 Applebred/Appalbred core. (Appaloosa derived from Thorougbred)

NostalgicAslinger wrote on 2024-03-13, 12:54:

Thanks for the info! So the yield quality was not so bad for the P5 core. Maybe the very eralier ones with the FDIV Bug need mor […]
Show full quote

amadeus777999 wrote on 2024-03-12, 09:39:

All P60 cpus that I have tested worked at 66mhz without a voltage raise(no matter if the VRM section was installed or not).

Thanks for the info! So the yield quality was not so bad for the P5 core.
Maybe the very eralier ones with the FDIV Bug need more voltage with 66MHz, because it was reported again and again how bad the yield rate was with the first Pentium...
It is actually nonsensical that 5.15V is then required for 66MHz clock, because the waste heat of the P5 is not that low either.

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be due to them being a later SX948 variant but one had the fdiv bug so it must have been earlier. (SX948s are, by now, nearly all that's left of the early Pentiums)

I had one cpu at 75 rock solid and flimsily at 80mhz - this was done via a board that supported clock crystals(unfortunately the board in question had very conservative stock timings).
At 75mhz the original Pentium did pretty well in build engine games.

Hello, I’m digging this topic out: do you know how rare is an Athlon Thunderbird 600 MHz Socket A/462 (FSB 100/200 MHz)? It is referenced on Wikipedia and CPU-World but I can’t find any available. Once again, I’m looking for the Socket A version, not the Slot A.
Was it only an engineering sample?

amadeus777999 wrote on 2024-03-14, 13:34:

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be […]
Show full quote

NostalgicAslinger wrote on 2024-03-13, 12:54:

Thanks for the info! So the yield quality was not so bad for the P5 core. Maybe the very eralier ones with the FDIV Bug need mor […]
Show full quote

amadeus777999 wrote on 2024-03-12, 09:39:

All P60 cpus that I have tested worked at 66mhz without a voltage raise(no matter if the VRM section was installed or not).

Thanks for the info! So the yield quality was not so bad for the P5 core.
Maybe the very eralier ones with the FDIV Bug need more voltage with 66MHz, because it was reported again and again how bad the yield rate was with the first Pentium...
It is actually nonsensical that 5.15V is then required for 66MHz clock, because the waste heat of the P5 is not that low either.

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be due to them being a later SX948 variant but one had the fdiv bug so it must have been earlier. (SX948s are, by now, nearly all that's left of the early Pentiums)

I had one cpu at 75 rock solid and flimsily at 80mhz - this was done via a board that supported clock crystals(unfortunately the board in question had very conservative stock timings).
At 75mhz the original Pentium did pretty well in build engine games.

None of my socket 4 boards supported 75mhz but all had a strange 80mhz FSB

Needless to say none of my chips would run at 80mhz.

I’ve always believed Intels rapid drop of socket 4 was a kick in the pants for early adopters, they should have at least released one other vintage speed grade for s4 when s5 released

There are many CPUs that could be classed as rare, some of which Im looking for so I have no intention of mentioning them here as it would make them sought after which would drive the cost up

It probably would have been hard to squeeze out a higher speed at 5V. As none of these boards supported 3.3v, the only possibility was what they did with the overdrives.

rmay635703 wrote on 2025-08-02, 23:09:

None of my socket 4 boards supported 75mhz but all had a strange 80mhz FSB […]
Show full quote

amadeus777999 wrote on 2024-03-14, 13:34:

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be […]
Show full quote

NostalgicAslinger wrote on 2024-03-13, 12:54:

Thanks for the info! So the yield quality was not so bad for the P5 core.
Maybe the very eralier ones with the FDIV Bug need more voltage with 66MHz, because it was reported again and again how bad the yield rate was with the first Pentium...
It is actually nonsensical that 5.15V is then required for 66MHz clock, because the waste heat of the P5 is not that low either.

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be due to them being a later SX948 variant but one had the fdiv bug so it must have been earlier. (SX948s are, by now, nearly all that's left of the early Pentiums)

I had one cpu at 75 rock solid and flimsily at 80mhz - this was done via a board that supported clock crystals(unfortunately the board in question had very conservative stock timings).
At 75mhz the original Pentium did pretty well in build engine games.

None of my socket 4 boards supported 75mhz but all had a strange 80mhz FSB

Needless to say none of my chips would run at 80mhz.

I’ve always believed Intels rapid drop of socket 4 was a kick in the pants for early adopters, they should have at least released one other vintage speed grade for s4 when s5 released

Anonymous Coward wrote on 2025-08-03, 03:38:

It probably would have been hard to squeeze out a higher speed at 5V. As none of these boards supported 3.3v, the only possibility was what they did with the overdrives.

rmay635703 wrote on 2025-08-02, 23:09:

None of my socket 4 boards supported 75mhz but all had a strange 80mhz FSB […]
Show full quote

amadeus777999 wrote on 2024-03-14, 13:34:

All nine 60mhz cpus I tested over the years ran flawlessly at 66mhz no matter the board(generic Intel or custom). This could be due to them being a later SX948 variant but one had the fdiv bug so it must have been earlier. (SX948s are, by now, nearly all that's left of the early Pentiums)

I had one cpu at 75 rock solid and flimsily at 80mhz - this was done via a board that supported clock crystals(unfortunately the board in question had very conservative stock timings).
At 75mhz the original Pentium did pretty well in build engine games.

None of my socket 4 boards supported 75mhz but all had a strange 80mhz FSB

Needless to say none of my chips would run at 80mhz.

I’ve always believed Intels rapid drop of socket 4 was a kick in the pants for early adopters, they should have at least released one other vintage speed grade for s4 when s5 released

Even in 1992 The process lithography could have easily turned out chipsets that operated at a hybrid 4v and 5 volt tolerant result, there is good evidence the existing 5 volt Intel chipsets already could run undervolted to 4 volts stock without issue .

There were 4 volt rated cpus (non-Intel) so intel even being slightly forward looking could have supported a handful of speed grades up to about 83mhz which is the 5 volt speed limit (let alone a theoretical 4 volt core design)