Your intended upgrades are all relatively small bumps in performance.

Your existing CPU may be a better overclocker if your motherboard supports extensive FSB /voltage manipulation than your target CPUs and is worth exploring first.

My personal experience during the socket A era was that many motherboards would become unstable when using faster chips than listed and that PSU quality was much more important than before.

You could get 2500+ Barton and try setting FSB to 400mhz. That will make it 3200+.

rmay635703 wrote on 2022-11-28, 20:39:

Your intended upgrades are all relatively small bumps in performance.

Your existing CPU may be a better overclocker if your motherboard supports extensive FSB /voltage manipulation than your target CPUs and is worth exploring first.

My personal experience during the socket A era was that many motherboards would become unstable when using faster chips than listed and that PSU quality was much more important than before.

That's good to know, thank you. I checked the BIOS and it allows adjusting the FSB from 166 to 199. No options for voltage. If the system could run at it, I could get 12 x 199 for 2388 MHz on the current CPU. Even though it's a small boost I do think it would be cool to have the extra L2 cache. I want to keep it somewhat stock so it can be a stable platform for testing AGP cards, as it has a full 8X slot. One of Biostar's manuals lists the Athlon XP as supported and tested, so I should be good there.

I'm hopeful the PSU I have will be alright. I picked it up from my local recycler who tested it and surprisingly even labeled it as "good for Socket A". I have a simple PSU tester but nothing that can test under load except the PC + stress tests. I have newer and more powerful ATX PSUs around in case I need to test them though.

I found some really great info about Thorton vs Barton. In short, the 512K L2 cache and 400 MT/s upgrades helped but didn't make a big splash, and Barton still lagged behind contemporary P4s which could reach 800 MT/s (quad pumped vs AMD's dual pumped FSB).

Fifth-generation Athlon Barton-core processors released in early 2003 featured PR ratings of 2500+, 2600+, 2800+, 3000+, and 320 […]
Show full quote

Fifth-generation Athlon Barton-core processors released in early 2003 featured PR ratings of 2500+, 2600+, 2800+, 3000+, and 3200+. While not operating at higher clock rates than Thoroughbred-core processors, they earned their higher PR-rating by featuring a total of 512 KiB L2 cache and, in some models, a faster 400 MT/s front side bus.[18] The Thorton core was a variant of the Barton with half of the L2 cache disabled and thus functionally identical to the Thoroughbred core.

By the time of Barton's release, the "Northwood" Pentium 4 had become more than competitive with AMD's processors.[19] Unfortunately, due to the architecture of AMD's processor caches, an L2 cache increase to 512 KiB did not have nearly the same impact as it did to Intel's line. Only an increase of several percent was gained in per-clock performance.[18] The PR rating became somewhat inaccurate because some Barton models with lower clock rate weren't consistently outperforming their higher-clocked Thoroughbred predecessors with lower ratings.[19]

The other improvement, a higher 400 MT/s bus clock, helped Barton gain some more efficiency. However, it was clear by this time that Intel's quad-pumped bus was scaling well above AMD's double-pumped EV6 bus. The 800 MT/s Pentium 4 bus was well out of Athlon's reach. In order to reach the same bandwidth levels, the Athlon bus would have to be clocked at levels simply unreachable.[18]

The K7 architecture had scaled to its limit. Maintaining performance equivalence with Intel's improving processors would require a significant redesign.[18] AMD would soon la´unch Athlon 64.

By Ramya, Friday, May 9, 2008

Source: https://aboutamd.blogspot.com/2008/05/barton- … nd-thorton.html

In any case, it's an inexpensive upgrade and I love to benchmark, so in the future I will post my comparison between the parts. :)

Cosmic wrote on 2022-11-28, 21:44:

That's good to know, thank you. I checked the BIOS and it allows adjusting the FSB from 166 to 199.

This is just a linear clock generator, so if you set 199 MHz then you will still have PCI/FSB divider set to 1/5 resulting PCI running out of spec (39.8 MHz). This will probably cause stability issues. Usually keeping PCI under 36.5 - 37.5 MHz is best compromise between performance and stability.

Instead of using BIOS you should use jumpers (JCLK1 in manual) on board to change FSB to 200 MHz. This will also force PCI/FSB divider 1/6 and PCI @ 33.3 MHz.
Bear in mind that you'll get 2.4 GHz which might be too much for Thorton on stock voltage. But you can at least try if PC will boot Windows and run some stability tests.

I also have a KT600 motherboard and I overclocked an XP2500+ to 3200 speed and even a little beyond. The issue quickly becomes one of heat, and how to get rid of it; your CPU cooler looks decent (hopefully it has a copper insert?), but it will quickly become overwhelmed by the heat of an overclocked Athlon XP. Back in the day, a CPU cooler like an SLK 900 was a worthy investment since overclocking an XP2500 to 3200 speed could save a lot of money on the CPU, and with something like the SLK 900 you could even exceed 2.2 GHz. But those coolers have all gone the way of the Dodo bird, and on the rare occasion when they do show up for sale on Ebay, people want a premium for them. It's actually cheaper now to buy the highest speed Socket A CPU - the 3200+ was almost $500.00 at launch in 2003, but now an SLK 900 would likely be double the price of a 3200+ or Sempron 3300+ on Ebay (they renamed it after the Athlon 64 was released). My Thermalright AX7 is a decent cooler, but I wanted to get a little more out of the system and my most cost effective upgrade was to buy a 3300+.

Ah the good old days of the pencil trick to get more voltage.

The Semptron “should “ clock higher with less heat than an XP chip, but it likely would need more juice to hit 2400mhz (depending on how late of a chip it was)

I believe your motherboard can be overclocked to FSBs speeds inbetween 333 & 400 from within windows using CPUCool or similar

Worth noting AMD made several socket A chips that could clock to the 3200/3300+ speeds that wore different names

In addition to the misnamed semptron 3300…
For a while there were a bunch of socket A “Geodes” that were 100% identical to the XP chip but were just run at lower bus/clock speeds . These also easily overclock to 3200+

I would throw a wide net to find a few good examples that will hit your desired speed overclocked or not.

Cosmic wrote on 2022-11-28, 19:40:

[*] CPU: AMD Sempron 2800+, 2.00 GHz, 256K L2, "Thorton"

http://web.archive.org/web/20080501075721/htt … lon-e23.html#L2
pencil on L2 bridges will unlock full 512KB of cache

http://web.archive.org/web/20080503211655/htt … pron/s10-e.html

I believe that L2 unlock won't work for CPU with locked multipier (like all retail Semprons). At least I haven't had luck with my Applebread Duron 1400, which was also multiplier-locked (connection of bridge was ignored for L2 cache while it worked for "mobile mod").
But it won't hurt to try just for fun 😁.

rmay635703 wrote on 2022-11-29, 00:25:

The Semptron “should “ clock higher with less heat than an XP chip, but it likely would need more juice to hit 2400mhz (depending on how late of a chip it was)

They are similar clockers to coresponding XPs models as they are just rebranded / renamed XPs that run on FSB 166 (200 for the fastest one IIRC).
Semprons on Thorton / Barton core should do 2.2 GHz on stock vcore and everything else is just a lottery. You might get a gem that will do 2.4 GHz without touching voltage or a garbage that will require 1.85 V for 2.3 GHz. There might be some "golden" series like in case of early XP 1700+, but I doubt that anyone cared about that in Sempron era. 😁

@Cosmic: If your CPU won't be stable @ 2.4 GHz, you can try to do a "mobile mod" which will allow you to control mobile multiplier of your CPU using sofware like WCPUid. But not every KT600 supports it.
To do it you need to connect L5 [2] bridge like is written in link provided by @rasz_pl:

Mobile
When L5 [2] bridge is shut, a dynamic multiplier change function can be used by CrystalCPUID.

In that case you'll be able to set eg 11x multi and run FSB @ 200 to get 2.2 GHz.