ABIT KT7A-R rev1.0 + AthlonXP 1500+ ... ich teste gerade!

Hab seit gestern einen XP1700+ in meinem KT7A-Raid Rev 1.0 stecken (auch BiosVersion 64).
Ist gestern den ganzen Abend stabil auf 1650 gelaufen (max getestet auf 1740, aber da war der FSB für die Grafikkarte offenbar schon ein Problem.

Hoffe er wird nicht nach einiger Zeit instabil wie bei Goldlocke, war ja noch nicht so lange durchgängig up, immer wieder rumgetestet, Bios verstellt, usw...

Derzeit läuft aber gerade RC-5 (seit ca. 8:00 morgens) und ich werd eh sehn ob er am Abend noch rennt, abfackeln sollte er eigentlich nicht mit der Wasserkühlung, ist sogar kühler als mein TB1,2@1,4...

Why does only the KT7A(-RAID) version 1.3 support the AthlonXP processor?

There has been much heated discussion about the failure of early KT7A motherboard revisions to support the AthlonXP and Duron (Morgan core) processors. These discussions have also been confused by incorrect statements on the ABIT (China) website, suggesting that the 64 BIOS release provided support for these processors. This is not true, and the issue is more complicated. The difference between the earlier Athlon processors and the new AthlonXP processor lies in the processor boot process. You can see the relevant boot timing diagrams on page 35, figure 10 of the Athlon Model 4 Data Sheet (for the older Athlons) and page 43, figure 12 of the Athlon Model 6 Data Sheet (for the Athlon XP). During the boot process, the motherboard must sample three data lines called FID(0:3) in order to determine the CPU multiplier setting. In the new AthlonXP processor, the timing requirements for this sampling has changed, and these lines only become valid 100ns after the Vcore power OK (PWROK) signal is asserted.
This means that if the motherboard does not apply an extra circuit to delay the NB_RESET# signal after the FIDs are fully valid, the motherboard may potentially get the wrong values from the FID lines. This is the reason why the earlier motherboard versions cannot support the AthlonXP processor. Getting the wrong FIDs may not always result in the system being unable to boot up - but it may result in the motherboard using the wrong parameters to initialise the processor, causing system instabilities under certain circumstances. On the earlier Athlon processors the FIDs are always valid after the Vcore of the processor is valid and this is why no motherboard has a problem supporting the highest frequency of old Athlon.
In practice, this change is very subtle. Many AthlonXP processors will work fine on earlier motherboard revisions. Others will work fine for some percentage of the time, but will require a hard reset during boot if this timing assertion is violated. However, only revision 1.3 of the motherboard contains the necessary hardware modifications to fully support AMD's revised timing specification. This is why ABIT cannot officially support the AthlonXP processor on earlier motherboard revisions - although if you can live with the occasional hard reset there is no problem. Furthermore, ABIT have made attempts to improve the timing compatibility with the 64 BIOS release, but still cannot officially support the processor on these revisions.
No BIOS release will allow ABIT to properly support the AthlonXP on motherboard versions prior to v1.3. However, in practice, the timing difference is sufficiently subtle that many processors will work fine, and others will simply require the occasional reset during boot. Once the processor has booted, the stability should usually be the same on all motherboards.
In some respects ABIT are simply being more honest than other motherboard manufacturers. In most cases the AthlonXP will appear to work OK - or have occasional instabilities. However, rather than pretending that a simple BIOS update can resolve the problem, ABIT are only officially claiming that the board with the proper additional circuitry can support these new processors.
Since originally posting this question, a few people have emailed me suggesting that the problem might be resolvable by unlocking the processor's multiplier using the L1 bridges and then setting the multiplier manually using SoftMenuIII. I can confirm that this solution will not work as this does not resolve the problem of the FID lines being unstable. If the bypass circuit incorrectly samples the high-low status of the FID lines, the power up issue will remain and hence this solution is not workable. However, locked or unlocked, a reset should always work as the PWROK line is always valid after the initial cold start.