It's great for Windows 98 and early Windows XP games in dual boot. Pair it with Athlon 3400+, NVidia GeForce 4 or FX (due to Windows 98) and Sound Blaster Audigy 2 ZS.

The socket 754 platform is really nice for building a powerful Win9x rig. Using one of the higher end Athlon64s, even CPU intensive games like Deus Ex will run smoothly.

Be wary of this issue though. To check if your system is affected, install WinXP alongside Win98 and compare the performance in 3D games and applications.

Joseph_Joestar wrote on 2022-08-02, 05:19:

Be wary of this issue though. To check if your system is affected, install WinXP alongside Win98 and compare the performance in 3D games and applications.

According to my findings, the issue was introduced (probably alongside a microcode update), sometime at the beginning of 2005.
If the BIOS version is from 2004 (even late 2004), then Win98 performance is most likely unaffected.

bloodem wrote on 2022-08-02, 06:18:

According to my findings, the issue was introduced (probably alongside a microcode update), sometime at the beginning of 2005.
If the BIOS version is from 2004 (even late 2004), then Win98 performance is most likely unaffected.

Nice find!

I guess it also depends on when a particular manufacturer decided to integrate the problematic microcode. For example, the earliest BIOS for my Asus K8V-MX is version 0112 from 2005/08/09 and that one still works fine on Win9x. Any further BIOS updates (versions 02xx) introduce the performance bug.

Joseph_Joestar wrote on 2022-08-02, 06:29:

Nice find!

I guess it also depends on when a particular manufacturer decided to integrate the problematic microcode. For example, the earliest BIOS for my Asus K8V-MX is version 0112 from 2005/08/09 and that one still works fine on Win9x. Any further BIOS updates (versions 02xx) introduce the performance bug.

Yes, indeed, the reverse is not valid (i.e. if the BIOS version is from 2005, then it doesn't necessarily mean that it's affected by this).
Also, I should mention that this only seems to be the case for motherboards with the K8T800 chipset. It doesn't seem to affect boards with the K8T800 Pro chipset.

bloodem wrote on 2022-08-02, 07:00:

Also, I should mention that this only seems to be the case for motherboards with the K8T800 chipset. It doesn't seem to affect boards with the K8T800 Pro chipset.

Interesting. It does seem to also affect the K8M800 variant, which is what my Asus K8V-MX uses.

BTW, another potential issue with socket 754 motherboards is the problematic support for SATA hard disks and/or SSDs. From what I can tell, the VT8237(R) southbridge only works properly with SATA I drives. If you try using a SATA II or a SATA III drive, you will encounter issues. This was apparently resolved in the VT8237R Plus revision.

Joseph_Joestar wrote on 2022-08-02, 08:11:

Interesting. It does seem to also affect the K8M800 variant, which is what my Asus K8V-MX uses.

Don't think I've ever had a motherboard with the K8M800 chipset. IIRC, the only real difference between K8M800 and K8T800 is the integrated graphics, so it would make sense that other than that, they probably share the same blocks of code when it comes to the chipset initialization (ergo, the same weaknesses).

In fact, don't ask me how I know 😁, but BIOS revisions that were destined for K8T800 Pro also work fine with the K8T800 (which is how I 'fixed' the Win98 AGP performance bug on some of my motherboards).

Joseph_Joestar wrote on 2022-08-02, 08:11:

BTW, another potential issue with socket 754 motherboards is the problematic support for SATA hard disks and/or SSDs. From what I can tell, the VT8237(R) southbridge only works properly with SATA I drives. If you try using a SATA II or a SATA III drive, you will encounter issues. This was apparently resolved in the VT8237R Plus revision.

Not only on s.754, but wherever the southbridge VT8237 is mounted. I have the same problem on MSI KM400AM (s.462) - I had to software-switch the Samsung SpinPoint HD250 disk to SATAI mode, otherwise the controller refused to recognize it.