Martin85 wrote on 2025-10-27, 11:23:Also, right now it has a Radeon 9200 with 128MB. What would be a good period correct graphics card? I have many, from banshees, to voodoo1 and 2, Matrox,...
I mean I have a Radeon 9100, so yours is even better... 😁
Martin85 wrote on 2025-10-27, 19:36:BTW: the idea (although it may sound crazy) was to have the creative drive II with the CT4860 (so PCI) and maybe also some roland midi (if I manage to get my hands on it). As what concerns the OS, was thinking about windows ME. 😁
I am aware that this combination might bring some headaches, but want to relive the experience I had when I was 16. 😝
I mean if you want Windows ME that's fine, but try to find appropriate drivers that are specifically for Windows ME and not Windows 98 (unless you want the authentic blue screen experience 😁 ). Look at some videos from Phil's Computer Lab if you need more info on setting up.
Ran into another quirk with my KT7A-RAID.
Under DOS 6.22, I was getting quite noticeable and distracting crackling noises from the MK1869 XTREME (ESS ES1869 and Gravis UltraSound PnP) sound card I am using. It was especially noticeable when playing Duke3D.
The FAQ website by Paul that I linked last page was quite useful for resolving this issue.
It suggested disabling Delay Transaction and PCI master Read Caching in the BIOS. This is done by default with newer BIOS revisions. This reduced the crackling by a little but it didn't fix the issue.
The site also suggested reducing the I/O voltage or overclocking the FSB slightly. In my case, reducing the I/O voltage made it much worse but increasing the I/O voltage by one tick (0.10v) pretty much fixed it. It is night and day compared to before. I MAY hear a random crackle sometimes (I'm not sure if I'm imagining it) but that's very rare, whereas before it was crackling multiple times every few seconds.
I also tried increasing the FSB but that seemed to bring back the crackling even with the above fix. So you may need to adjust both the FSB and I/O voltage to sort it out.
So to summarize, with my Barton-M 2500+, I am using 9x multiplier (With the 5-4 PIN mod, so it's actually set to 17x multiplier), 133 FSB, CPU voltage set to 1.65v (May be able to lower it to somewhere between 1.6v-1.65v, but 1.65v is 100% stable so haven't bothered yet), 3.50v I/O voltage and the rest of the BIOS settings are set to what is recommended on Paul's website.
Shiftyy wrote on 2026-01-08, 18:26:So to summarize, with my Barton-M 2500+, I am using 9x multiplier (With the 5-4 PIN mod, so it's actually set to 17x multiplier), 133 FSB, CPU voltage set to 1.65v (May be able to lower it to somewhere between 1.6v-1.65v, but 1.65v is 100% stable so haven't bothered yet), 3.50v I/O voltage and the rest of the BIOS settings are set to what is recommended on Paul's website.
I just about wrapped up my years-long KT7A build last night. I also have an MK1869 with the PCMIDI MPU-401 add-on (though not the extreme version). So I thought "Hey, why not add a NIC -- the finest quality of course", so I went with an Intel 82559 PCI card. I immediately started having issues which I thought I solved by specifying a different IRQ for it in the BIOS (it seemed to demand IRQ 11), but then began the application crashing in Windows. I said to myself "heck no, I can live without a NIC in this thing". I can try something else, but honestly, I have front USB ports (from the motherboard header) so I will use those for any serious data transfer like I always do (I do find the VIA USB to be quite good, thought it is only USB 1.x).
Lesson learned -- rumours about VT8363A (and other VIA) bad PCI compatibility are true, and if you don't need it, don't use it. VT82C686B PCI-ISA bridge is excellent OTOH, and I have no issues at all with my MK1869 (albeit I only tested in Windows, and I have no crackling at all). I am running a Thunderbird 1.4Ghz (at 1.3Ghz). I use Thunderbird because I don't want SSE which I think causes issues with nVidia on this platform. Yes, I sacrifice multiplier throttling, but I don't need that. This is strictly for DirectX 8 gaming on Windows and maybe some DOS on the side.
(Decommissioned:)
mockingbird wrote on 2026-01-12, 14:47:Shiftyy wrote on 2026-01-08, 18:26:So to summarize, with my Barton-M 2500+, I am using 9x multiplier (With the 5-4 PIN mod, so it's actually set to 17x multiplier), 133 FSB, CPU voltage set to 1.65v (May be able to lower it to somewhere between 1.6v-1.65v, but 1.65v is 100% stable so haven't bothered yet), 3.50v I/O voltage and the rest of the BIOS settings are set to what is recommended on Paul's website.
I just about wrapped up my years-long KT7A build last night. I also have an MK1869 with the PCMIDI MPU-401 add-on (though not the extreme version). So I thought "Hey, why not add a NIC -- the finest quality of course", so I went with an Intel 82559 PCI card. I immediately started having issues which I thought I solved by specifying a different IRQ for it in the BIOS (it seemed to demand IRQ 11), but then began the application crashing in Windows. I said to myself "heck no, I can live without a NIC in this thing". I can try something else, but honestly, I have front USB ports (from the motherboard header) so I will use those for any serious data transfer like I always do (I do find the VIA USB to be quite good, thought it is only USB 1.x).
Lesson learned -- rumours about VT8363A (and other VIA) bad PCI compatibility are true, and if you don't need it, don't use it. VT82C686B PCI-ISA bridge is excellent OTOH, and I have no issues at all with my MK1869 (albeit I only tested in Windows, and I have no crackling at all). I am running a Thunderbird 1.4Ghz (at 1.3Ghz). I use Thunderbird because I don't want SSE which I think causes issues with nVidia on this platform. Yes, I sacrifice multiplier throttling, but I don't need that. This is strictly for DirectX 8 gaming on Windows and maybe some DOS on the side.
I have an Intel 82558B NIC in mine and it works without any problems.
Here are the PnP settings I use. This will force the MK1869 to use the correct IRQ/DMA settings:
IRQ-9 > Legacy ISA
IRQ-11 > Legacy ISA
IRQ-15 > Legacy ISA
DMA-6 > Legacy ISA
Also disable anything you don't need such as printer or com ports.
Please note that the above settings are for the XTREME version of the MK1869, but I believe it should work the same for the regular variant.
Also make sure that you enable "Force Update ESCD" after making changes to your PnP settings and install the Via 4in1 chipset drivers and the PCI latency patch by George.
I haven't tested the crackling issues in Windows 9x/XP as I'm using the Audigy 2 ZS for EAX support. The MK1869 XTREME is dedicated to DOS 6.22.
Shiftyy wrote on 2026-01-13, 00:10:Also make sure that you enable "Force Update ESCD" after making changes to your PnP settings and install the Via 4in1 chipset drivers and the PCI latency patch by George.
I haven't tested the crackling issues in Windows 9x/XP as I'm using the Audigy 2 ZS for EAX support. The MK1869 XTREME is dedicated to DOS 6.22.
Thanks, much appreciated, but I don't want to overload this poor VIA platform with too much complexity. Took enough time to get it stable as it is. I will test with a 3Com NIC just to satisfy your curiosity though 😉.
(Decommissioned:)
Fair enough, but I wouldn't declare this board has PCI problems until you have tried all of the above.
Shiftyy wrote on 2026-01-13, 00:50:Fair enough, but I wouldn't declare this board has PCI problems until you have tried all of the above.
So my report with the 3Com (3C905CX-TX-M) is that it works perfectly fine for me on the KT7A, as opposed to the Intel 82559. I will leave it in. Thanks.
(Decommissioned:)
With 686B, I would be most concerned about silent disk corruption. Maybe using a PCI disk controller would be a workaround. I would probably try to come up with a way to stress and verify the storage subsystem under load (with a game runnning.)
swaaye wrote on 2026-01-16, 19:27:With 686B, I would be most concerned about silent disk corruption. Maybe using a PCI disk controller would be a workaround. I would probably try to come up with a way to stress and verify the storage subsystem under load (with a game runnning.)
I have been using an Abit KT7A v1.3 for more than 20 years and I have never experienced disk corruption. I have DOS 6.22 and Win98 SE on the 1st 2 GB FAT16 partition as well as Win XP (SP 0) on the 2nd ~75 GB NTFS partition.
The system also has an SB Live 5.1 but I have the PCI Latency Patch by George E. Breese installed. If I remember correctly such disk corruption could only be reproduced by copying big chunk of data between the 2 IDE channels.
But I also have another disk on the 2nd IDE channel with Tiny Win Vista installed and I copied data between the 2 channels many times without problems.
Well maybe you should explore the possibility in depth. It would be interesting. Although so many things can cause disk corruption with Windows 9x. Maybe running an NT OS even prevents the problem.
That's quite the run with a KT7A. I still have the Abit BF6 that I bought in 1999 but it gets used only maybe a few hours a year. I recapped it around 15 years ago! Among the Abit 440BX boards was also the BE6-II, which is the BF6 but with an onboard Highpoint HPT366. A friend had one of those boards. That Highpoint chip was definitely best avoided.
swaaye wrote on 2026-01-16, 19:27:With 686B, I would be most concerned about silent disk corruption. Maybe using a PCI disk controller would be a workaround. I would probably try to come up with a way to stress and verify the storage subsystem under load (with a game runnning.)
I have been using the HighPoint HPT370 controller on my KT7A-RAID for my data drives. I assume this lets me avoid the 686B issues without needing to waste a PCI slot on a disk controller?
Shiftyy wrote on 2026-01-17, 04:11:I have been using the HighPoint HPT370 controller on my KT7A-RAID for my data drives. I assume this lets me avoid the 686B issues without needing to waste a PCI slot on a disk controller?
If the HPT370 works well for you then sure. I think the 370 was fairly solid.
swaaye wrote on 2026-01-17, 04:31:Shiftyy wrote on 2026-01-17, 04:11:I have been using the HighPoint HPT370 controller on my KT7A-RAID for my data drives. I assume this lets me avoid the 686B issues without needing to waste a PCI slot on a disk controller?
If the HPT370 works well for you then sure. I think the 370 was fairly solid.
My only complaint would be it's running a bit slow. Speed tests showed it maxing out at around 60-70MB sec. Being UDMA100 it should be closer to 100MB sec. Though the issue could be the StarTech IDE to SATA adapter I'm using.
The slower speeds doesn't matter in the end.
Shiftyy wrote on 2026-01-17, 05:54:swaaye wrote on 2026-01-17, 04:31:Shiftyy wrote on 2026-01-17, 04:11:I have been using the HighPoint HPT370 controller on my KT7A-RAID for my data drives. I assume this lets me avoid the 686B issues without needing to waste a PCI slot on a disk controller?
If the HPT370 works well for you then sure. I think the 370 was fairly solid.
My only complaint would be it's running a bit slow. Speed tests showed it maxing out at around 60-70MB sec. Being UDMA100 it should be closer to 100MB sec. Though the issue could be the StarTech IDE to SATA adapter I'm using.
The slower speeds doesn't matter in the end.
The same is true for the integrated UDMA100 IDE controller. it is maxing around ~70 MB/sec. The limiting factor is definitely not the HDD since I tried a 256 GB Samsung 860 EVO with the help of a SATA2IDE adapter
(https://www.aliexpress.com/item/1005006917394 … .61761802eDjOA6)
And it produced the same result.
I think the max you can practically move over PCI is 90-100MB/s so that's not so bad. Even running a SSD on UDMA 33 is a great experience on these old systems. The SSD benefit is mostly the ultra fast random read/write performance and access time compared to a HDD.
aren't the via 4-in-1 drivers supposed to nerf things slightly so the 686b ide corruption becomes a non-issue?
the hpt-37x has always been a solid performer for me running stripes on 3 separate mfr boards under all flavors of windows. linux compatibility is not so well served, but there are fixes for its quirks if you dig for them.
Shiftyy wrote on 2025-12-16, 14:05:I have a KT7A-RAID v1.0. It works fine with a Duron 1000. However, when I tried a Barton XP-M 2500+, it would refuse to post. […]
I have a KT7A-RAID v1.0. It works fine with a Duron 1000. However, when I tried a Barton XP-M 2500+, it would refuse to post.
The only activity would be the fans spinning and nothing else. No beep codes, even when no GPU or RAM were inserted.
I tried multiple BIOS revisions, a full mobo recap, different power supplies, PIN mod. Nothing worked.
Then I found out that a bloke named Paul created a "FAQ" website for this board. This was done back in 2001-2003 and no longer exists so it wasn't easy to find. However, archive.org to the rescue!
https://web.archive.org/web/20031203090528/ht … wfaqs.cfm?fid=2
I recommend that everyone with a KT7/KT7A/KT7A-RAID motherboard should take the time to read the site and its many pages. It contains a lot of good information and you are bound to learn something new.
Anyway, I found this little snippet under the "Processor" section:
So after giving up and putting the KT7A back into its box, I decided "One more try", pulled it back out and hooked it up.
The reset worked. For the first time I managed to get the motherboard to post with the Barton XP-M 2500+. Step 1. Power-on the computer. Step 2. Wait a few seconds Step 3. Hit the reset button.
Now I need to come up with a way to automate this, because the case I had planned to put it in, does not have a reset button! Lol.
Anyway, so far in this thread, no-one has mentioned Paul's FAQ website, nor the above fix, so I thought it was worth posting.
Amazing find! I also found out independently about the reset after a cold boot, didn't think it worth mentioning (in retrospect I should have). Glad you made it work 😀
