dj_pirtu wrote on 2022-04-12, 16:51:

Look what I've got today! Two v1.2 KT7A-RAIDs. […]
Show full quote

Look what I've got today! Two v1.2 KT7A-RAIDs.

Going to test these on weekend... with Geforces. 😉

Edit: first one on testbench, geforce 6800gt gives blue screen for nv4_disp.dll like my previous KT7A when playing FlaOut2. And PS/2 keyboard port is dead. Lights blink on keyboard when turning on but that's it. Mouse-port works. Fuses are OK.

Edit2: second tested, same thing with geforce, bsod with nv4_disp. It's going to look like a software problem if all my four KT7A (KT7A v1.1 & v1.3 and KT7A-RAID v1.2 & v1.2) mobos dont like geforces...

Edit3: this fully working mobo takes FSB like crazy 😮 with default voltage 152MHz. Runs 3dmarks fluently (with Radeon X800Pro). My previous boards took maybe 145MHz and that's it.

Maybe some clues here for you... https://www.overclockers.com/forums/threads/p … rce2-gts.16526/

I gotta say this thread is a bit amusing, I had this board 20 years ago and it was crash city. It would blue screen all the darn time

Zeerex wrote on 2022-04-20, 17:11:

I gotta say this thread is a bit amusing, I had this board 20 years ago and it was crash city. It would blue screen all the darn time

Then something went wrong with your board that is not typical.
I have been using this board for 20 years without blue screens 😀
It's worth looking at the OS installation date. It's the original XP service pack 0 and it has been running without problems for almost 20 years:

Wow dude, you will need to throw that thing a birthday party soon.

For the first six months I had one of these motherboards I ran Windows 98 SE, and it seemed very unstable. When I installed XP in the spring of 2002, the installation halted and I discovered that I had to relax the memory timings on the generic inexpensive SDRAM I'd purchased from Tiger Direct to be able to install XP. No doubt that Windows 98 would have worked much better if I hadn't set the RAM as though it was high quality overclocking memory rather than the low end RAM that it actually was. And I had no idea about checking stability with Memtest or anything of that nature back then.

Repo Man11 wrote on 2022-04-20, 19:44:

For the first six months I had one of these motherboards I ran Windows 98 SE, and it seemed very unstable. When I installed XP in the spring of 2002, the installation halted and I discovered that I had to relax the memory timings on the generic inexpensive SDRAM I'd purchased from Tiger Direct to be able to install XP. No doubt that Windows 98 would have worked much better if I hadn't set the RAM as though it was high quality overclocking memory rather than the low end RAM that it actually was. And I had no idea about checking stability with Memtest or anything of that nature back then.

MEMtest is a friend, and Prime95. Couple hours of those and you can say it's quite stable. But with VIA there's always AGP-stability in concern too so gaming and 3DMark gives the final statement. 😀

The KT7A 1.3 should overclock up to 150-MHz FSB CL2 and more (never pushed it any further to be honest).

With a proper Bios even the 2600+ XP-M is getting detected and you can undervolt it
Re: My 3DMark01 Mega Thread

I can run 3DMarks @152FSB but MEMtest reboots halfway of test pass. 151FSB same thing. 150FSB fully stable. 😎

KT7A-RAID v1.2, Athlon XP-M Barton, Radeon X800Pro.

i remember having an abit kt7a back in the day. as i recall i hated it...maybe it was an early board revision--i don't remember. wish i could get my hands on one today for an athlon xp isa build. raid would have been a must for striping, but now with ssd so ubiquitous i'd disable the extra controller for a faster boot. here is my pecking order of athlon xp isa boards:

1. abit kt7a later revision
2. biostar m7mia
3. epox ep-8kta3
4. abit kt7a early revision
5. iwill kk266 plus (my current retro build)
6. iwill kk266 later revision

mockingbird wrote on 2022-04-06, 20:34:

GeForce3 Ti500 - crash to desktop GeForce4 Ti4200 8x - crash to desktop GeForce 4 Ti4600 (NV25) - won't load Windows when driver […]
Show full quote

GeForce3 Ti500 - crash to desktop
GeForce4 Ti4200 8x - crash to desktop
GeForce 4 Ti4600 (NV25) - won't load Windows when driver is installed
...
Next I will have to try some of my other Socket462 KT133A boards.

I received my KK266+ this week and re-capped it. I just ran Flatout2 on it, albeit with a Palomino 1700+ (it won't support more than that, even with the pin mod).

Flatout2 was perfectly stable. Didn't ever crash, launched perfectly on the first try and played well through the racetrack.

I'm going to now start enabling stuff in the BIOS that the KT7A had enabled, to see if I can get it to crash like on the KT7A.

i have an athlon xp palomino 1700, 1900, 2000, 2100 and they all worked fine with my iwill kk266+. bios update gets the higher ones recognized correctly, but they all posted even before that. maybe you have the earlier kk266 non-plus revision? i think they are green vs yellow.

stanwebber wrote on 2022-05-06, 04:08:

i have an athlon xp palomino 1700, 1900, 2000, 2100 and they all worked fine with my iwill kk266+. bios update gets the higher ones recognized correctly, but they all posted even before that. maybe you have the earlier kk266 non-plus revision? i think they are green vs yellow.

Sorry, I meant Palomino in general, not the 1700+ in particular.

I also read that the trade-off when it comes to excess heat with anything higher than a 1700+ isn't worthwhile.

i overclock my athlon 2100 to an easy 140mhz fsb with no vcore adjustment. i use a copper skived fin heatsink with an 80 > 92mm fan adapter and a temp controlled fan that doesn't register enough rpms to even be detected by the board. speed is speed.

bios reports my fully warm cpu temp is 35c. since that's cooler than my body temp i'm not sure how accurate that is.

stanwebber wrote on 2022-05-06, 15:57:

i overclock my athlon 2100 to an easy 140mhz fsb with no vcore adjustment. i use a copper skived fin heatsink with an 80 > 92mm fan adapter and a temp controlled fan that doesn't register enough rpms to even be detected by the board. speed is speed.

bios reports my fully warm cpu temp is 35c. since that's cooler than my body temp i'm not sure how accurate that is.

On the KK266+ --- not very acurate.

They cut the thermistor leads too short so that it doesn't touch the bottom of the CPU... Ah, but don't play around with it, if it works, it works...

Please test Flatout2 for stability... Flatout2 seems to be the KT7A's achiles heel...

Next up, I plan to test the Azza KT3-AV which I know works with Barton, so we'll find out if this fault is also present with this model as with the KT7A...

If anyone with a higher KT7A revision wants to contribute, please test Flatout2 with a Barton 2800+ or higher, though the differences between KT7A revisions are most probably insignificant.

dj_pirtu wrote on 2022-04-07, 05:13:

In FlatOut2 it's crashing in minutes or when idling in desktop.

mockingbird wrote on 2022-05-06, 03:29:

I received my KK266+ this week and re-capped it. I just ran Flatout2 on it, albeit with a Palomino 1700+ (it won't support more than that, even with the pin mod).

Flatout2 was perfectly stable. Didn't ever crash, launched perfectly on the first try and played well through the racetrack.

Finished re-capping my AZZA KT3-AV today. Installed a Barton 2800+, BIOS only allows up to 12.5x (so 12.5x133 = 1600mhz~), so I'll have to pin mod tomorrow to get 2133Mhz.

But Flatout2 runs perfectly stable.

The evidence of some design flaw with regard to power delivery to the AGP on the KT7A is growing.

After the AZZA I'll be testing a Biostar M7VKD.

Did you use the same PSU on both (KT7-AV/KT7A) ?

cde wrote on 2022-05-12, 08:19:

Did you use the same PSU on both (KT7-AV/KT7A) ?

Yes... Fully re-capped Enermax PSU... Can you please test with Flatout2? I've not seen a single report of anyone else running this game stable with the KT7A.

UPDATE:

The KT3-AV experiences the same issue as the KT7A in Flatout 2 when the 2800+ Barton is run at the correct frequency (16x133 = 2133Mhz). The game crashes to desktop during the race.

So I conclude now that it's the KT133A that is not fully able to handle very high frequencies, and this isn't a motherboard problem.

So stick with 1.6Ghz for this platform and not more.
I did more testing, this time with an AXMH2500 week 0413 on the Azza KT3-AV (yes, post 2003 week 28 mobile chips are unlocked as well unlike their desktop variants)... Everything was perfectly stable on Flatout2 at 2133Mhz.

So the problems lie with with the AXDA2800DKV4C I was using and the AXMH2400FQQ4C (Mobile Barton 2400+)... Why then did the 2800+ crash on Flatout2 at stock speed and voltages? I imagine that the voltage regulation on these boards isn't good enough for these later models (remember, these boards came out when sub-Ghz processors were mainstream), and the higher rated, better-binned 2500+ I was using is more able to withstand the voltage droop or fluctuation.

I'd reckon that this 2500+ would also be stable at 2133Mhz (2800+ speed) on the KT7A, and I'll probably test this some time in the near future to confirm.

mockingbird wrote on 2022-05-12, 18:12:

UPDATE: […]
Show full quote

UPDATE:

The KT3-AV experiences the same issue as the KT7A in Flatout 2 when the 2800+ Barton is run at the correct frequency (16x133 = 2133Mhz). The game crashes to desktop during the race.

So I conclude now that it's the KT133A that is not fully able to handle very high frequencies, and this isn't a motherboard problem.

So stick with 1.6Ghz for this platform and not more.
I did more testing, this time with an AXMH2500 week 0413 on the Azza KT3-AV (yes, post 2003 week 28 mobile chips are unlocked as well unlike their desktop variants)... Everything was perfectly stable on Flatout2 at 2133Mhz.

So the problems lie with with the AXDA2800DKV4C I was using and the AXMH2400FQQ4C (Mobile Barton 2400+)... Why then did the 2800+ crash on Flatout2 at stock speed and voltages? I imagine that the voltage regulation on these boards isn't good enough for these later models (remember, these boards came out when sub-Ghz processors were mainstream), and the higher rated, better-binned 2500+ I was using is more able to withstand the voltage droop or fluctuation.

I'd reckon that this 2500+ would also be stable at 2133Mhz (2800+ speed) on the KT7A, and I'll probably test this some time in the near future to confirm.

Would you mind running memtest and see if it throws any errors?
I'm thinking maybe there's some incompatibility problems on some SocketA ABIT boards when using later AXP's:
Re: Yet another Abit de-Teapofied (almost)

bofh.fromhell wrote on 2022-05-13, 20:25:

Would you mind running memtest and see if it throws any errors?
I'm thinking maybe there's some incompatibility problems on some SocketA ABIT boards when using later AXP's:
Re: Yet another Abit de-Teapofied (almost)

Sure, but Memtest isn't going to show anything Prime95 didn't, and Prime95 passed while Flatout2 did not.

But I think it's going to be stable when I test it with the Mobile 2500+. I'll make sure to run it at the same voltage as my KT3-AV.