MSI K7N2 Delta ILSR. The manufacturer claims support for all Athlons from 600MHz to 3200+.
This is a story that may have some useful parts. If not, I get to vent.
To tell the end: The board does not now work.
To start at the beginning: It's a refurb from Newegg, $80 shipped. By all accounts, their "refurbs" have a high no-work instance, and sometimes are missing accessories (as Newegg says.) I got all the accessories, which is like a stack of ten items with this deluxe model, some unique, so I lucked out there. The board and items looked perfect. Only the mobo was unsealed. I figured a refurb naturally had been set up for a different CPU, etc., so would likely need a CMOS reset to work, which it did. (First I just tried setting the FSB jumper to 100Mhz.) It took about twenty go-rounds of CMOS resets and reset button pushes before it would boot (in the middle of which I set the memory speed jumper to 100MHz.) Rather troublesome, but resetting the CMOS hasn't worked completely reliably or regularly for me with the other Athlon mobo I have that OCs from the BIOS, or if switching CPUs.
So the board was up and running on a box outside the case and using my extra, but known good, parts; like I do normally before committing to a new mobo. I was using my old 800MHz TBird and a Radeon 7200 PCI video card. I reinstalled XP over itself on one partition to get XP to work. I played with my new toy for about 8 hours. Ran prime95 for a couple of hours at 9.5 x 100. Blasted away in Max Payne for a couple hours. The mobo OCed to 1066 (8x133, not stable in XP), but 7x133=933 was fine with prime95. Put the memory up to 133, then async to 166 and 178. Prime95 for an hour. But memory at 200 was a no-boot. Reset the CMOS with about as many tries as before
Now here is what I did not like at all. I wanted to try the CPU at 6x 166=1000 or maybe 5.5 x 166 =913, or even 5x 200=1000. There is nothing below 7 in the BIOS. It has from 7 to 13 (and Hardware default), that's it. It doesn't go very low for voltage either. I was out of interesting options. Oh well, it is getting late. On to the "real" CPU, a 1700+ modified from an 11 multiplier to a 15 (very long story elsewhere), and which has been OCing many months to 2000MHz. To get ready for the CPU exchange, I set the BIOS to Hardware multiplier and Hardware voltage, memory ratio to 1:1. I set the jumpers to 100Mhz FSB and 100MHz memory. That way the mobo should boot perfectly for the next CPU at 15x 100 = 1500, very close to the factory speed of 1467 for 1700+. Since I had so much trouble resetting the CMOS, I made sure it booted from a complete power-off a couple of times with the 800.
But it would not boot the 1700+. After many power off/ons and button resets, I resigned myself to to resetting the CMOS, which again was ridiculously flakey. It eventually boots, but it boots at 2400+ (WHA???), which is 2000MHz not 1500. Unfortunately, without overvolting, my 1700+ is barely stable enough to boot at 2000MHz . I would get into the BIOS to set working settings and the computer would crash before I made it out of the BIOS. A few times I made it out of the BIOS, and it was still identified as 2400+ even though I set the multiplier as low as it would go (7x), which made no sense.
At 2AM nothing made any sense. I had reset the CMOS and it wouldn't boot over and over, so I called it a day. In my sleep, with my brain in neutral, it was all clear: the mobo must be setting the FSB to 133 and the multiplier to 15 (15x 133 =2000), so I would set the memory jumper to 133 and try booting again. Awake, I didn't get the sleep logic about the memory jumper, but the memory always worked at 133 IAC, so I did it, and the board immediately booted. At 1800+. Wha??? Why? It was always 2400+ before. ( Yes, it is the right jumper in the right position according to the manual. ) But I'll take it. Ran prime 95 for hours. Blasted awhile in Max Payne. Pulled stuff across the LAN. Played some MP3s. Everthing is perfect. The board boots fine from power-off.
Now for OCing. I upped the FSB to 133 and tried a few low multipliers like 9 and 10 to be safe, but none would work except 7 (which mysteriously gave me 2000MHz), and I had to do the time consuming CMOS reset that worked so sporadically. Eventually it penetrated my dense brain that the "deluxe" Delta ILSR mobo was not setting the fifth multiplier bit. (OBSOLETE PIECE OF JUNK!). Therefore my CPU was keeping that bit high, and 9 was really 17, 10 was 23 and so forth. No wonder they didn't work. I confimed that by trying it with FSB 100. Yep, 9 gives 17x 100 =1700MHz, etc. (7 gives 15x , BTW.) Now that I had overcome my misconception that the mobo was a contemporary design, things were falling into place.
The lowest multiplier this mobo will set (without a hardware mod) is 13 (would be a 5x on CPUs factory set to multpliers below 13x). Because of the way the multipliers translate, there is no 13.5, 14 or 21 without multipliers below 7x. 2000MHz is close to my CPUs ultimate speed. So 2000/13 = 154Mhz is the maximum FSB. To get 166FSB I'd need 12x, and a mod.
So I was trying out my new knowledge, but went over the CPUs limit and crashed. Since then, nothing I can do will get it to boot. Not switching CPUs or DIMMs. Not moving jumpers. And the CPUs/memory still work perfectly elsewhere, just as they did in this mobo after I can got it to post once.
Pathetic that I'm going to have to RMA it because it won't boot with a CMOS reset. I even took the battery out and closed the jumper for a couple of hours. I see people with this no-boot situation and people tell them to get rid of that crappy memory (like Crucial) and get some Wanker Ubermench ZZZ which has never been known to fail. Dang, where is some Wanker Ubermench when I need it!
Story and vent over.
This is a story that may have some useful parts. If not, I get to vent.
To tell the end: The board does not now work.
To start at the beginning: It's a refurb from Newegg, $80 shipped. By all accounts, their "refurbs" have a high no-work instance, and sometimes are missing accessories (as Newegg says.) I got all the accessories, which is like a stack of ten items with this deluxe model, some unique, so I lucked out there. The board and items looked perfect. Only the mobo was unsealed. I figured a refurb naturally had been set up for a different CPU, etc., so would likely need a CMOS reset to work, which it did. (First I just tried setting the FSB jumper to 100Mhz.) It took about twenty go-rounds of CMOS resets and reset button pushes before it would boot (in the middle of which I set the memory speed jumper to 100MHz.) Rather troublesome, but resetting the CMOS hasn't worked completely reliably or regularly for me with the other Athlon mobo I have that OCs from the BIOS, or if switching CPUs.
So the board was up and running on a box outside the case and using my extra, but known good, parts; like I do normally before committing to a new mobo. I was using my old 800MHz TBird and a Radeon 7200 PCI video card. I reinstalled XP over itself on one partition to get XP to work. I played with my new toy for about 8 hours. Ran prime95 for a couple of hours at 9.5 x 100. Blasted away in Max Payne for a couple hours. The mobo OCed to 1066 (8x133, not stable in XP), but 7x133=933 was fine with prime95. Put the memory up to 133, then async to 166 and 178. Prime95 for an hour. But memory at 200 was a no-boot. Reset the CMOS with about as many tries as before
Now here is what I did not like at all. I wanted to try the CPU at 6x 166=1000 or maybe 5.5 x 166 =913, or even 5x 200=1000. There is nothing below 7 in the BIOS. It has from 7 to 13 (and Hardware default), that's it. It doesn't go very low for voltage either. I was out of interesting options. Oh well, it is getting late. On to the "real" CPU, a 1700+ modified from an 11 multiplier to a 15 (very long story elsewhere), and which has been OCing many months to 2000MHz. To get ready for the CPU exchange, I set the BIOS to Hardware multiplier and Hardware voltage, memory ratio to 1:1. I set the jumpers to 100Mhz FSB and 100MHz memory. That way the mobo should boot perfectly for the next CPU at 15x 100 = 1500, very close to the factory speed of 1467 for 1700+. Since I had so much trouble resetting the CMOS, I made sure it booted from a complete power-off a couple of times with the 800.
But it would not boot the 1700+. After many power off/ons and button resets, I resigned myself to to resetting the CMOS, which again was ridiculously flakey. It eventually boots, but it boots at 2400+ (WHA???), which is 2000MHz not 1500. Unfortunately, without overvolting, my 1700+ is barely stable enough to boot at 2000MHz . I would get into the BIOS to set working settings and the computer would crash before I made it out of the BIOS. A few times I made it out of the BIOS, and it was still identified as 2400+ even though I set the multiplier as low as it would go (7x), which made no sense.
At 2AM nothing made any sense. I had reset the CMOS and it wouldn't boot over and over, so I called it a day. In my sleep, with my brain in neutral, it was all clear: the mobo must be setting the FSB to 133 and the multiplier to 15 (15x 133 =2000), so I would set the memory jumper to 133 and try booting again. Awake, I didn't get the sleep logic about the memory jumper, but the memory always worked at 133 IAC, so I did it, and the board immediately booted. At 1800+. Wha??? Why? It was always 2400+ before. ( Yes, it is the right jumper in the right position according to the manual. ) But I'll take it. Ran prime 95 for hours. Blasted awhile in Max Payne. Pulled stuff across the LAN. Played some MP3s. Everthing is perfect. The board boots fine from power-off.
Now for OCing. I upped the FSB to 133 and tried a few low multipliers like 9 and 10 to be safe, but none would work except 7 (which mysteriously gave me 2000MHz), and I had to do the time consuming CMOS reset that worked so sporadically. Eventually it penetrated my dense brain that the "deluxe" Delta ILSR mobo was not setting the fifth multiplier bit. (OBSOLETE PIECE OF JUNK!). Therefore my CPU was keeping that bit high, and 9 was really 17, 10 was 23 and so forth. No wonder they didn't work. I confimed that by trying it with FSB 100. Yep, 9 gives 17x 100 =1700MHz, etc. (7 gives 15x , BTW.) Now that I had overcome my misconception that the mobo was a contemporary design, things were falling into place.
The lowest multiplier this mobo will set (without a hardware mod) is 13 (would be a 5x on CPUs factory set to multpliers below 13x). Because of the way the multipliers translate, there is no 13.5, 14 or 21 without multipliers below 7x. 2000MHz is close to my CPUs ultimate speed. So 2000/13 = 154Mhz is the maximum FSB. To get 166FSB I'd need 12x, and a mod.
So I was trying out my new knowledge, but went over the CPUs limit and crashed. Since then, nothing I can do will get it to boot. Not switching CPUs or DIMMs. Not moving jumpers. And the CPUs/memory still work perfectly elsewhere, just as they did in this mobo after I can got it to post once.
Pathetic that I'm going to have to RMA it because it won't boot with a CMOS reset. I even took the battery out and closed the jumper for a couple of hours. I see people with this no-boot situation and people tell them to get rid of that crappy memory (like Crucial) and get some Wanker Ubermench ZZZ which has never been known to fail. Dang, where is some Wanker Ubermench when I need it!
Story and vent over.