K8N Neo4-F

[DoobieOnline]

New 1.73b and 1.73b "mod" bios available at lejabeach.com. I have no idea what the changes are and no info is listed. Use at your own risk!

 

[feelthaflo]

I'm also having the underclock problem with BIOS 1.5, but only when I try to increase the voltage. Trying 1.71b now, hopefully it works as well as it did for everyone else. Thank you.

 

[feelthaflo]

Voltage settings appear to work, but can't get 2.6. Running now at 2.4, which is acceptable (since it's free), but if I go too much higher I get BSoD with "Driver_IRQL_Not_Less_or_Equal." Did I just get a lower quality CPU than others here or can this be fixed with a driver update?

 

[Cobalt]

OCing can sometimes corrupt the OS (so I've heard?), so keep it at 2.4ghz.

 

[DoobieOnline]

I got curious and tried the 1.73b bios. The only difference I could see is that it doesn't have the 140, 150, or 180 mem dividers but it does have support for several dividers higher than 200. I didn't test the higher dividers, though. I wanted the 140 and 150 dividers for OC'ing so I switched back to the official 1.6.

On another note, I picked up a 3700+ Sandy on a hot deal (now dead) from Monarch and it arrived yesterday. I was able to get it up to 2.75GHz using 11x250 on the Neo4-F but the dividers were doing something strange. Using 166 divider at 11x250 caused cpu-z and Everest to show mem running 192 even though bios showed 207. I switched methods and started using the 10x multiplier. So far I've gotten it stable at 10x260 and the 166 divider puts the mem where it should be at 215. I'm only using 1.45V to get to 2.6GHz so I feel pretty good about that. I'll definitely keep trying to get more out of it using the 10x multi and would love to hit 2.8GHz but I just don't know if that will work out since I topped out at 2.75GHz using the 11x multi and 1.55V. The San Diego core runs a bit warmer than the Venice but only to the tune of 3 or 4C. I haven't run any benchmarks but I expect to see slight increases since the Sandy has 1MB L2 cache. I'm not too familiar with how changing the multiplier affects the mem divider but it was a surprise to see the mem running slower than anticipated using 11x and 166.

 

[feelthaflo]

Could Vcore fluctuations cause my BSoD? It goes at least .02 either way. 1.71b let me set Vcore in BIOS, but if I want it higher than that, the error is multiplied by the percentage as well, increasing the error even more. I've used a generic 500W and an Antec 350SP, both with about the same results. Does this motherboard naturally have a wandering Vcore line, or can I rectify the problem with an Antec TPII430 or Neo480?

 

[DoobieOnline]

My Vcore fluctuates about the same as yours. Ranges from .01 over to .04 under. Kinda sucks because I have to run Vcore a little higher than I'd like to. I'm using a Tagan 480W that has great rails, too. Many motherboards have Vcore that fluctuates, though. The only one I've ever used that was always rock-steady is the Soltek NF3 250GB board.

Set your Vcore a little higher and see if it clears up the BSOD problem. You might try upping your chipset and mem voltage a bit to see if that helps, too.

 

[Cobalt]

My Vcore fluxes as well, so I had to bump it to 1.6v (says 1.58 in BIOS). I could of had 2.7ghz stable at 1.55v but the fluxes caused reboots in Sandra and 3DMark05, and oddly enough NOT in Prime95 stress tests.

 

[Serpentor]

Originally posted by: DoobieOnline
I got curious and tried the 1.73b bios. The only difference I could see is that it doesn't have the 140, 150, or 180 mem dividers but it does have support for several dividers higher than 200. I didn't test the higher dividers, though. I wanted the 140 and 150 dividers for OC'ing so I switched back to the official 1.6.

I'll second that, 150 divider on the 1.6 bios is very stable, CPU-Z now correctly states my RAM speed whereas it showed simply blank on the 1.3 bios. I only infrequently crash playing CS:S, but only on very hot evenings. I may actually back down from 2.7ghz as my DVI LCD caps me at 60hz/60FPS anyway, so for most of today's games I won't be seeing the extra frames the overclock could bring me, even at 1280x1024 4xAA/AF.

I noticed WinXP taking longer to load with the 1.6 bios, I even tried reformatting and it didn't do the trick. Normally the XP load screen stays on for 1 second or less (the blue loading scroll doesn't have time to appear or moves only a cm or so). Now it's at least a few seconds and the blue scroll makes about 4 complete passes. I know I'm being picky.. but is there a fix or app that let's me see why XP is stalling a bit, there's no HD activity to account for it?

 

[Cobalt]

Try TuneXP.

 

[brikis98]

ok, a question. I have the following setup:

MSI Neo4-f
AMD Athlon 3000+ (Venice core)
2x512MB PC3200 Kingston ValueRam

I can run Prime95 stable at 290FSB and x9 multiplier, or 2610Mhz (idle temp 38C, load temp 48-52C). However, unless I use the 100 divider for my RAM, the system won't boot. Even 133 won't work. CPU-Z gives me the following timings for my RAM:

Frequency: 145.0Mhz
CAS Latency: 2.5
RAS to CAS: 2
RAS Precharge: 2
Cycle Time: 5

I used Zebo's guide, and tested that my motherboard can go up to 320FSB when CPU multi is x6 and up to 310 when x9 multiplier (both times RAM at 100). I have not been able to follow his advice for testing my RAM, as I don't have a floppy drive to run memtest. However, its pretty cheap value RAM and probably can't do more than 200 or 210.

I guess I'm supposed to "consolidate" now, but am not sure how. I thought that I should be able to use the 133 divider with 290 x9, but I can't, and was wondering if any1 else with this mobo had the same issues? And, perhaps more importantly, does it really matter performance wise if I can use the 133 divider? Or is getting to 2.6Ghz going to overshadow that?

 

[brikis98]

anyone?

 

[DoobieOnline]

Hey brikis, what timings are you trying to use at 290 FSB and 133 divider? That would put your mem at 192MHz, which is close to its rated speed so you may need to relax your timings a bit. Try 3-3-3-8-2T for grins and if that works try 3-3-3-8-1T to see if your stable. Also, bump your VDIMM up to 2.7 (which is a perfectly safe voltage). If you can get stable at 2.7V and 3-3-3-8-1T, then try lowering your timings and see if you can get to 2.5-3-3-7-1T. Hope this helps!

 

[brikis98]

Originally posted by: DoobieOnline
Hey brikis, what timings are you trying to use at 290 FSB and 133 divider? That would put your mem at 192MHz, which is close to its rated speed so you may need to relax your timings a bit. Try 3-3-3-8-2T for grins and if that works try 3-3-3-8-1T to see if your stable. Also, bump your VDIMM up to 2.7 (which is a perfectly safe voltage). If you can get stable at 2.7V and 3-3-3-8-1T, then try lowering your timings and see if you can get to 2.5-3-3-7-1T. Hope this helps!

thanks for the reply. my timings were set to auto (only one i set manually was 2T), so maybe you're right, and the mobo was trying to do timings that were too tight for 192Mhz. I did bump the mem voltage up to 2.7 and even higher, which didn't help. i'll try entering slower timings manually and see if that helps.

btw, which is more important for game performance: tight timings (2.5-2-2-5) at 145Mhz or loose timings (3-3-3-8) at 192Mhz? Does it primarily depend on the game? Is there a noticeable difference?

 

[DoobieOnline]

Not really sure about how the timings will affect your games since the speeds are so far apart. My guess is that 192 would be preferable. If you can get it stable at 192 loose, try running a couple of game benchmarks to see how it compares to 145 tight.

Are you using the 2 mem slots closest to the cpu?

Edit: I just remembered that the 133 divider is broken on many if not all of the official bios. Are you still using 1.71b? cobalt has verified the 133 divider works on 1.71b. For grins, bring your FSB down and try the 166 divider (use 9x240 which would put your mem right at 200MHz). That way you'll know for sure if it's the mem or a flakey 133 divider).

 

[brikis98]

Originally posted by: DoobieOnline
Not really sure about how the timings will affect your games since the speeds are so far apart. My guess is that 192 would be preferable. If you can get it stable at 192 loose, try running a couple of game benchmarks to see how it compares to 145 tight.

Are you using the 2 mem slots closest to the cpu?

Edit: I just remembered that the 133 divider is broken on many if not all of the official bios. Are you still using 1.71b? cobalt has verified the 133 divider works on 1.71b. For grins, bring your FSB down and try the 166 divider (use 9x240 which would put your mem right at 200MHz). That way you'll know for sure if it's the mem or a flakey 133 divider).

I am using the 2 mem slots closest to the CPU, and the mem is recognized as running in dual channel.

I am on the 1.71b bios. once i get home from work, i'll try running the 166 divider with the settings you mentioned. i'll also try running 133 but with lower FSB settings. thanks for the advice

 

[brikis98]

ok, so, some wierd test results.

the 133 divider DOES work. I was able to boot with at 280x9 (2520) using the 133 divider, or the RAM at 186Mhz. However, anything higher than that (even 282) won't boot. I tried very loose timings (3-3-3-8) but anything above 280FSB would not boot with the 133 divider.

With the 100 divider, I can boot at 300x9 (2700), although I run at 290x9 (2610) for more stability. At this setting, my RAM is only at 145Mhz. I can use very tight timings though (2.5-2-2-5) and everything runs well. So, it seems like my RAM is crappy.

Here comes the wierd part: I can boot and run stable at 240x9 (2160) with the 166 divider, or the RAM at 200Mhz. The timings are loose again (3-3-3-8), and everything works.

So, my RAM can clearly do 200Mhz. Of course, I kind of knew this since at stock settings (200x9, no divider) that's what it runs at. And, the 133 divider works. However, if I try to use the 133 divider and get my RAM to run at anything higher than 186Mhz, the system doesn't boot.

What the heck is going on? Is this a mobo issue? A RAM problem?

And if I can't get it resolved, am I better of running 290x9 (2610) with my RAM at 145Mhz (100 divider) and tight timings (2.5-2-2.5) or running at 280x9 (2520) with my RAM at 186Mhz (133 divider) and loose timings (3-3-3-8)?

Thanks.

Edit ---> forgot to mention, the memory timing is 2T; i tried 1T in various configs and had random crashes and BSODs, so I went back to 2T.

 

[DoobieOnline]

I wish I could say what's wrong but I can't figure it out. Did you ever try the 1.6 bios? If not, give it a go and see if it helps. One of the great things about 1.6 is the additional memory dividers not found in 1.71b. You get 140, 150, and 180 dividers to try.

 

[brikis98]

Originally posted by: DoobieOnline
I wish I could say what's wrong but I can't figure it out. Did you ever try the 1.6 bios? If not, give it a go and see if it helps. One of the great things about 1.6 is the additional memory dividers not found in 1.71b. You get 140, 150, and 180 dividers to try.

ooh, that's pretty sweet. does the 1.6 bios have the multiplier issue, where in windows, it always reverts to x5?

 

[Cobalt]

I'm using the 166 dividers on my RAM now, running it at 250mhz, CAS 2.5 and 2T.

 

[DoobieOnline]

brikis, here's a good article showing higher speed to be of more importance than tight timings - link. It's for 2x1GB, but the results should be the same for 2x512MB.

I haven't had the problem of settings reverting to 5x HTT in Windows with any of the bios versions. Are you running Cool N Quiet or using Core Cell? I haven't used either, but that's the only thing I can think of that would do that.

 

[brikis98]

i had that issue with the 1.5 bios. Cool N' Quiet was disabled, Core Cell was installed. With the 1.71b bios, the issue went away. I'll try out the 1.6 bios next time i get a chance and see what kind of results I get with it.

 

[wanderer27]

Originally posted by: brikis98
ok, so, some wierd test results.

the 133 divider DOES work. I was able to boot with at 280x9 (2520) using the 133 divider, or the RAM at 186Mhz. However, anything higher than that (even 282) won't boot. I tried very loose timings (3-3-3-8) but anything above 280FSB would not boot with the 133 divider.

With the 100 divider, I can boot at 300x9 (2700), although I run at 290x9 (2610) for more stability. At this setting, my RAM is only at 145Mhz. I can use very tight timings though (2.5-2-2-5) and everything runs well. So, it seems like my RAM is crappy.

Here comes the wierd part: I can boot and run stable at 240x9 (2160) with the 166 divider, or the RAM at 200Mhz. The timings are loose again (3-3-3-8), and everything works.

So, my RAM can clearly do 200Mhz. Of course, I kind of knew this since at stock settings (200x9, no divider) that's what it runs at. And, the 133 divider works. However, if I try to use the 133 divider and get my RAM to run at anything higher than 186Mhz, the system doesn't boot.

What the heck is going on? Is this a mobo issue? A RAM problem?

And if I can't get it resolved, am I better of running 290x9 (2610) with my RAM at 145Mhz (100 divider) and tight timings (2.5-2-2.5) or running at 280x9 (2520) with my RAM at 186Mhz (133 divider) and loose timings (3-3-3-8)?

Thanks.

Edit ---> forgot to mention, the memory timing is 2T; i tried 1T in various configs and had random crashes and BSODs, so I went back to 2T.

Hey, did you get this one figured out ? I'm a noob to OCing but I've been digging into it and I may have an idea of what's going on here.

What is your HT (or LDT or what ever MSI is calling it) multiplier set to ? This should be a multiplier value between 1-5, with 5 being the default.

If your HTT=280, CPU mult =9, MEM divider = 133 here's what I see:

HT(LDT)=3 -> HyperTransport = HTT*HT(LDT) = 840. CPU @ 2520, Mem @ ~187 Mhz
This (HyperTransport) is kinda low, so I'm thinking you're using this:

HT(LDT)=4 -> HyperTransport = 1120. CPU @ 2520, Mem @ ~187
HyperTransport is above the 1000 it should be, and from what I've heard this will cause all kinds of flaky problems. So if this is the case I'm not suprised you're having issues raising the HTT even higher (so Mem won't run faster).

I'm thinking if your MB can do HTT @ 320, you may be able to try this:

HTT=300, CPU mult =9, MEM divider =133, HT(LDT)=3.
This should give you HyperTransport = 900, CPU = 2700, and Mem = 200.
You may even be able to bump HTT upto 305.

If HT(LDT) is currently at 4 (god forbid it's 5), then this may solve your problem.

Of course as I said, I'm an OC noob, so I may be way off here

 

[brikis98]

Originally posted by: wanderer27

Originally posted by: brikis98
ok, so, some wierd test results.

the 133 divider DOES work. I was able to boot with at 280x9 (2520) using the 133 divider, or the RAM at 186Mhz. However, anything higher than that (even 282) won't boot. I tried very loose timings (3-3-3-8) but anything above 280FSB would not boot with the 133 divider.

With the 100 divider, I can boot at 300x9 (2700), although I run at 290x9 (2610) for more stability. At this setting, my RAM is only at 145Mhz. I can use very tight timings though (2.5-2-2-5) and everything runs well. So, it seems like my RAM is crappy.

Here comes the wierd part: I can boot and run stable at 240x9 (2160) with the 166 divider, or the RAM at 200Mhz. The timings are loose again (3-3-3-8), and everything works.

So, my RAM can clearly do 200Mhz. Of course, I kind of knew this since at stock settings (200x9, no divider) that's what it runs at. And, the 133 divider works. However, if I try to use the 133 divider and get my RAM to run at anything higher than 186Mhz, the system doesn't boot.

What the heck is going on? Is this a mobo issue? A RAM problem?

And if I can't get it resolved, am I better of running 290x9 (2610) with my RAM at 145Mhz (100 divider) and tight timings (2.5-2-2.5) or running at 280x9 (2520) with my RAM at 186Mhz (133 divider) and loose timings (3-3-3-8)?

Thanks.

Edit ---> forgot to mention, the memory timing is 2T; i tried 1T in various configs and had random crashes and BSODs, so I went back to 2T.

Hey, did you get this one figured out ? I'm a noob to OCing but I've been digging into it and I may have an idea of what's going on here.

What is your HT (or LDT or what ever MSI is calling it) multiplier set to ? This should be a multiplier value between 1-5, with 5 being the default.

If your HTT=280, CPU mult =9, MEM divider = 133 here's what I see:

HT(LDT)=3 -> HyperTransport = HTT*HT(LDT) = 840. CPU @ 2520, Mem @ ~187 Mhz
This (HyperTransport) is kinda low, so I'm thinking you're using this:

HT(LDT)=4 -> HyperTransport = 1120. CPU @ 2520, Mem @ ~187
HyperTransport is above the 1000 it should be, and from what I've heard this will cause all kinds of flaky problems. So if this is the case I'm not suprised you're having issues raising the HTT even higher (so Mem won't run faster).

I'm thinking if your MB can do HTT @ 320, you may be able to try this:

HTT=300, CPU mult =9, MEM divider =133, HT(LDT)=3.
This should give you HyperTransport = 900, CPU = 2700, and Mem = 200.
You may even be able to bump HTT upto 305.

If HT(LDT) is currently at 4 (god forbid it's 5), then this may solve your problem.

Of course as I said, I'm an OC noob, so I may be way off here

thanks for the help...

with FSB at 290 and CPU mult of 9, I had my LDT at 3. The HTT is then a bit low, but I've read to keep it below 1000. i basically use an LDT of 3 for anything above an FSB of 250/260.

as i've said, using the 133 divider i can't set my FSB higher than 280 or the system won't boot. with the 100 divider, i can get the FSB up to 310 or 320 (with the x6 multiplier) so it's not the mobo. with the 100 divider and x9 multiplier, i can get the FSB to 300 or 310, so it's not the CPU. and, as i mentioned, with the 166 divider, i can get the FSB to 240, so my RAM can definitely do 200Mhz.

this is why i'm confused: i can't figure out what the limiting factor is.

my last remaining guess is a fault 1.71b bios, so i still need to try 1.6. unfortunately, i haven't had time. i'll post results once i do.

 

[wanderer27]

Well, my theory was off on that

So you can run at CPU=2520 & Mem=~187 (FSB=280), or CPU=2610-2790 & Mem=145-155 (FSB=290-310) ?

And with the 133 divider 280 is tops, even with a lower CPU multiplier, right ?