What am I doing wrong? E6400 can't 400FSB

S

Surly73

[H]ard|Gawd
Joined
Dec 19, 2007
Messages
1,782
So...

I've mostly been about stability, and have never bothered to overclock anything (unless it came that way like a BFG GPU). I understand the concepts (I'm an EE) but haven't followed it with an enthusiast passion. For instance I don't have all the "good" steppings and revs memorized.

I have to say, though, that reading recently about people getting 400FSB from E6400s with stock voltage and cooling got me interested in a free upgrade from 2.1GHz to north of 3.0.

I've read the C2D sticky in this forum, searched a number of other sites, and read this http://www.thetechrepository.com/showthread.php?t=30

E6400 stepping 6 revision B2
OCZ XTC DDR2-800 Platinum rev 2 running SPD timings
Asus P5B-E (the earlier version with stepping C1 of the P965) BIOS 1601
Arctic Cooling Freezer 7 Pro, Q-Fan set to "performance"
Antec NSK-4400 (I think) w/ SilenX 120mm, Q-Fan set to "performance"
Enermax EG465P-VE
8800GTS/512 @stock
Audigy 2ZS
Leadtek TV2000XP tuner
Seagate 7200.10 750G

- PCI-E locked at 100
- PCI locked at 33.3
- Vmem auto
- FSB term auto
- Vcore auto or as specified

My problem is that I don't hit instability, I'll go from stable to completely unable to POST.
FSB333 1:1 is no problem (memory is actually underclocked)
FSB366 1:1 at least booted
FSB400 completely fails to POST, tried CPU VCore auto, 1.295, 1.3875

Based on the article I referenced above about ASUS chipset strapping, I tried FSB 399, 400 and 401 - all failed to POST.

- I don't see this being a Vmem issue, the memory isn't even overclocked yet at 1:1.
- I've used both AUTO and increased Vcore to what should be perfectly reasonable for the speeds I'm trying.
- I've played with FSB termination voltage past mid-range - no change.
- I don't see an MCH voltage setting in my BIOS
- Since it fails to POST it isn't a heat issue. At FSB333 my temps are 35C-50C (TAT)
- I haven't been disabling C1E or SpeedStep because I want those features and I'm doing pretty mild OC compared to the experience of others with this same MB and CPU. If 400FSB 1:1 was stable I wouldn't even try anything higher.
- I know the PSU may be a borderline, but it's high quality and I feel I should have instability before getting to a no-POST situation if it's the PSU. As far as I know, the GPU doesn't throw 100% load during POST anyways plus that would be on the 12V rail. At stock clocks, even with the new GPU, the system is 100% stable.

Do I just have a dud somewhere?
 
Some updates and additional questions. I've read a bunch more guides, some specific to P965, Asus P5B series boards and E6x00 C2Ds.

I was eventually able to get my system to POST at 400 by a surprising method. I originally had the BIOS set to obey SPD for memory timings. CPU-Z indicated that SPD was setup for 5-5-5. When I manually set 4-4-4-15 (the rated timings) or 5-5-5-15 (supposedly the SPD timings) it POSTs fine. Go figure.

Recall that I have OCZ XTC Platinum Rev2 DDR2-800 memory. During all attempts I was running 1:1, which means at 375 it was actually underclocked. Am I correct in my understanding that FSB400 1:1 will actually have the memory running at its rated clock speed and overvolting shouldn't be necessary?

I noticed reduced stability at 400 1:1 with the rated 4-4-4-15 timings, which I find odd. Perhaps my memory has a defect...

I don't consider an evening of 40 reboots fun, or to be my hobby, so I only did so much work on this last night. I would like to try to get 400 working because of the strapping issue for the Northbridge on P965s. Note also that I have a v1.01G P5B-E so it has C1 stepping P965 which cannot adjust the NB, SB, MCH, ICH voltages. I'd rather not go with 375, for example, and have the NB heavily overclocked.

For whatever reason, it looks like I need more CPU Vcore than average to pull this off. I was getting flaky results (couldn't get through bootup) with anything less than 1.4V@400 even though some guides considered 1.375 to be "guaranteed" starting points and more than adequate on the E6400. :confused: It made it through POST at stock Vcore (which some people here say they are using) but that's about it.

If I lower the CPU multiplier I will reduce the need to overvolt the CPU right? Is my best bet to go for 400-425 (to keep the NB mildly OCed and voltages down), multiplier of 7 instead of 8 to reduce Vcore and see how things end up? I also believe that running 400ish 1:1 means I'm actually running the memory at its rated speed so no overvolt is required there.
 
Need more voltage.

My E6400 did 3.2GHz @ stock, 3.4GHz @ 1.41 and 3.5GHz @ 1.44
 
Well, that's what I mean. Mine won't even do 3.0GHz at stock, or at 1.375, but only at 1.4.

I guess I have a "below average" E6400? Or perhaps something else is still causing the crashes. I'd be happier with 3.0@stock than [email protected]. I'd like to keep the heat and reliability under control and the upgrade from 2.1 to 3.0 would be "enough" for me.
 
A couple of things, manually set the vmem to what ocz says it should be for the rated 400mhz speed. Also, lower the cpu multiplier to 6 or 7 and see if that allows you to run stably. If it does, then you need more voltage
 
skadebo said:
A couple of things, manually set the vmem to what ocz says it should be for the rated 400mhz speed. Also, lower the cpu multiplier to 6 or 7 and see if that allows you to run stably. If it does, then you need more voltage
Click to expand...

For the rated speed it says 1.9-2.1V. Here's my thing - if the system is stable on "auto" voltage at stock clocks but with an asynchronous multiplier to reach the rated DDR2-800 shouldn't it also be stable up to 400MHz in 1:1 since it's really the same clock speed? I'll try it anyways - not arguing, just trying to confirm what I believe to be true.

I will also be trying the CPU multiplier to see if that reduces voltage/increases stability.
 
I see, and you're correct. That may mean that the motherboard isn't able to work at 400mhz FSB so maybe the northbridge needs more volts.
 
skadebo said:
I see, and you're correct. That may mean that the motherboard isn't able to work at 400mhz FSB so maybe the northbridge needs more volts.
Click to expand...

It may, which I'm unable to do on a v1.01G P5B-E. However, when I read http://www.thetechrepository.com/showthread.php?t=30 it leads me to believe that 400 or 401 should be "easy" as ASUS has set up the strapping to change before 400. At any rate, 400 should be easier than, say, 375 because of the strapping issue, for instance.

This has led to my frustration over an inability to do what others are doing "easily" :( I don't want to OC "at all costs" (ie. 1.5V+ Vcore), but I was hoping for an easy 3.0-3.2 with good Vcore, power consumption and temps.

Thanks,
 
I have a very similar setup (p5b, E6400 &ddr2 800), and did quite a bit of work on my overclock last night, so its all still fresh in my mind. The main thing I had problems with was my memory timings and speed. It seems I was trying to run the memory too fast and with too tight timings (the PC would not even post and resetting the CMOS jumper didn't put things back to normal!!). Luckily I had some spare PC7200 around I could slip in to get things going again. So what I ended up doing was loosening the timings from 4-4-4-12 to 5-5-5-15, keeping the memory divider 1:1 and then stepping up the FSB 10 mHz at a time. Started the vcore at its lowest setting (around 1.213?) and then tested for stability with prime95 for only a couple of minutes when it booted.
If I got an error I rebooted and stepped up to the next vcore level, and tested again.
Ended up FSB@375 for 3GHz, Memory @ 750MHz with vcore at 1.325.Would have probably gone further too as with both cores running 100% with prime95 I was only at 55C but I ran out of time.
 
Thanks for the post. I may also try running with looser-than-spec memory timings even though I'm not overclocking the memory to see what happens. Do you happen to know the rated timing on your RAM? Was 4-4-4-12 spec, or tighter-than-spec?
 
It was Geil Black dragon 4-4-4-12. I thought in theory it should run up with those timings up to 800MHz but in practice it was not to be. I my try to tighten them later but at the mo just don't have time...
 
I have the same problem with my E6400, not quite the same board though ( G33-DS2R ), but everything right up to 400 is fine. At 400 it's insta-no-post for no reason, have you tried going over? Mine works at 415 and 420 and requires...a rather large amount of voltage, which pushes the temperatures higher than I like.
 
Try giving the MCH +0.1V. I have to give vMem at least 2.0V for it to work at 800Mhz (same RAM as you). Your post does not indicate that you have 2 or 4GB of memory. I have to back timings off to 5-5-5-15 for stability with 4GB RAM, and have to give another +0.1V to the MCH (+0.2 total). Also try backing your CPU's multiplier to 7 to see if the FSB is the problem...
 
dook43 said:
Try giving the MCH +0.1V. I have to give vMem at least 2.0V for it to work at 800Mhz (same RAM as you). Your post does not indicate that you have 2 or 4GB of memory. I have to back timings off to 5-5-5-15 for stability with 4GB RAM, and have to give another +0.1V to the MCH (+0.2 total). Also try backing your CPU's multiplier to 7 to see if the FSB is the problem...
Click to expand...

Not sure if all comments are addressed to me.

I cannot adjust MCH voltage on this board, I can adjust FSB termination.

I have 2x1GB DDR2-800 OCZ Plat rev 2.

Backing off the multiplier is on tonight's list :)
 
Upped mem to 2.0, upped FSB term to 1.4, set FSB for 400, timings to 5-5-5-15.

CPU wasn't stable at 8x until 1.3875 in the BIOS, which was 1.336 idle and 1.264 under load according to cpu-z. Orthos always failed on the second core - interesting.

Temps in TAT hit around 60C, Orthos hit mid 50s C, ran for about 20 minutes before I stopped it. I noticed a bit of a hot/fried electronics odor that I'm highly tuned to (as an EE) and immediately set everything back to stock timings and voltage.

None of the temps or voltages were in a danger range as far as I'm concerned during the test. The main 120mm fan in the case continues to spew a very faint fried smell (the PSU fan does not) and the machine is working fine... Let's just say I won't be leaving it unattended and on for a while. I opened the case and did not observe any split or distorted caps, nothing was overly hot to the touch, no localization of any smell from the board. All the onboard voltage readings and temps are great.

Maybe the board in my SilenX 120mm itself is coincidentally letting go :D

Hmph
 
I wasn't planning to take it all apart again tonight, but I did anyways. Right after shutdown I shoved my sniffer in the case and smelled the updrafts from all the major component areas, smelled the DIMMs etc... Any smell I could detect was located behind the CPU and nowhere else. This was either the CPU Vcore regulators or the 120mm fan.

It looks like its the fan itself. This SilenX fan is only 1 month old and is already a warranty replacement for a 10 month old original that came apart - the bearing wore so much it simply separated. (and this isn't a 24/7 system). I guess shutting down Q-fan and having it run full tilt while working on OC caused it trouble.

Sheesh. We can put men on the moon but finding a good fan...
 
Surly73 said:
stopped it. I noticed a bit of a hot/fried electronics odor that I'm highly tuned to (as an EE)
Click to expand...

Second time you've mentioned this in this thread. This has absolutely no bearing on the discussion. None of your degree applies to computer hardware installation and troubleshooting :) It also shows you are the man at connecting Vcc to ground. :) :D

edit: FWIW, OCZ Plat Rev 2 is speced at 4-4-4-15 at 800 mhz.
 
dook43 said:
Second time you've mentioned this in this thread. This has absolutely no bearing on the discussion. None of your degree applies to computer hardware installation and troubleshooting :) It also shows you are the man at connecting Vcc to ground. :) :D
Click to expand...

...and that I'm an expert at making, as well as detecting, foul odours. :D
 
You must log in or register to reply here.
Back
Top