3.0C Overclock Status

[mongre556]

My first post to these forums but I have been lurking for some time.

Based on various independent information sources I decided to try my hand at overclocking.

I selected the following

Asus P4C800-E Deluxe
Antec True Power 430W PS
Intel 3.0C (Retail, got an SL6WK L348A652, luck of the draw)
OCZ 3700 Gold Rev 2
Zalman 7000A-Cu Heatsink
Artic Silver 5

Now I will mention that I am not going for the highest possible overclock, I just want to hit the specified maximum FSB for my RAM at 233FSB. I figure I am doing pretty well at what appears to be the average 3.0C overclock. For me it is a tradeoff, I want to keep power and heat as low as possible, but get slightly better performance than what is available today stock.

I had the system for a week before overclocking it, to let the Artic Silver cure. Played games pretty steadily on it for extended periods, but as the AS site said gave it periods of cool down.

On monday though I decided to overclock. I had no problems stepping from 205-210 and now I am running at my target of 233 and am performing Prime95+Folding@Home tests plus my own regular gaming use.

I did have a failure of Prime95 last night at the auto voltage setting so I increased the voltage to 1.6 as a first step. Then I had a game crash tonight so pushed it to 1.625, then a few hours later another crash to desktop. Now I cannot of course rule out the game just crashing, it is generally pretty stable under my old system(AMD 2400+), but the occasional CTD is not unheard of.

To be sure I am running the Prime95+Folding tests to get a more accurate read.

My settings are the following (otherwise I run the defaults)

233FSB
1.625Vcore
2.75Memory Voltage
Spread Spectrum (disabled but had it enabled prior to this last change)

That is about all I have tweaked...memory is all auto as I am happy running it at its default settings. My goal is to stress the processor, but have plenty of flex left in my other components.

Under load I run

1.488 VCore with occasional increase to 1.504 (~2% flux)
52 Deg C CPU, 35C Motherboard

According to the poll on this site I am just above what most people have for temperature under load. I am pretty sure I can shave a couple of degrees of that too, just need to make some minor cooling adjustments. Keeping the system relatively quiet is also a goal.

I am kind of curious about the voltage at load. From what I can tell this is the same voltage under load I got when the system was running at spec 3.0C with default voltage. Before I over volted a bit manually in the bios I dropped below 1.488. Does this seem reasonable?

 

[batboy]

Asus is known to undervolt during load, but that seems more than normal in your situation. As long as you aren't going above 1.6v actual measured vcore during load, you can probably try using the 1.65v BIOS setting. What are your 5v and 12v rails doing during load? That Antec PS should be ok, but maybe it's weak?

Make sure your case ventilation is good. Maybe rearrange power and IDE cables so they aren't blocking air flow.

I have a couple suggestions. Update your BIOS if you haven't done so yet and make sure you lock the AGP/PCI to default 66/33 in the BIOS.

 

[Edward2]

WELCOME TO THE FORUMS !!!

I just switched to an Asus P4C800-E Deluxe mobo from an Abit IC7-G mobo, and I saw the same voltage drop under load problem that you described. I decided to do the "droop" volt mod for my mobo, but that may be more than you want to do right now from the sounds of it. My CPU voltage would drop from 1.6 volts at idle to about 1.48 volts under full load.

I agree with Batboy, you can try increasing your CPU voltage to 1.65 volts in the BIOS. However, your voltage may still drop too low under full load.

Try adding a coupe of 120mm fans to your case if you have room. They supply fairly high flow rates and are not too noisy. If you can reduce your mobo temp by 5-6 degrees, your CPU temp should also drop several degrees.

 

[mongre556]

Thanks for the replies.

I do lock the AGP/PCI at 66/33, I left that out of my original post but that was one of the first things I set, even prior to overclocking. Was not looking to stress my AGP card. It is a ATI 9500Pro and I have it set to 1.5volts.

The case I have can't support 120mm fans unfortunately, however the two parallel intake fans can be replaced with higher flow fans at not too much of a noise increase. They are effectively ducted to the outside air so should be fairly effective with increased flow.

My latest increase to 1.625 resulted in a hung system overnight. I clocked the chip back to 3.3 to see if I can get predictable results while at work. VCore is still 1.625 and still drops to 1.488-1.504.

I had found 220FSB to be pretty stable before, many hours of use with no problems. 220FSB was my previous step before 233FSB.

The Other rails seem fairly steady. The 5 volt varies by only one or two thousandths of a volt according to Asus Probe.

I have not flashed the bios on this board as yet however that is probably my next step.

I haveread that V1014 is the Bios to use rather than the newest V1016?

 

[Sophisticated]

increase your vdimm to the max....yes put vcore at 1.65....use a diff ratio like 5:4 = 333 on that board...then see what happens..if all good then you can try and OC a bit more but since u have bad flux then u prolly wont get that far

 

[Edward2]

My P4C800-ED mobo came with BIOS v1014. I tried v1015 but it made my computer LESS stable, so I went back to v1014. I have not tried v1016 yet.

Is your 3.0C a 12-cap or 30-cap CPU (# of rectangles on bottom of CPU)? If your CPU is a 12-cap, then you might have a hard time reaching 233 FSB without either raising the Vcore a lot or doing the "droop" volt mod. The reason is that a 12-cap CPU would probably need 1.55-1.60 volts to reach 3.5 GHz and your CPU voltage is dropping too much under load (1.48-1.50 volts).

If you feel comfortable doing it (I am, but everyone is different), try increasing your CPU voltage to 1.70 volts in the BIOS, and then see how low the voltage drops at 100% load. If your voltage will hold 1.55-1.60 volts, then let it run for a couple of hours at 100% load and see if your CPU is stable. If your voltage will not hold 1.55 volts, then you might as well set it back to 1.60-1.65 volts in the BIOS.

 

[mongre556]

I will set to 1.65V VCore when I get home. I had not considered that what the BIOS setting is could be a fair bit different than the actual voltages supplied.

I had been pretty conservative with the memory voltages but I will try modifying that as well.

I have noticed that Asus Probe does not supply in the information on the vdimm actual voltage. Do any tools report this information for the P4C800E-Dlx?

I have tried to get cpuz but www.cpuid.com has been down for me at home and at work since last week for me.

 

[mongre556]

My P4C800-ED mobo came with BIOS v1014. I tried v1015 but it made my computer LESS stable, so I went back to v1014. I have not tried v1016 yet.

Is your 3.0C a 12-cap or 30-cap CPU (# of rectangles on bottom of CPU)? If your CPU is a 12-cap, then you might have a hard time reaching 233 FSB without either raising the Vcore a lot or doing the "droop" volt mod. The reason is that a 12-cap CPU would probably need 1.55-1.60 volts to reach 3.5 GHz and your CPU voltage is dropping too much under load (1.48-1.50 volts).

Ahhhh, light goes off. The little brown blocks, cap=capacitor.

Well I did not count them (just wrote down all of the numbers on the IHS). I shoulda grabbed the digital camera..

I do remember thinking how many 'caps' there were on my processor ( as many as can fit on the bottom) because it was different than a picture of an earlier P4 3.0C in a review I read that day (written last summer).

Having looked at some more pictures of 12caps (there are spaces where caps could go) and remembering mine had no spots open for more caps I am almost positive I have more than 12, so presumably the CPU is a 30cap.

 

[Edward2]

What I found was that I would set the voltage in the BIOS to 1.60 volts. I would get 1.65 volts at idle and only 1.50 volts at full load. So keep an eye on both your idle voltage and your full load voltage.

AFAIK, there are no programs that will actually read the Vdimm. Apparently the mobo does not make this voltage reading available. You have to measure it directly on the mobo with a multimeter.

Sounds like you got a 30-cap CPU. Hopefully it will be one of the good ones. I never tried running mine at 233 FSB. I went directly to 250 FSB. It may be that your CPU needs just a little more voltage at full load in order to remain stable.

Good luck and let us know how it goes.

 

[Lowkey]

Hello, i am a newbie in P4 OC.

I got two P4 FSB800 3.0Ghz "SL6WK" 30 caps.

Both are locked in a way.

I tried to raise vcore, but i get better results with lower vcore.

3600Mhz are stable at "auto" Vcore (1.525). I tried 1.7 Volt, but i got errors while running prime95.

So i want to know about bios settings... is there a guide for bios options?

Is it better to use divider at 1:1 at low mem timings or 4:5 with high timings?

sometimes i get errors at 3600Mhz, sometimes not. Its strange.

 

[mongre556]

Hello, i am a newbie in P4 OC.

I got two P4 FSB800 3.0Ghz "SL6WK" 30 caps.

Both are locked in a way.

I tried to raise vcore, but i get better results with lower vcore.

3600Mhz are stable at "auto" Vcore (1.525). I tried 1.7 Volt, but i got errors while running prime95.

What are your running voltage settings? Motherboard? Memory? Vdimm setting, Powersupply, Temperatures etc? As someone that is also new to OCing you definitely need to provide these basic bits of information. It really helps others to try to spot the problems.

 

[mongre556]

Here is my update.

I ran all day at 1.625 Bios (1.61-1.488 actual) at 220FSB no problems. That was about 8 hours. I am pretty sure I can drop that Vcore even more and run stable at 3.3 since no voltage setting below 1.65 even changed my vcore under load. So I know I can get that much out of this 3.0C without too much effort.

I then tried to run at 1.65 Bios (1.64 - 1.52 actual) at 233FSB, as suggested, after about an hour Prime95 complained.

This is after I set the VDimm to 2.85V as suggested as well.

I then set the Vcore to 1.675 (1.68-1.552 actual). I ran prime95 for about an hour, it did ok, need to do the long test though since most of my failuers seem to come at between 2 and 4 hours. I am testing at 230FSB at the moment, if it holds I will try for 233FSB again.

My temps are holding well. I believe my biggest cooling issue is air exchange between the outside and the inside of the case. With the case closed 52C, open, 43C under load. So my Heatsink is performing well, if there is enough delta between ambient air and the CPU temp.

My intake fans are still pretty low CFM. So I am going to address that with higher CFM intake and a speed controlled exhaust fan with a high potential CFM I can adjust to suit my heat/noise tastes.

I also noted that most of the cpu's in the database at overclockers.com has a ~3.5 3.0C around the 1.6-1.7 range. It looks like I am still in the average for this CPU. My 3.3 stable at stock voltage also seems to line up with what I see there.

So my question is what are the risks/downside of running a 1.68 idle voltage? I have read that the issue is more running the loaded voltage above 1.6. That is not that case as mentioned above, I am slight over 1.5 under load.

 

[Lowkey]

System1

SB61G2 (i865)+ 250Watt SilentX
P4 3,0 Ghz SL6WK 30 Caps
2x512Mb Twinmos Winbond BH-5

3600Mhz "primestable"
Hypertreading is on

System2

SB75G2 (i865)+ 250Watt SilentX
P4 3,0 Ghz SL6WK 30 Caps
2x512Mb PC400 MDT

"230 at 1:1 primestable"
Hypertreading is on

testing 234 (3500Mhz)....

 

[aNTiChRisT]

250w psu? Not good for overclocking! I would upgrade to at least 350w truepower -- and if you want a nice GPU in there too 450w.

I'm about to start testing my 12cap sl6wu 3.0 now. Wish me luck

~t0m

 

[Lowkey]

No!

ssf barebones are using "better" psu and even the 200Watt psu would be enought to run a monster pc.

I had antec truecontrol 550Watt last year, but there is no reason to buy a besser psu at all. Good cpu & mainboard will run with even a worse psu.

 

[Edward2]

mongre556, sounds like you just might be stable at 3.5 GHz with 1.55 volts under load. In fact, you could probably go even higher (3.6-3.7 GHz) if your idle voltage wasn't so high. 1.68 volts at idle is about as high as I would recommend for 24/7. One thing you could do is run the Folding client on your computer, and that way your computer would never sit at idle. Your CPU temp of 43C at load is very good.

Of course the other option would be to perform the "droop" volt mod to your mobo, but I don't know if you want to try soldering your mobo or not.

Lowkey, have you checked your voltages yet using MBM5 or another program? I would be curious to see how your 12v rail is doing while the CPU is under 100% load.

 

[batboy]

Overclocking requires good power supplies. With a P-4 if you want to O/C, you need a minimum of 350 watts and I recommend more.

 

[mongre556]

mongre556, sounds like you just might be stable at 3.5 GHz with 1.55 volts under load. In fact, you could probably go even higher (3.6-3.7 GHz) if your idle voltage wasn't so high. 1.68 volts at idle is about as high as I would recommend for 24/7. One thing you could do is run the Folding client on your computer, and that way your computer would never sit at idle. Your CPU temp of 43C at load is very good.

Of course the other option would be to perform the "droop" volt mod to your mobo, but I don't know if you want to try soldering your mobo or not.

I wish it was stable, but it is not, even at 3.375 (225FSB). After maybe 3 hours of gameplay, or a couple runs of prime95 torture test I get a crash to desktop in the game, or an error in the prime95 torture test + folding.

So I am going back to a baseline. I want to make sure everything runs stock very stable, especially my RAM. I will also try some 5:4 testing to remove the RAM from the equation once I have confirmed the system is at least a 3.0.

I did have once concern early on with the RAM in that I expected it to be able to run 2-2-3-6 at 200FSB as it can run at 2.5-3-3-7 at 233FSB. It failed almost immediately in prime. That was at auto voltage however.

Running folding is not really an option to keep the voltage down, I do not want to make any special allowances like that. The temperature is good, with the case side off, hehe. I will get that addressed though, it is good to know my Heatsink is working well. As for the droop mod, I would have to think long and hard on that one. I was not expecting to have to modify my motherboard. I still have a few things to try before I go there.

 

[aNTiChRisT]

An asus board, right? That means his load temp of 43 is more like 50-55, sorry to burst your bubble.

And No, you cannot run a P4 (90w) and run all devices on your PC, and THEN overclock. Not only may you damage something, if your rails are low you will become unstable.

~t0m

 

[mongre556]

An asus board, right? That means his load temp of 43 is more like 50-55, sorry to burst your bubble.

And No, you cannot run a P4 (90w) and run all devices on your PC, and THEN overclock. Not only may you damage something, if your rails are low you will become unstable.

~t0m

My understanding is that with the latest asus probe that the temperatures are fairly accurate. Also you cannot make a blanket Asus board statement like that since the sensors used for temperature (on board, on die) differ from model to model and socket to socket. At least back it up with a link to an analysis showing a systemic problem with Asus boards mis-reporting the temperature readings with the latest Asus Probe software.

I would also argue your comment about overclocking and power is not accurate and overly broad. You can certainly run your other devices and overclock, with a sufficiently powerful PS.

I am not though sure if you are responding to the other poster that seems to have hijacked my thread, or me, or both of us.

However if it is directed at me I have no bubble to burst, temperature is only for reference, and since I have been able to have the temperature move by a relative 10 degrees without effective stability (I have had more success at higher temperatures so far) it is not the limiting factor as far as I can tell at the moment.

In addition I am seeing what I expect to see with the Heatsink I have chosen and the thermal paste I have used and given the rooms ambient temperature.

No it is my vcore flux and confirming proper settings for my RAM that is of interest at the moment.

For example one of the OCZ tech guys seem to think that the 3700 Gold rev 2 prefers 2.65 and users have gotten better results at a lower voltage not higher. Just to add one more little spice into the boiling pot of rumour and speculation that is overclocking.

Also thanks for all the help so far Edward2 and the others, I may not get a great overclock but I am learning a lot and I will figure out what my limiting factor(s) is.