Power supply is major key to overclocking 1.6A.

[WarCon]

I am currently running a 330watt Enermax (voltage modded) and it can supply 12A of 12V current. I just received my Antec 400watt supply that supplies 18A of 12V current. If must of been defective (sent it back already), because at default settings it drained the 12V rail, down to 11.31v (full load). It wouldn't even let me boot at 150 @ 1.75vcore (which I am able to do right now stable, but my 12v rail drops to 11.31 under full load and thats below spec so I tuned it back so the supply stays in spec.) It barely made it to the same speed I am running now (it actually didn't change its output much though, as it stayed at 11.31 under full load @ 132fsb 1.55vcore.

Anyway, I hope the next one I get is good or I am going to have a fit..............

I mean its 2.03spec and its an ATX12V standard, I really hope that isn't typical for that supply.

....................edit......................
Added plus for you P4B266 owners. Holding the Insert key while you start the machine after it won't boot from too high of settings does work for this mobo. It must start it at a slower speed. Haven't had to use the cmos jumper yet.............

 

[f13dfx]

What are your memory settings?

What are your idle and load temps?

TIA

 

[WarCon]

At 132, which I am running right now so that my 12v rail isn't going below spec, I am running spd timings but 3:4 CPU/Memory Ratio or its at 176 or really DDR352 speed. I am only at 1.525vcore (which actually reads right at 1.5vcore @ idle)

I can run stable all the way to 152fsb with 1:1 CPU/Memory Ratio @ 2 2 2 5 timing, but my 12v rail is running at 11.31v and thats too low for me. Thats why I ordered the Antec 400w, but the one I received was much worse than the Enermax I am currently running. It ran 11.31v on the 12v rail at default 100fsb @ full load. I couldn't boot at 150fsb (I can boot all the way to 160 now just not Prime95 stable).

I am hoping that it was just a bad supply or I will somehow need to try a different one. It was rated for 6 more amps on the 12v rail, so I really don't think it was ok. Might of just not been calibrated right, but wasn't going to void my warranty the first day I have something that I might need to send back. I just shipped it.

Oh and as far as temps go, I am sure they play absolutely no part in my overclock since I have never seen a temp over 40C no matter how high I have gone. I watercool and these processors must use a lot less power than an overclocked T-bird. Now for my T-bird, temp was one of the keys for higher speed, but not these Woodies. I think my 1ghz @ 1.4ghz was well over 100watts coming through a tiny little core and could hit 43C measured by a not so accurate socket thermistor. Most of the time it runs 28C now, at its current settings it can hit 33C full load. I used Nevin's (Arctic Silver) calibration method and it says I need to lower those readings by 3C, I don't because they look so low already.

My walkaway and come back idle temps are 25C (2C above the system). Ambient room temp around 19-21C.

 

[Poof]

WarCon - are you using MBM to get your readings? If so, it seems to me that something is very wrong other than the fact that with Asus boards, the CPU temp never goes below 40C in MBM and when I've gone into the BIOS to get the reading for the 12V+ rail at idle, it reads 11.77V - 11.84V (this is for an Antec 300W 303XP). Yet once in windoze and idle, it supposedly drops by almost 0.5V down to 11.33V according to MBM. Start up a load and it always drops to 11.19V for me.

Personally I think the algorithms coming out of the ASUS board to MBM are way off.

Only way to really know I guess is to slap a voltmeter on somewhere.

 

[jiffylube1024]

See my thread on this same topic.

 

[WarCon]

Honestly, with the troubles in bios in that section (temperature monitoring is a joke), I am tending to believe MBM. I can watch the 12V rail slowly diminish as I increase the FSB. This is expected behavior, unlike the readings my bios gives me. I can see in MBM the correlation between the vcore and the 12V rail. As the 12V gets lower and lower, the vcore follows suite. And this powersupply that I just had, that was much inferior to what I already have (am really hoping it was just a fluke), showed 11.97v in bios (which is typically not idle conditions), and idling at 11.45 at default in windows. It showed up in the vcore and ability to overclock.

I really wish they would address these issues with my mobo. The USB issue is a real pain, and the sensors being so screwed up. I just think they released the board before they finished the bios........... Now several months later there has only been one update on my C revision of the P4B266 and that only comes with the board, its not even available for download. I am not really complaining since most of the functionality is there and it overclocks great. I may have to study their voltage controller and see if there is some kind of mod available to boost vcore. I still want a 12V rail that will hold its own to the rated specs (+/- 5% right?).

Oh and I really am quite certain overclock ability on these is directly tied to voltage. I was running Prime95 at 155 (1.75vcore) and I pressed the button to open my DVD and right then it errored. So the stability was directly tied to the additional droop caused by the small amount of current used to open the tray. I think it probably made the vcore droop and forced an error. I am thinking a stability mod should be available for this board. Probably a small capacitor from the feedback line to ground to add some stability to the feedback circuit or something along that line. I don't think all of the Pentium 4 boards even have this issue though. If the circuit had been designed correctly, it shouldn't droop until the need exceeds the current capacity of the source. The vcore varies too wildly. I think the feedback loop has been somehow made dependant on the source voltage.

Maybe one of you very bright and industrious individuals out there could figure the voltage control circuit out and see if the feedback loop could be smoothed, or if it is simply 12V rail current issues.

Sorry for so long a post. Just kinda bummed, was hoping a 400watt Antec would solve my power needs, but either I got a bum supply (will know after the RMA returns) or I am going to have to live with 132fsb and high memory speed (for DDR memory).

 

[exp]

Can't you adjust the pots on that Antec?

 

[ICEFuZiOn]

Hmmm Interesting Read....

I got my 1.6A@133FSB Stable...Anything over will crash in 3dmark with P4B266-C. Im in Normal Mode..isnt normal keep the pci/AGP at spec? and Turbo1 37/78 or something like that?
Is it my PSU? Im using EZplug and have a Sparkle 300watt PSU, in MBM it looks like it says the +12.00 Rail is running at 11.37..is this Bad? im running the Default 1.500 (1.60 in MBM)VCORE. COuld it be my PSU that isnt allowing me to up FSB and get Stability? I can Boot into WinXP as far as 145FSB on 1.55Vcore..but it crashes within min's...

I have 256mb Crucial PC2100 Memory, would it be good to invest in Some Expensive Corsair PC2700?


Any Help would be gr8tly cool!

 

[JellyBaby]

WarCon, what gives you better benches, a 132 mhz fsb with the memory running at 176 mhz.....or a 150 mhz fsb with memory also running at that speed? I'm especially curious to know which gives you the best memory scores.

 

[WarCon]

When my 12v rail hits into the 11.30's is when I start having it fail Prime95 (it can be several hours into testing when it fails, but it still fails). I am running my Crucial DDR2100 256meg stick at 176 with spd settings. I could also run it at 152fsb with 2 2 2 5 timings. I think your memory is fine. You might try upping your vcore if your cooling is fine. I have a watercooler, so I feel a little better about using high vcores than I would with air-cooling. (Still a little nervous, til I see if these chips are going to start failing in 6 months from high vcore or not).

I really think that when the droop in the 12volt rail is too extreme, is where your going to be limited with your overclock. I realize that there are other limitations: Poor overclocking chips, bad mobos, memory issues and all sorts of misc problems caused by peripheral's and their issues with higher than average PCI/AGP bus speeds.

I believe your assuration about Normal mode and Turbo mode to be correct. I am running Turbo1 without issue.

exp - I wasn't really wanting to open the supply that I just received and tweak it. I thought a supply with 18A (my current Enermax is only 12A) of 12volt current shouldn't of been drooping like it did. I figured that if I was going to tweak the supply, it would have to be a good one to start with...... (I have an uncommon sense of right and wrong. I figure if I use something beyond its clearly stated user agreement, then its mine whether its broke or working. I don't return stuff that "I" break). The supply was drooping badly at default speed, so I tried to overclock but had little success. My current supply performs much better.

ICE- If your setup is doing what mine is then it is very possibly your supply. BTW, don't sparkles have pots too? I am not sure, but I think I remember hearing someone say they do.

 

[WarCon]

Jellybaby, actually I get better CPU related scores from the 150fsb (which is to be expected). But my memory scores really climb with the 132fsb. At 152 (my highest stable point - so far ), my 3DMark is 8239. At 132 (with memory pumped) its 8046.

At 150 (memory 2 2 2 5 1:1), my PCMarks are:
CPU: 5772
Memory: 5086
HD: 1025

At 132 (memory SPD 3:4), my PCMarks are:
CPU: 5107
Memory: 5129
HD: 1019

Sisofts CPU at 152 - 4619/2977
Memory - 2285/2185

Sisofts CPU at 132 - 4012/2587
Memory - 2658/2532

As far as system feel (can't measure it), I am not so sure it isn't smoother at 132. All my games play really well here and at 152, so thats hard to say for sure.

Hope the data helps...........

 

[Uclagamer_99]

hopefully asus will update hte p4b266-c bios sometime in the near future

 

[WarCon]

You would think so, heck the regular one is at 1006 right?

..................edit..............
Sorry, the official is 1005, and there is a beta 1.005.004.

 

[Poof]

Right now, I'm running SETI on mine @150MHz bus and 1.55V Vcore. I'll be out of town for a few days so I'll let it cook at that. I was able to POST and do stuff in windoze @160MHz bus but running SETI would blue screen VNC.

This is all with the retail HS & Arctic Silver 3 (plus caseless).

I still think that MBM with the -c mobo is crap. Yes, I'll see movement in the Voltages with MBM and I'll see movement upwards above 40C when at load, but there is no way that a caseless CPU at idle and not OC'd, is at 40C in my room (where I try to keep the ambient at 15C - 18C). Plus I have CompuNurse temp probes on my OC'd XP & T-bird and @idle, they're running ~26C - 30C.

 

[RalfHutter]

Poof - Are you using the "Asus 2" sensor in MBM as your CPU temp sensor? It is the one to use. The "Asus 2 (CUSL2)" sensor adds around +10-14C temp compensation for some other boards.

 

[Duvie]

maybe it is the aus p4b266 boards...cause voltages run similar in bios and windows with asus probe on my sis...I believe them...idle my 12v does 11.9 st 1.7v and less...at 1.775v it does 11.84 at idle...at full load it does 11.64v avg. mbm confirmed this the short time I had the program installed...


quit bitch'n about the temps...everybody has seemed to figure it out that the temps at idle are wrong because the probe stops at 40c...but at full load the temps will climb to accurate temps...mbm will monitor the same sensor yu guys...bios at idle 38c asus probe 40c....big deal???

 

[prontospyder]

Poof, as others have stated, you need to use the Asus 2 sensor on MBM5

I have a P4B266-C too and the temps range from 20C to 43C in MBM5.

 

[JellyBaby]

<< Hope the data helps >>

Yep, sure did, thanks WarCon. I'm creating a rig very similar to yours and wasn't sure which route to take. If my 1.6A will handle a 150+ fsb that's what I'll go for.

 

[ICEFuZiOn]

I just bought a Enermax 350watt EG365P-VE FMA PSU..this hopefully will help my Stability problems....I hope so
Its +12 Rail is MAX 20A and my Current Sparkle 300watt only provides 11A MAX...

Hopefully this Enermax will be worth the cash cow it is....

 

[WarCon]

20A should be sufficient (dang well better be because thats 240watts of power)...........

 

[F155mph]

Warcon,

I totally agree with you. I was using a Powermax 500w PSU and I can't get it to run stable at 149fsb. So I replace the Psu with an Antec 400w and now I can run stable at 149. But when I touch anything while Prime95 is running I get errors.

 

[jiffylube1024]

<< I just bought a Enermax 350watt EG365P-VE FMA PSU..this hopefully will help my Stability problems....I hope so
Its +12 Rail is MAX 20A and my Current Sparkle 300watt only provides 11A MAX...

Hopefully this Enermax will be worth the cash cow it is.... >>



I just got the same PS and so far it's a lot better than my old 300W PS (which had 8A max on the +12 V line .. what were they thinking!?). Actually I got the one without fan speed control for cheaper (exact same specs minus fan control).