805 at 3.8GHz on Thermaltake Bigwater
- Thread starter Pandobear
- Start date
Pandobear said:
Blue078 on XS said:
I presume this means the same pencil mod I done on my P5ND2 will also work as they use the same chip.
http://www.xtremesystems.org/forums/showthread.php?t=94659
This thread explains the pencil mod, good luck.
Arthalen said:
I got throttling at 70c so I disabled it and it kept on rising so I thought better not let it carry on and cancelled the test. Is 80c still ok ?
I noticed the north and south bridge heatsinks get hot enough to burn my finger when touched, so I put small fans on the north and an old 586 heatsink on the south
I think the voltage in the bios goes all the way to 1.8v or so, whats the limit there ?
Edit:
I did the mod pretty roughly, I couldnt get the pencil lead fine enough to actually fit inbetween the tiny legs so I taped it in place with sellotape. 0.5mm lead is too big :/
There does seem some improvement but still have a drop of 0.05v under load.
full pass on s+m at stock speeds and voltage which is now showing as 1.39v
Last edited:
According to the manual my boad limit on Vcore is 1.7, but i'd need a vapochill before even thinking of going that high lol.
On side note, I'm just upping my overclock. Specced it up another 5FSB to 3.7Ghz now that I've done the Vdroop mod.
Voltage is still cream smooth stable, 1.225 Volts set in BIOS, 1.32 volts reported by Speedfan. This is doing 32M SuperPi at the moment without issues, ran through 1M no problems. I'm about to full load test after the 32M Super Pi to see how things run.... *grits teeth against the temperature rises*
I think the thing about this mod I love the most is the fact that under load the voltage actually rises slightly thanks to the lower resistance, which although it means more temps it does mean more stability.
Thank god for ghetto is all I can say
On side note, I'm just upping my overclock. Specced it up another 5FSB to 3.7Ghz now that I've done the Vdroop mod.
Voltage is still cream smooth stable, 1.225 Volts set in BIOS, 1.32 volts reported by Speedfan. This is doing 32M SuperPi at the moment without issues, ran through 1M no problems. I'm about to full load test after the 32M Super Pi to see how things run.... *grits teeth against the temperature rises*
I think the thing about this mod I love the most is the fact that under load the voltage actually rises slightly thanks to the lower resistance, which although it means more temps it does mean more stability.
Thank god for ghetto is all I can say
@obi:
I'm sorry to say mate, but I'd have another go at that pencil mod. That voltage seems very choppy to me, and the alarm bells ring when you said you still get a vdroop of 0.05 - this is exactly what was happening to me BEFORE I did the mod :/
Here's a speedfan piccy after the fact for me:
As you can see, a lot smoother...... 1.32 rock solid stable before I start testing, and then starts to peak UP a tiny bit (0.01, 0.02) while doing 32M SuperPi.
EDIT NOTE: Do not disable Throttling. Throttling is your thermal friend, your performance adversary. It tells you need better thermal management or to drop the voltage off a little and will keep your chip happier in the long run.
I'm sorry to say mate, but I'd have another go at that pencil mod. That voltage seems very choppy to me, and the alarm bells ring when you said you still get a vdroop of 0.05 - this is exactly what was happening to me BEFORE I did the mod :/
Here's a speedfan piccy after the fact for me:
As you can see, a lot smoother...... 1.32 rock solid stable before I start testing, and then starts to peak UP a tiny bit (0.01, 0.02) while doing 32M SuperPi.
EDIT NOTE: Do not disable Throttling. Throttling is your thermal friend, your performance adversary. It tells you need better thermal management or to drop the voltage off a little and will keep your chip happier in the long run.
To be fair the chart does look a lot better under normal stress, 2 copies of sp2004 looked smoother. Superpi only tests 1 core at a time, so its more a speed test then stress and I use winrar benchmark to test multi-thread speed.
I will probably give it another go but I'll have to take the motherboard out and get my eye 1 inch away as its so miniscule.
Also my heatsink doesnt really cut it despite being almost a kilograms worth of copper, I need one of those heatpipe jobbies
I will probably give it another go but I'll have to take the motherboard out and get my eye 1 inch away as its so miniscule.
Also my heatsink doesnt really cut it despite being almost a kilograms worth of copper, I need one of those heatpipe jobbies
@Raikiri: Thats real weird - Mine doesn't throttle until 70C .... but then youvce read my old thread about the lies Speedfan II reports about my stackcooled motherboard 
Well, the good news for me is 32M Pi just passed at 3.7Ghz, so I'm starting the Dual Stress Prime of Death and destruction tests now *sigh*
Well, the good news for me is 32M Pi just passed at 3.7Ghz, so I'm starting the Dual Stress Prime of Death and destruction tests now *sigh*
Well its under way, looks like I can add another 100Mhz to my overclock as temps seem stable at around 67C.
Voltage stability is rock solid at 1.340 - 1.345. Much goodness.
*Keeping fingers crossed Prime passes a good few hours of testing *
Voltage stability is rock solid at 1.340 - 1.345. Much goodness.
*Keeping fingers crossed Prime passes a good few hours of testing *
I have just done the pencil volt mod, could not have been easier. The oddest thing is that now it actually runs 3.8 GHz at stock voltages, which are obviously 0.1v higher than set, but this is a full 0.15 under where I had to have it before. Runs cooler now also, obviously because of the lower vcore. What I don't understand is what is the downside of this mod? There must be one otherwise surely Asus would do it as standard? Anyways, thanks guys vcore does not move more than 0.02 now, and only upwards! woohoo
Pandobear said:
Just fantastic isn't it ??? Vdroop seems to be limiting factor, do this mod , and Im stable at 3.7 now with less volts.
I will probably go to 3.8 tomorrow night.. or maybe tonight.... ahhh fubar it im gonna go for it... Prime stable and no throttling in sight here
Last update before I hit bed - running fine at 3.8Ghz on Full Dual Prime load - temps same as what they were at 3.7, no voltage change necessary (yet lol), voltage is super stable for me
Gonna head off to bed.
Ooh, look at that, my 100th Post too! Yay me (lol).
Gonna head off to bed.
Ooh, look at that, my 100th Post too! Yay me (lol).
Really excellant processor for the money. I was going to change for a Conroe as soon as they are available, but I think that I will hold off for a little while, wait and see which one will really give an improvement over the 805. Also not many of the new chipset boards have many sata ports. I really need a minimum of six internal, but most boards seem to be going for the esata, not sure why, can't see much use for it. The P5LD2 has the four intel sata which I use for a four drive RAID 5, and then I have to use a 3112 Silicon Image PCI card for the two raptors as the board has only one more internal sata, and it is a real mess to have a esata to sata cable running from the outside of the case back in.
One other interesting thing with the overclocking is that at 180 FSB I lose the intel raid controller, comes back at 190FSB. Doesn't matter whether I lock the PCI and PCIe buses or not, weird.
My 2nd attempt was no good, I thought it worked at first because I snapped the very thin tip of the pencil off nicely between the chip legs however I must have been a bit liberal with the graphite as it was really unstable afterwards.
I even removed the pencil lead and it didnt stop the freezing, so I had to use a rubber on the mainboard to remove any traces of graphite which seems to have worked now. No damage done phew
S+M is stable at 100% but shows a deviation of .03v at stock speeds. I'll try again later I guess
I even removed the pencil lead and it didnt stop the freezing, so I had to use a rubber on the mainboard to remove any traces of graphite which seems to have worked now. No damage done phew
S+M is stable at 100% but shows a deviation of .03v at stock speeds. I'll try again later I guess