Posted: Sat Apr 12, 2003 1:57 am
ABit KX7-333
1.1V to 1.8V in 0.025 steps
Happily running an Athlon 1800+ @ 1.60V
-John
1.1V to 1.8V in 0.025 steps
Happily running an Athlon 1800+ @ 1.60V
-John
Discussions about Silent Computing
https://www.silentpcreview.com/forums/
https://www.silentpcreview.com/forums/viewtopic.php?f=28&t=2872
http://www.soltekusa.com/product/showpr ... 1045186570To help you in your quest for the almighty MHz, Soltek has included what is known as RedStorm Overclocking. Simply put, this gives you a lot of power when it comes to settings within the BIOS. The AWARD BIOS has some decent overclocking options, but the additions made by the folks at Soltek make it even better. To give you an example of what I mean, you get the ability to adjust Vcore voltages from 1.1 – 1.85v in 0.025v increments, DDR voltage settings from 2.5 – 2.8v on 0.1v increments, AGP voltage settings of 1.5 – 1.8v in 0.1v increments and VDD voltages of 1.6 – 1.8v in 0.1 increments. You can also adjust FSB settings to upwards of 200MHz for those who really want to push their systems to the outer limits.
Ralf, that's amazing! I'm really excited to try this now... can I please ask you a few quick questions?Ralf Hutter wrote: I'm running my PIII-S 1.4Ghz at 1.23V right now, 100% "32hrs ofPrime95" stable. Default for it is 1.45V. I haven't tried any lower yet, I kinda have a feeling that more than 15% undervolting at default FSB would be pushing things too much. At this (1.23V) Vcore the Prime95 load temp is only 3°C hotter than the idle temp!
Still running at 1.23Vcore. Temps are real cool so I haven't felt the need to push the envelope. I've used about four other PIII-S 1.4 CPUs and their lowest stable Vcore seems to range around 1.25-1.325V, that's why I don't want to push this any lower.al bundy wrote:Ralf, that's amazing! I'm really excited to try this now... can I please ask you a few quick questions?Ralf Hutter wrote: I'm running my PIII-S 1.4Ghz at 1.23V right now, 100% "32hrs ofPrime95" stable. Default for it is 1.45V. I haven't tried any lower yet, I kinda have a feeling that more than 15% undervolting at default FSB would be pushing things too much. At this (1.23V) Vcore the Prime95 load temp is only 3°C hotter than the idle temp!
Are you still running your 1.4Ghz rig at 1.23V, or have you tried reducing the voltage still further? (I'm really hoping that you tried to push it to a voltage even lower than 1.23V - wow!)
What are your temps at this undervolt, both idle and under load? Also, are you getting these results using the SLK800 and a 5V Panaflo FBA08A12L, or with some other cooling solution?
The max Intel temp spec for the PIII-S 1.4 is 69C I believe, so you have plenty of temp room to work with there... which leads me to wonder this next question:Ralf Hutter wrote:Still running at 1.23Vcore. Temps are real cool so I haven't felt the need to push the envelope. I've used about four other PIII-S 1.4 CPUs and their lowest stable Vcore seems to range around 1.25-1.325V, that's why I don't want to push this any lower.
CPU temps vary with ambient temps (of course!) and are about 5-6°C above the case temp at idle. Full load temps are about 3-4°C above idle temps. My case temps vary from high 20°C in the early morning to mid 30°C in the late afternon. CPU idle temps run from 35-40° during a typical day and load temps are proportionally higher.
These temps are with the SLK-800 + 7-volted M1A as listed in my sig.
I think that your motherboard reads the temp from the internal diode, right? Because I'm getting even lower temps...Ralf Hutter wrote:CPU temps vary with ambient temps (of course!) and are about 5-6°C above the case temp at idle. Full load temps are about 3-4°C above idle temps. My case temps vary from high 20°C in the early morning to mid 30°C in the late afternon. CPU idle temps run from 35-40° during a typical day and load temps are proportionally higher.
These temps are with the SLK-800 + 7-volted M1A as listed in my sig.
This shows the quite high leak current of 130nm Pentiums. Are you able to run it passive at idle? My Palomino is able to do it with a Alpha 8045.Ralf Hutter wrote: CPU temps vary with ambient temps (of course!) and are about 5-6°C above the case temp at idle. Full load temps are about 3-4°C above idle temps.
I'm sure it would run OK with a 5V L1A but I haven't tried it. I can't hear the 7V M1A in the enviornmemt that I'm in anyway so I'm not in any hurry to try and make this any quieter.al bundy wrote:The max Intel temp spec for the PIII-S 1.4 is 69C I believe, so you have plenty of temp room to work with there... which leads me to wonder this next question:Ralf Hutter wrote:Still running at 1.23Vcore. Temps are real cool so I haven't felt the need to push the envelope. I've used about four other PIII-S 1.4 CPUs and their lowest stable Vcore seems to range around 1.25-1.325V, that's why I don't want to push this any lower.
CPU temps vary with ambient temps (of course!) and are about 5-6°C above the case temp at idle. Full load temps are about 3-4°C above idle temps. My case temps vary from high 20°C in the early morning to mid 30°C in the late afternon. CPU idle temps run from 35-40° during a typical day and load temps are proportionally higher.
These temps are with the SLK-800 + 7-volted M1A as listed in my sig.
With those excellent undervolted temps, do you think you might actually be able to instead use an L1A at around 5V on the cpu after all? Now that would be very silent indeed... What do you think, and have you tried that?
Well Ralf, look what you've done... now I just have to try it out!Ralf Hutter wrote:I'm sure it would run OK with a 5V L1A but I haven't tried it...
Just be careful down at 1.3v. Like I said, that's the low range of what most PIII 1.4's will run stable. If you are going to try low Vcores you should test your stability just like if you were overclocking (running low Vcore is kind of like OCing, you're trying to find the max stability at a certain Vcore, in this case a low Vcore) by running Prime95 for 12+ hours. If it doen't run Prime95 100% stable you should increase the Vcore by .025V and try again.al bundy wrote:I'm going to buy an undervoltable Tualatin motherboard to use with my PIII-S 1.4, and set vcore to 1.3V or less.
Please do feel free to share any further experience you've had doing this, that you think could help me succeed with this project... and wish me luck!
I'd do something like 1.4v under Prime95 for an hour or so, then try 1.35-1.375V for an hour until Prime 95 finally errors out. Then boost your Vcore by .025-.5v and try again for 5 hours or so. The reason I say to give it at least 12+ hours is that I've had it fail after 8-10 hours but never after 12+. I usually run a new system (or any clocking/Vcore mods) for 24-48 hours before I declare it stable. I may go overboard but I place a very high value on stability. It's just not worth it to tweak the last few Mhz out of a CPU or try and undervolt to save a degree or two if you run the risk of corrupting your data.al bundy wrote:Thanks for the advice Ralf,
OK, perhaps as low as 1.3V might be a bit too ambitious to try out first.
Perhaps I should instead start out at some mild undervolt level, test prime95 for many hours (wow, is 12+ full hours really necessary?), and then lowering the voltage a little more if I receive stable results...
Do you agree with this strategy, and if so, what initial undervoltage level do you recommend that I should start with please? Also, what readings or behaviour will I see with prime95, that will indicate when I have undervolted too much?
Nope on the SilentTek. You can monitor temps and speed, but the Windows software adjustment feature is not yet (as of this post date & and it looks, from their forums that it has been a know problem for many months) developed by their engineers, despite the SilentTek utility available for download for this mobo from the AOpen website AND SilentTek listed in the product description. Wish I knew before I bought. Hope this helps others. FWIW, it might be good to check other AOpen boards carefully (ie AOpen eforums and readme files) prior to purchase just in case.johnc wrote:There are many nice options.
If your friend wants a P4 with on-board VGA, sound and LAN, the AOpen AX4GPRO would be a good choice.
It offers continuous cpu core voltage control from 1.10 to 1.85 volts in 0.025 increments AND SilentTek. $99, delivered, from NewEgg.
Add a 1.7 GHz P4 for $54 (delivered, from Newegg), and a rather powerful, inexpensive and quiet machine is well on its way.
John Coyle
nan0dog wrote:Keep us posted Dan. I see you have a 1700+. Is this the 1.5v version. I have heard that the 1.5v version can run stable @ 1.1v. with no underclocking. I was going to get an abit nf2 board as these appear to offer 1.1 vcore but I think i'll wait and see how you get on. What I would really like is a micro-atx solution.
Oh thanks for the link. Downloading now.nan0dog wrote:
I wonder will shuttle ever sell the motherboard on it's own ( FN45 ). I love the board but i'd prefer low profile case like the pundit or smaller.
I see Soltek are selling a new cube, the EQ3701 with two 5" bays This might be undervoltable as their nforce2 atx boards undervolt to 1.1v. I can't download the manual at the moment.
From the AOpen Forums:britannica wrote:The AOpen MK79G-N is a possiblity. It does come with Ezclock which allows adustment of cpu clock and FSB from within Windows, but Vcore has to be adjusted in BIOS. Worth posing the question on the AOpen eforum Motherboard section.Zergling wrote:Any undervoltable microATX board?
Also has SilentTek fan control software.
ß
Which is sad because I thought maybe this may would actually work. Is there any reason why I can't find a microATX board that undervolts? Is it that there isn't "room" on the board for all functionality? Or is it just that they manufacturers don't see this as a big enough market to spend the time on? Frankly I don't see why ATX is still the biggest market. How many people use 5 PCI slots?The manual for this board claims the Vcore is adjustable from 1.1 to 1.85Vbut there is no option in the CMOS set up for Vcore ajustment.