ABit KX7-333

1.1V to 1.8V in 0.025 steps

Happily running an Athlon 1800+ @ 1.60V

-John

MikeC, the Soltek DRV4 has CPU voltage in the range 1.10 - 1.85V.
Hope that helps.

Any undervoltable microATX board?

According to Tweaktown it's undervoltable. Voltage adjustable from 1.1 to 1.85



http://www.soltekusa.com/product/showpr ... 1045186570

http://www.tweaktown.com/document.php?d ... 47&dPage=8

The Epox 8k9a2+ is undervoltable.

Does 1.400 - 1.850 in 0.025v increments
and 1.850 - 2.100 in 0.050v increments

(Currently running my xp2400+ at 1.55v on this board).

Ralf, that's amazing! I'm really excited to try this now... can I please ask you a few quick questions?

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?

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.

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:

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?

I think that your motherboard reads the temp from the internal diode, right? Because I'm getting even lower temps...

I have 1.3 GHz Celeron II, with ABit BX-6 rev.2 (slot-1). I use PowerLeap's adapter. My board doesn't support reading directly from the diode (it would work if someone soldered few SMD resistors to the board), but from external sensor. I have this sensor under the retail heat sink (using the retail fan too).

Temps are 35-37°C, depending on the ambient temp and there is only a small idle/load variation; (tried playing some games, encoding MP3s and Sandra's processor burn-in tests) only 1-2°C. I haven't (yet) undervolted the processor, the adapter has settings down to 1.1V (stock is 1.5V). SpeedFan allows me to control one fan, if I throttle down the CPU fan to 65% (at which point it's inaudible compared to the PSU), I get temps around 38-40°C.

I have an older version of Enlight 7237, with 235W PSU. No extra fans (I'd like one in the back, but there is no mount for it (yet... )). Do You have any extra fans in your case?

I'm going to undervolt when I get more free time...

Cheers,

Jan

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.

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.

Currently the only time I can (even barely) hear anything from this system is around 3-5AM when things are real quiet around here. After that the ambient noise level of people's daily activities far exceeds the very low noise output of this system.

Well Ralf, look what you've done... now I just have to try it out!

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. I'll attempt to cool the processor with an SLK800A and a 5V Panaflo L1A. I'll retire the TUSL2-C for the time being. This is kinda exciting for a 'silent PC freak' like myself!

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!

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.

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?

Just a sidenote to testing stability with prime95... In some cases you might find that your machine will run 12+ hours of prime95 fine, but then crash in other applications. Or, prime95 might crash within 10 minutes, but then your other applications can run faultlessly.

So even though prime95 can give a good indication of stability, it isn't the be all and end all. Remember to test it out in some other capacity too (I'm thinking of games myself).

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.

I've been OCing for a long time now and have always relied on Prime95 and Memtest86 to determine stability and have NEVER been bit yet (knock wood). If the box I'm working on happens to be a gaming box I'll usually run some looped 3dmark too.

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.

Just ordered this very undervoltable board. Take a good look at the BIOS screen shots near the bottom. Lots of clearance at the CPU socket. As many BIOS options as you could ask for since it's an overclocker's board.

http://www.lanaddict.com/review.php?ID=67



Dan

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.

It runs at (default) 1.6 volts on my current board. Palominos run at 1.7 I believe. Many Soltek boards have a wide voltage range both under and over the default voltage. Usually when you read a review they only discuss how high the voltage can be goosed because most boards widen the range on the high side and neglect the low side. Not so with many Soltek boards.

I ordered a new power supply since it has that auxiliary P4 power connector so it may take a week for my under volt report. Auxiliary P4 power connector is supposed to make for a more stable board. I got mine here on discount: http://www.newegg.com/app/listProduct.a ... 11&order=0

Abit nf2 board..... I don't know the voltage range. You should go to NewEgg and check out the customer comments on it, other nforce and on the Soltek I got. Here's a shortcut: http://www.newegg.com/app/viewproduct.a ... on=nForce2





Dan

At least somebody in the forums at Sudhian (http://forums.sudhian.com/messageview.cfm?catid=43&threadid=36454 found this in their manual. The manual for the new Athlon model (SN45G) doesn't seem to be online. Does anybody know if this is undervoltable too?

Woohoo! Undervoltable mainboards are the only thing that can make an SFF really quiet (unless you go with a VIA CPU).

Bye egghat

The manual is online now @ http://www.shuttle.com/new/support/down ... odel=SN45G

I'll put you out of your misery. 1.1v to 1.0 vcore Excellent.
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.

Oh thanks for the link. Downloading now.

The Pundit is somewhat cool, but nonstandard. I'll guess we'll have to live with the fact, that the really small computers will be bundled with a case and a PSU.

I'll check out the Soltek one. Have you seen a review somewhere on the net with noise measurements?

bye egghat.

Don't know if it's on the list...

MSI K7D dual socket A with MPX chipset. From 1.1 -> 1.85v, but it undervolts by 0.4-0.5 resulting in 1.05-1.8v.

Speaking of MSI... I've heard only positive things about their new 865 Neo2 boards. But does anyone know if they are good for undervolting too?

Abit NF7-S v2.0 is CPU undervoltable. In 0.025V increments, if I recall from 1.1V upwards. Currently we have two NF7-S boards running AMD T-BredB 1700+:s at 1.30V (1.6V default) with a slight 5% overclock. Both CPUs are stable through Prime95 and pass memtest86. Note: I do not know whether this degree of undervolting is normal with 1700+ processors though; it is quite possible that we have gotten very lucky.

Edit: we're just breaking these processors in and will experiment more with them soon, including undervolting further while increasing clock speed.

Btw every cheap motherboard support undervolting and overclocking!
You just have to unlock the processor and do some wireing to the socket

here http://www.overclockers.ru/news/newsite ... 1050266349
and somewhere here http://www.ocinside.de/index_d.html

From the AOpen Forums:

http://club.aopen.com.tw/forum/viewmessage.asp?forumID=17&MessageID=41861



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 biostar m7ncg is a nforce2 board that will undervolt to 1.3v I believe.
You should make sure that you get revision 1.4 as 1.2 seems to have a lot of problems. I hope Abit make an nforce2 matx board as all there boards seem to undervolt to 1.1v.