Shuttle AK31V20 – Budget DDR Upgrade

A cheap DDR upgrade, and what it gets you.–John R. Abaray

Finally finished this. Life is full of little problems. I wanted to compare my A7V133 to the AK31V20 and see how much bang for your buck you could get. A shipping problem cost me a week. I had to go back to DDR memory school. After memory school, I had to wait for a different stick of RAM to see if what I had learned was correct. This turned into another article because what I had learned needed to be passed on sooner rather than later.

Then I wanted to use the board everyday for a couple of weeks to make sure there were no more surprises. The severe hard drive corruption with the 1T command (S205 bios) I experienced after crashes at high FSB speeds left me a little gun shy. I’m starting to see a hint here and there this can happen with other DDR boards if you don’t understand advanced memory timings.

It could be I’ve just received one to many emails from a disappointed reader with a sad story. I feel the industry has too much influence on hardware sites. It’s no secret how I feel about some, but not all reviews.

If you rely on the Mobo Makers to send you free boards to test, they have influence on you. If someone tells me it’s not free because they send it back, I find that a bit ludicrous. What do you call a board you don’t have buy to test? My dictionary says free is “not costing or charging anything.” Do you really think if a reviewer points out problems and flaws they will continue to get free boards to test? If anyone reading this is naive enough to believe otherwise, I have a bridge I want to sell you cheap. You can charge tolls and never have to work again.

I am an independent author. I buy everything I write about so I can speak my mind freely with a clear conscience. I don’t work for anyone or get paid anything by anybody with this exception: I work for the aspiring performance enthusiast and those building their first computer. Last but not least, those short on cash looking to get the most for their money.

Everything I write or say is public domain and it may be used anyway the reader wishes. You don’t have to ask my permission for anything. If there is a secret set of unwritten rules between hardware sites as an independent author, I am not bound by them. Anything I write, including emails, can be posted anywhere, anytime. I’m not ashamed of anything I say, nor do I have anything to hide.

It’s been said I’m holier than thou – I ‘ve seen it in print. While this is quite flattering, I don’t believe I’m holier than thou. I might be holier than some. It’s been inferred I live in a glass house. Quite to the contrary as an independent buying my own gear, I live in a steel reinforced concrete bunker.

In view of a recent minor controversy, I felt it necessary to make a position statement. For whatever reason, there are some who assumed I work for OverClockers.com. You should never assume anything. It could be your “Dumb thing of the moment.” I’m sure this will clear things up. From now on, there will be a disclaimer along with all my articles.

I won’t bore you with pages of technical specs – you can find them easily enough if you want them. Instead, I’ll tell you what I like about the board:

Six PCI slots and four DIMM slots – enough to satisfy anyone.
A soft BIOS with all the tweaks you could ask for, with one exception – no VIO adjustment. However, you can boost the RAM voltage to 2.70v. Overclocking options are as good as it gets. No voltage mods needed (except for the ultra hardcore) on this board.

If you’re a performance enthusiast, you can get plus 0.275v over Vcore default. If you connect the L7 bridges on a Duron, default will be 1.85vcore. Add the plus 0.275 and you can squeeze the absolute last Mhz out.

The onboard sound is usable but not enough to satisfy an audiophile. The Northbridge hsf is a deluxe version compared to the one on my A7V133 – a flat plate with nothing underneath it. This one has some fins and thermal paste underneath it. Only one problem: there is a capacitor along side it, so no Blorb.

The Clear CMOS is a jumper located where you can easily use it and it works. This is great compared to the solder short points on the A7V133 which initially didn’t always work.

No it doesn’t have RAID. The question is what does RAID do for you?

Does it make your system boot any faster? No.

Does it make you games play better? No.

So what does RAID do? If you’re loading really big files, it will do it faster.

It makes your mobo more expensive,

It will give you great HD benchmarks.

You can mirror a second hard drive. Just what I need – two screwed up hard drives when I crash while maxing out the board.

You can have bragging rights at your favorite forum and say “I’ve got RAID”.

No I am not a big fan of RAID, though it does have a few benefits for some users. There was a lot of hype about RAID and many were lead to believe it was going to give them something special. With a few exceptions, the average user isn’t going to see any difference.

Getting the board up and running – no problem. Using the board at high FSB speeds with aggressive memory settings – big problems.

If you haven’t read the first part of this article on what the reviewers don’t tell you, I suggest you do it now.

After building computers for six years, I don’t expect the big whoosh anymore when I upgrade to the latest advance in technology. However, this board did give me a little swish. I decided to compare it to my A7V133 to give you an idea of what you can get with a budget DDR board. I used the same CPU and video card.

I don’t use business app benchmarks. Remember, these articles are geared for the “New Guys”. If I wanted a business machine, it certainly wouldn’t be an overclocked DDR board. I use SiSoft Sandra, 3Dmark, and Quake timedemo. Simple but useful and free. If I can run these without any problems, then the board is doing what I want it to do.

I compared the boards at the same FSB as my A7V133 and also the practical (?) max out for the AK31V20.

The AK31V20 shows a twenty five percent ALU increase at the same FSB over the A7V133; over a thirty percent increase in FPU. 155 FSB shows a small increase over 145 FSB. This is from a board in the ninety dollar(US) price range with a lowly Duron.

I really thought I would get another five or six fps. 155 FSB showed a miniscule increase in normal – not worth mentioning. A little disappointment, but it didn’t ruin my day.

Again as with memory, we see a significant increase over the A7V133. And again, very little increase in performance from 145 FSB to 155 FSB.

Thanks to a good reader response, I found some are now getting S907 BIOS with their board, some are getting S20B. I tried it and ALU/FPU scores are lower than S907. Thanks to DaveB who informed me he got his board with S906 – he said he could also get memory scores in the 700/900 range, the only difference being the extra memory commands. Mine came with S205.

I also found some got a different BIOS when requested and others got no reply, like myself. I always get complaints users often get no reply from mobo makers’ tech service . This is nothing new, you just have to deal with it.

If you are a seasoned overclocker, you will really like this BIOS. If you are new to the game, it has everything you need to thoroughly screw up your system. If you like to make a change in BIOS to see what happens, you may regret it.

It’s difficult to write anything up to date about computers. What’s true today may not be true tomorrow. What’s up to date at the time can be ancient history in a few weeks, or even a few days. Thanks to all for their emails.

Simple benchmarks can be very useful tools and they can also lead you astray, as I was with aggressive memory commands. I could benchmark with SiSoft Sandra and even play Quake by upping the AGP drive strength from DA to EA. 3DMARK would crash every time, leaving me with a corrupted hard drive.

On the other side, by looping 3DMARK, I could find the maximum FSB for stability. At 158FSB, I could run it once and then crash. At 155FSB, if I got through the second one, it would crash on the third – usually in high polygon count eight lights. Stable enough to use for some gaming and surfing, but I wouldn’t keep my checkbook records on it.

Since there was so little difference in scores between 145FSB and 155FSB, I settled with 145FSB. I used the board everyday for two weeks doing 3DGaming and surfing with no signs of instability. A 750 Duron@ 1015Mhz (145 x 7.0). I did manage to get ALU/FPU scores as high as 746/968 – nice eye candy to look at but totally useless.

I lost some emails. I might as well fess up: The board seemed stable at 155FSB because I had been using it for three days. Then I remembered I hadn’t looped 3DMARK and crashed. I forgot I had downloaded a few emails. Shame on me. I had been exchanging emails with someone, so if you’re reading this, I apologize for no reply.

I’ve already mentioned hard drive corruption with aggressive memory commands at high FSB speeds. I might also add, with a second hard drive connected, I also found it had some corrupted files.

If you don’t already know it, there is a problem with the KT266 chipset and USB devices over about 138 or 140 FSB. Word is it will work with a third party card. Keep this in mind. If you’re on a budget, I don’t think you’ll want to buy one. I tried a USB joystick at 145FSB and no go.

Rebooting when making changes in FSB speed up or down would sometimes produce a Windows registry error message and it would disappear on another reboot. There were times when making an FSB change would cause the dreaded black screen on reboot and I would have to Clear CMOS. If I just shut it down and let it sit for a few minutes, there were times I didn’t have to clear CMOS.

This board is a very good value for the money. At this time, it can compete with any of the expensive, big-name boards regardless of chipset. Judging from the response I got from potential AK31 owners, Shuttle is going to sell a lot of these boards.

I keep feeling I missed something. Maybe it’s just an old lab rat’s intuition. It could be some insignificant thing in BIOS that would affect performance. The feeling is there and I can’t shake it.

The motherboard makers are responding to our community as it continues to grow. The boards I am seeing now were beyond my wildest dreams in the old K6-2 days. The price to be paid is a thorough understanding of BIOS and memory commands.

Your board may do better or worse. I know they do vary, but what the range is would be an interesting project. With my R&D background, I figure about a thousand boards would cover random chance. Buy a hundred boards at a time with spaces in between. Possibilities and probabilities always fascinate me. Now all I have to do is win the lottery.

article, I said this:

“I keep feeling I missed something. Maybe it’s just an old lab rats intuition. It could be some insignificant thing in bios that would affect performance. The feeling is there and I can’t shake it.”

I believe intuition is real. The problem is your mind can play tricks on you, so intuition isn’t always reliable. Thanks to ICherub and others, I found out what it was. My intuition was correct and it was in BIOS.

This is the way it happened:

I used the same CPU as in the A7V133 to try and make all things as equal as possible. The first thing I did was to max out the 750 Duron. The L7 bridges were already connected and my default Vcore was 1.87. I was able to get up to 1027 MHz, so I felt no need to raise the voltage.

As usual, after an article I began to get emails. A number of them said using the plus Vcore setting they could not get above 1.85 Vcore. Anything above 1.85 Vcore would revert back to default. They also said they couldn’t get a definitive answer from Shuttle.

The first thing I did was to try it myself and couldn’t get plus 0.1 Vcore. The next thing I did was to try minus 0.1 Vcore and it worked. A light bulb went off over my head:

I needed a CPU with a low Vcore default. If you have never used defogger paint to connect bridges, I can tell you this: I would rather take a kick in the butt than wipe them clean. I was going to run down and buy another Duron when I remembered I had a stock Duron in my CPU junk box with 1.60 default Vcore.

I tried upping the Vcore a notch at a time and it works up to 1.85 Vcore but not 0.1 over. The minus works, but who cares? So it works, but not the way I thought it did. I’m big on cliches: “If it sounds to good to be true”. It was and shame on me for being in a hurry to get out an already delayed article and not checking it out. I should have listened to my intuition and sat on the article for a few more days until I figured out what it was I missed.

Now there is a price to be paid: Ignoring a problem or an error does not make it go away. The real test of integrity is admitting it. If I am going to maintain my integrity with the readers, I have to be honest.

The road to hell is paved with good intentions and I am no exception to the rule. Some users bought the board on early information, thinking they could get the plus Vcore over 1.85V. I have to admit it also influenced my decision to buy the board for an article.

ICherub asked me if I assumed it was possible because it was in BIOS? YES, It was in the BIOS and I didn’t try it, so it does qualify as an assumption anyway I look at it. A little self deprecation can be good for the soul. It was a “Dumb thing of the moment”.

I do not work for anyone or get paid for anything I write. Everything I write about I pay for out of my own pocket. Everything I write in articles or say in emails is Public Domain for the reader to do with as they wish.