Maxing Out The A7V133 – Getting The Bugs Out

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Have you ever heard “The players may change but the game is the same”? There are a lot more players now but the game is the same. Lurking in the forums, I often find a post cursing one board or another. For some reason, everyone expects a new board to work flawlessly. This has never been my experience. Go to the various Mobo manufacturers sites and dig around in the archives. You will find numerous board revisions and BIOS updates.

Going back to my K6-2 days, I still remember the Ali 5 chipset and my GA-5AX board. While Ali finally came up with a AGP driver that worked, the GA-5AX AGP slot didn’t supply enough juice for a high end video card. Some users even had their boards die on them trying to get it to work. A few got lucky and it worked (All motherboards are not created equal). I spent endless hours on that one. I wound up using a Voodoo 2 PCI card (Sigh).

There were five versions of the board. There were twelve BIOS releases including the original. That was then and this is now. Same game, just a lot of new players.

If you’re a seasoned Overclocker this is not an article for you.

You won’t find anything new here. This is an article for newcomers to our community – “New Guys” as I like to call them. I don’t use the word Newbies, as I have read the references in forums to “Dumb Newbies”. It is entirely possible the Dumb Newbies may actually be more intelligent than the poster; they just haven’t gone through the learning curve we all have to go through.

With new DDR boards coming out all the time, the price of the A7V133WOA is dropping to the point where it is close to being a value board. If you’re looking to recycle some old components or build a new system, this could be a consideration. I’ve also included some simple benchmarks to give you an idea of what kind of increase in performance you can expect.

The system as it stands:

  • A7V133WOA Rev 1.04
  • BIOS 1005
  • Via 4.31 drivers
  • OS Win4-10-98
  • Unlocked Duron 750
  • Vantec FCE 62540D W/Arctic Silver
  • 3d Prophet 2 MX
  • 3 128 MB sticks Corsair PC150
  • 2 Maxtor 15 gig 7200 rpm ATA100
  • Ancient Yamaha 724 sound card (I like Aureal 3d)

Getting the system up and running was uneventful. The only things I had to do was change the setting from optimal to normal while waiting for my PC150 and clear the CMOS once. I ran the system for three weeks with a slight FSB overclock with the original 1001c BIOS – I wanted to make sure everything was working before voiding any warranties.

The system ran flawlessly while the forums were full of rants. Initially, I wrote it off to inexperience and a few bad boards. As the din grew louder, I began to suspect there were some real problems. But, my board was running just fine.

What’s true today may not be true tomorrow. It wasn’t too long ago I was telling people if they wanted to do some overclocking, get a Duron because T-birds were poor overclockers. Then along came the 1 Gig T-bird AXIA and suddenly OCers were getting unheard of forty percent overclocks with air cooled T-bird systems. That’s about how quick my world changed when I unlocked my Duron and began using the Soft Menu for maxing out the system.

In an earlier article I said:

“If you want to be on the leading edge of technology or close to it, your going to be a Beta Tester for the Mobo manufacturers. Accept this and learn to deal with it.”

Truer words were never spoken. I didn’t have the sound card problem (which a patch has since fixed) because I was using the old Yamaha. What I did get was a lot of the dreaded black screens on reboots. It was getting to be a real pain clearing the CMOS with the two solder short points.

I was never sure I had the screwdriver squarely on both of them. Occasionally, clearing the CMOS didn’t work. My system is set up on one switch which isolates it from the AC line when I turn it off. I would just wait patiently (not really) about ten minutes and it would work when I turned it back on.

I got tired of screwing with the Soft Menu and went to jumpers and dip switches while waiting for a new BIOS to correct the problem. My Duron was happy at 1 GHz 133 x 7.5. Then one day something strange happened – bizarre is the only word to describe how I cured it. I was going to do a little surfing before going to sleep, and when I turned on the box all I got was a black screen. The fans were running but that was it. “Oh well I’ll fix it tomorrow” I thought.

Morning came and I couldn’t get it going. I thought the board had died on me. I was about to run down and get a KK266 when, for some reason, I tried going back to Soft Menu and the board came to life. I still don’t know why I did it except maybe the Mobo god put the thought in my head. Was this a chipset problem or a BIOS problem? Was it a one time fluke?

I left it on Soft Menu with my original setting of a slight FSB overclock. A week later it happened again. I know how to fix this right? I switched to jumpers and dip switches. Black screen and fans. What? I switched back to Soft Menu and the board came back to life again. HeHeHeH! I wonder how many RMAs this caused? If it happened to me, there was a good chance it happened to others. By now I was convinced it was a BIOS problem.

By this time, BIOS was up to 1004. I spent hours surfing, trying to get some info on the new BIOS and didn’t come up with much. So I decided to do something I almost never do: I went to the site where you can usually find the latest beta BIOS. I downloaded and flashed 1004_02D. I started playing with the Soft Menu. No more black screens on reboots. Now I can start maxing out the board right? Wrong again.

I wanted to max out the Duron before I started upping the FSB so I would know when to lower the multiplier. I set the Vcore to 1.85. I got 1.9. “Alright”, I thought, “Every little bit helps”. On reboots, it would go back to default Vcore. I can fix this – I’ll just connect the L7 bridges.

It worked, but I got the 1.85 I was supposed to get instead of the 1.9 I got with Soft Menu or jumpers. XXX XXX XXXXXXX XXX XX X XXXXX (CENSORED). With 1.9, I could benchmark at 1020 MHz and get into Windows at 1050 MHz with errors. With 1.85, it was 975 and 1010 MHz. It’s amazing what .05v can do sometimes.

I decided to try a trick that worked with an older KT133 board. I didn’t know if it would work with the KT133a. I left it on Soft Menu and moved all the voltage jumpers over. It worked – 1.9 Vcore, no problems on reboots.

Finally, maxing the board out.

The Quake timedemo1 framerates were dismal, so I went to the Viahardware site. Ah, here it is, 4.05b – greatly improves AGP performance. I installed it and timedemo1 went up to 123.4fps in 640×480 – reasonable for a GeF2mx. I almost forgot to mention I took off the stock Northbridge hsf and replaced it with a Blue Orb and Arctic Silver (I know, the Orb isn’t much better but every little bit helps).

Some lurking in the forums told me Joe Average was peaking out at 141-143 FSB with this board. The hardcore were claiming as high as 160 FSB. I started at 140FSB with the settings at Optimal and 2t2t2t. I upped it 1 MHz at a time. All I got was 1 MHz. I hit the wall at 142 FSB.

I changed the Vio jumper and got a 0.1v increase. I hit the wall again at 146FSB. I changed the setting to normal and left it at 2t2t2t. I hit the wall again at 151 FSB. Changing to 3t3t3t did not give 1 extra MHz, so it wasn’t the RAM. Going to a lower multiplier didn’t help. There’s a problem somewhere else.

I took a memory performance hit with normal setting that was not acceptable to me. For all practical purposes, I was maxed out at 145 FSB. I could benchmark and play Quake 3 (I can win) without any problems. I’m not much of a gamer. I bought Quake 3 for timedemo. Quake 3 is also my idea of a stress test. I don’t care if a board crashed three times in three days ruining Prime95 twenty four hours a day. All I care about is if my board is stable running my applications 24/7.

There were still things I could do to up the FSB. I could trick up the VIO pins to get a few tenths more voltage. Better cooling on the Northbridge. Some of the hardcore are even cooling the clock generator, but mine was just warm to the touch. Sometimes it’s not a good idea to stuff all the slots with RAM.

There are those who advocate one big stick in the slot closest to the Northbridge. However, the PCI and AGP speeds are approaching the point were I know I’m going to start stressing the hardware. The question is, “Is it really practical to go any further?” For me – no. I’m not interested in bragging rites. So, good ‘enuff.


At 1 GHz CPU speed, the GeF2mx is no longer CPU limited. Higher speeds will give you little increase in FPS. However, CPU speed is not everything. Memory performance is another thing to consider. I’ve chosen to show 100 FSB x 10.0 =1000 MHz, 133FSB x 7.5 = 1000 MHz, and 145FSB x 7.0 = 1015 (practical max out). For those of you with older systems, this will illustrate the dramatic increase you can get in memory performance. Benched using SiSoft Sandra 2001se.

Graph 1

I think the chart speaks for itself.

A little Quake timedemo1 in 32bit high detail for the gamers.

Graph 2

These are with the original 5.32 drivers that came with the card. Frame rates can be confusing depending on how timedemo is set up. 16 bit and low detail will bump up the FPS a bit. A little cooling, overclocking it, and trying other drivers should bring these up. Still reasonable for an MX. You can only expect so much from two rendering pipelines.

The MX-200 is a crippled version of the MX. The MX-400 may give you about ten FPS (more or less depending on memory speed) above the MX. My card came with 5.5ns sdram. Before you buy a video card, do a little homework – find out the speed of the memory and who makes it. Otherwise, you might wind up doing the “would have, could have, should have” bit.

Here’s another one that should get your attention if your using an older system.

CPU Bench

The Duron 750 – performance without the price. You can see what an overclocked Duron 750 does compared to a stock Duron 800.

Final Words and Conclusion

Just out of curiosity, I went back and read all the early reviews I could find. There was not one mention of any of the problems myself and others experienced. I know they had to have at least some of them. It really makes me wonder about the integrity of these sites. I’ve got an idea for a commercial called “The Power of Asus” – or any other major manufacturer., with Ed’s rants, was the only one that said anything. At the time I sent Ed an email. “Either I have the best board and you have the worst board or something else is wrong”. To be honest with you, if I had been using Win 2K and a SB live audio card, I never would have sent that email. As it turned out, with the right BIOS and drivers, everything works.

I make no claim to being an expert at anything. I’m just a very tenacious individual. When I have problems, years of building computers have taught me this lesson:

More often than not, a simple email or post in a forum just doesn’t get it. The Mobo god helps those who help themselves.

You may do better or you may do worse. “All mother boards are not created equal”. Manufacturing tolerances of components will dictate how good your board is. The right combination and you may have a superior board. The wrong combination and you have an inferior board. You pay your money and take your chances.

Comments on the article good or bad (if it’s constructive) are welcome.

Is an A7V133 a good board? Yes, with the right BIOS, drivers, and hardware.

Is the A7V133 a great board? No, I’ve never had what I considered a great board.

Was it worth the effort to get it right? Yes.

End of story.

John R. Abaray


Leave a Reply