The Cranky MSI K7TPro2A

Add Your Comments

The last MSI motherboard I attempted to review caught on fire. This one did a little better.

Actually, this board comes close at times to inducing you to set it on fire, but never quite crosses that line.

Our Philosophy

Our conclusions on products fall into three broad categories:

  1. This is good.
  2. Ehhh, you can do better.
  3. Throw it out the window.

We also believe a product should actually work reliably without endless tweaking, and at a reasonable level of performance. Tweaking to get the last bit of performance
is fine. Tweaking to get it to boot isn’t.

Specifications:

I see no point cutting and pasting these things from the manufacturer’s website, so if you want to see them, go here

Essentially, it’s a standard KT133 board with ATA100 support. Six PCI slots, no ISA. Three memory slots.

Yes, it has big capacitors. Reminded me of the ashcans people used to blow off when I was a kid. You don’t have a ton of room for a fan/heatsink, but you should have no problem with say, an Alpha PAL6035.

Installation

Putting the board in a case is routine. Outside of several quirks known to the KT133 chipset (somewhat higher voltage; turning off after a few seconds on an initial cold boot); initial core installation also routine.

I followed the installation order found here, and you should too, for any KT133 board. Very handy reference.

Problems

The Multiplier

If you provide the ability to change the multiplier in BIOS, it’s not unreasonable to expect that it will actually work. If the machine demands a cold boot before working, OK, but if that the case, it should work then.

The only way I’ve been able to change the multiplier consistently is to take the steps I described here.

Now sometimes I haven’t had to get into BIOS again. Some of you just needed to cold boot to change the multiplier. Some of you reported this trick didn’t work for you at all. Some of you have no problem at all.

However, just to show I’m not completely crazy, this reviewer ran into the same problem, so I am not alone.

What moves this problem from an annoyance to an irritation is that you have to go through this whole procedure every time your machine crashes (and this motherboard does crash).

Nor is this is not a matter of a bad pencil job (though if you run into this, that should be something you should check; the pencil marks do wear off). I thoroughly recoated my TBird before testing. Besides, if the pencil connections really weren’t there, they wouldn’t work at all, or just work sporadically. Not the case here.

BIOS

One of the first remedies for these sorts of issues is a new version of the BIOS. Since I’ve gotten this board, MSI has come out with new versions, even an overclocker’s version that lets you do a few things the regular BIOS won’t.

When we got the board, we flashed the BIOS to version 2.01. No problem then. I’ve tried numerous times to try other BIOS versions. They won’t work. Tried with the autoexec.bat, tried them without. Tried them using default settings. Nothing worked. The BIOS just loaded, and looked at me, made no effort to flash.

It’s not fun resetting your machine in the middle of a BIOS session, unless you like the rush of wondering for ten seconds, “Have I fried the BIOS chip?”

Again, others have had this problem, others have not. Some could fix it, some couldn’t.

I probably have flashed a mobo BIOS a hundred times. Until this machine, it was no big deal. I did what I was supposed to do, and it did what it was supposed to do. Not here.

Maybe the name of the flash program: AWFL should have told me something. 🙂

“I’m Not Going To Do It Until I’m Good And Ready!”

Give this machine something new to do, and it gets cranky. It usually shows its crankiness by not booting.

A typical example:

I had relatives coming over. They were having problems with a game, and wanted me to troubleshoot. To run the game, I needed sound. Hadn’t installed a sound card since I was so much fun with everything else, but now it was a priority.

Stuck in an Ensoniq AudioPCI card. Hit the power, no boot. Mind you, not a freeze during a boot due to IRQ conflicts; I could understand that. No boot at all. Did that a couple times, pounding on “Insert” keys and the like, same results.

Remembered the board did have on-board audio, good enough for my purposes. Loaded the audio drivers, initiated the audio in BIOS, all the things good little boys do.

Have you ever had to introduce your dog to a stranger, and had to let him sniff the new kid on the block for a while before he or she stopped barking? Well, that’s what this was like. Stick the card in, won’t boot. Pull the PCI modem, now it will boot. Being a completely unreasonable guy, I’d like
to have the modem and sound card both work at the same time, so here I am, in both Win98 and 2K yanking the cards in-and-out trying to coax it into booting and working with both devices at the same time.

Eventually, it did. That’s what makes this board so frustrating. You can’t call it good, and it’s too quirky to be OK, but it’s not quite “throw it out the window” material. If you are persistent enough, the motherboard will do what you want. I would prefer the thing working the first time. I could understand (if not like) it not working ever. I don’t understand it eventually working without any other changes.

Stability

Again, it seems like the machine has a learning curve. Had quite a few problems with start-up and shutdown at the beginning, right after making a change, problems I’ve not seen in other KT133 boards.

I’ve left it alone the last few days, and it’s purred along downright nicely. However, if overclockers are known for anything, they’re known for constantly fidgeting with equipment, and upgrading parts of it all the time. I can hardly recommend this board for that lifestyle.

Memory

Most of you would probably want to run your PC133 RAM at 133Mhz on this board. You might think the way to do that is to pick manual settings rather than let the machine decide what to do by SPD, and you would be wrong, just like I was initially wrong.

Set the memory manually, and you get memory scores from Sandra like 402/470. The reason why they’re bad is that MSI doesn’t really let you manually determine everything, and gives you less optimal settings on a few items.

If you let the motherboard determine by SPD, and your memory is in fact PC133 CAS2 RAM; it will in fact run it just that way with 4-way interleaving (at least that’s what WPCREDIT says, BTW, you can use the KX133 PCR as a guide in figuring out what the settings are on your KT board), and give you memory scores more like what you’d expect.

If your memory isn’t “officially” those wonderful specs, but you can run it that way on other boards; then it’s get out WPCREDIT and start tweaking.

Hard drive issues

The official (as of right now) Via drivers don’t really support ATA100. You end up with hard drives in PIO mode, which you really don’t want.

Fortunately, you can now download unofficial drivers that will do the trick. However, MSI still doesn’t provide them on their website.

In Win2K, it will do something odd, it will rename your IDE devices SCSI devices. Something else I couldn’t help noticing was that SiSoft Sandra hard drive scores using an identical IBM 75GXP on an A7V were significantly higher than that provided by the new Via driver. Not disgustingly so, about 10%, but certainly there.

Performance

Unless a motherboard is an absolute dog; you shouldn’t let a couple percentage points really affect your buying decision. The MSI K7TPro2A is not a dog. Nor is it a wonderchild. The benchmarking I did indicated that it was on par with other premium KT133 boards, not a reason to buy, not a reason not to buy.

Conclusions

No product lives in a vacuum. If you’re going to buy something, you need a reason to buy it.

The MSI K7TPro2A does not provide a compelling reason to consider it over other motherboards. It does not perform significantly better than even first-generation motherboards. It is not notably more stable than its competition in day-to-day operations.

On the other hand, the motherboard gave me at least some pretty good reasons not to consider it over other motherboards. It’s cantankerous. Sure, I can deal with it, and live with it, but why should you when you can get something else?

During a good part of the testing, I was also putting together a Duron system using an A7V. Took the same care in putting in the drivers and patches and all, but the stuff worked. The first time. If I didn’t know what I was doing, it wouldn’t have. I felt like I was dealing with a machine, not a two-year-old.

In my book, if you want to be recommended, you have to beat the champ. Maybe not knock him out, but at least beat him on points. Otherwise, why change? This board doesn’t do it.

The fact that I still consider the A7V the champ after four months of challengers bothers me. It’s not a great motherboard; it’s a good one. I certainly would have thought some challenger would have KO’d it by now, but nope.

I get the sense these designs are getting rushed, both by Via and the mobo makers. Something new every few months, and no time to truly refine the existing technology, since there’s something new that must be done.

Within weeks, we’ll see 133A designs. We’ll see DDR designs. These are bigger technology steps than changing from a slot to a socket, or ATA66 to ATA100. We hope the delays in these products will yield a more finished product, but we’re skeptical. You should be, too.

“You Don’t Know What You’re Talking About, My Board Works Fine”

Something odd has happened with the last two boards I’ve reviewed (this one and the KT7); there have been some pretty fundamental problems with the ones we’ve tested, but many have reported no problems at all.

However, some have reported the same kinds of problems we’ve faced. We actually find that more disturbing than if everybody or nobody faced these problems. We’re seeing a lack of consistency in these products, even after you strip out less-than-astute users.

For instance, the multiplier issue is pretty basic. It’s not something a user can screw up. Most aren’t having this problem, but some are, and the cause can’t be traced to something else.

If a board works perfectly for 85-90% of users, but has real problems for 10-15%, what do you say about it? If you pay attention to the 85-90%, a portion of those who buy on your say-so are going to have nightmares you helped induce.

If you pay attention to the 10-15%, some might argue that you’ve unfairly trashed a product.

Our attitude is you take enough chances overclocking. If you want to get into another crapshoot with your motherboard, fine, but at least know you’re getting into one.

“Too Many Negative Vibes”

Look at it this way, if we like it, it has to be good. 🙂

Email Ed



Leave a Reply

Your email address will not be published. Required fields are marked *