Sunday, January 28, 2007

MSI: Screwing up hardware once again.

Back when ATi first launched the Radeon Xpress chipset, i went out of my way to get one of the motherboards for AMD's Socket 754 Athlon64 processor. Reason why is that I already had a couple Socket 754 chips, and Asus's K8V-Deluxe was starting to get a little, funky, with stability. The Radeon Xpress, in addition to being newer, was proving to be an overclocking monster, easily keeping pace with Nvidia's Nforce Solutions.

So, when I went shopping, the only Socket 754 Radeon Xpress motherboard was MSI's RS480M. Got it in, and was promptly horrified by it. MSI didn't just have limited voltage controls, there were NO voltage controls at all. DDR 400mhz memory that ran at timings of 2,2,2,5 in the old Asus K8V-Deluxe, would only run at the stable speeds of 3,3,3,8. This isn't something that should happen with Athlon64 chips where the memory controller is on the processor, not on the motherboard. One last niggling problem was the location of the BIOS clear Dip-Switch, which was nestled right behind the front panel (power on, reset, hard-drive light) connectors. It was hands down one of the worst 3rd party computer motherboards I have worked with. That also ignores the speed issues with the ATi SB400 and SB450 chipsets, which had lousy Serial ATA and USB 2.0 transfer speeds.

Okay, so recently a client decided to go with an Apeiva X-Qpack case, something I've talked about before on this blog. I like it because of it's small size, and you can use standard add-in parts.

Because of my success with Asus's P5RD2-VM, a Radeon Xpress chipset, my client also elected to go with Radeon Xpress, but for Athlon64 Socket AM2.

Again, the only motherboard in this catagory, was from MSI. The K9AGM-FID. Now, having dealt with many other motherboard manufacturers before, it is fairly standard practice that when somebody launches a fairly horrible motherboard (like the RS480M), any further editions in the line-up generally have those issues resolved.

Nope. Not with MSI. The K9AGM motherboard shares the same layout as the old RS480M, which means the BIOS switch is still in one of the worst places I've ever seen a BIOS switch placed. And like the RS480M there are absolutely no voltage controls what-so-ever on the motherboard. Yes, that is right. No voltage controls on a Chipset that is designed explicitly to have excellent voltage controls.

Okay, so, why do Voltage controls matter anyways? Are not those controls only used to Overclock? No. The problem in this case is that my client also purchased a full gig of DDR2 800mhz Memory from Geil with timings of 4,4,4,12. Yes, the problem from the RS480M is mirrored in the K9AGM. The memory does not run at the rated timings. Rather, the memory will only run at 5,5,5,15. This is a significant problem on Socket AM2, where that timing difference means a consistent performance loss around 10%-15%. The problem is, the voltage that the K9AGM thinks the memory needs is not the voltage the memory needs.

Now, I for one am disgusted. I know that Radeon Xpress is a killer overclocker on both Intel and AMD platforms. I've worked with a couple of ATX motherboards also using the same Radeon Xpress as the K9AGM where I've gotten great overclocks.

Now, what MSI needs to do, and understand this is not a "would be nice if they did it," it is a "absolutely you are releasing this next week" is update the bios for the K9AGM to include the required voltage controls. Now, I don't frankly care if they are not as extensive as the controls found on BioStar's T-Force series. I really don't. I do care though about actually having something to work with.
Post a Comment