The KT7-RAID is the second motherboard we have looked at to offer an additional USB header to take advantage of the four supported by VIA's 686A southbridge. Even though ABIT did not go as far as ASUS to support seven ports, four seems to be somewhat of a sweet spot, so expansion should not be a problem.

In order to compensate for the missing USB hub, ABIT features an unprecedented four 3-pin fan headers, a Wake-On-Ring header, a Wake-On-LAN header and twin SMBus headers. Additionally, the KT7-RAID is the first board we have reviewed that includes an active cooling solution on the North Bridge. Now, we have had 440BX-based boards overclocked to 150MHz and i815 systems running at 166MHz without a problem so it is a bit perplexing why ABIT would need to add a fan to the VT8363 North Bridge. Nonetheless, since they went to the trouble of even adding thermal paste, we cannot complain about the additional cooling.

Perhaps the biggest difference between ASUS' A7V and the KT7-RAID is ABIT's inclusion of IDE RAID support. Whereas ASUS integrated the Promise PDC20265 ATA/100 controller, ABIT went back to HighPoint for their HPT370 (successor to the HPT366) ATA/100 RAID controller. This controller boasts RAID 0 (striping for performance), 1 (mirroring for data security), and 0+1 (striping and mirroring for performance and security). Of course, the main caveat to utilizing this feature is the added cost of additional hard drives. Using 0+1 means buying four disk drives - something reserved for those who do not need to consider money as an object. We would prefer a SCSI array if anything, but the IDE RAID support may come in useful for anyone dealing with budget video/audio production.

At the moment, it seems as though every consumer Intel platform is capped at a 512MB memory limit, making the 1.5GB ceiling of the KT7-RAID quite attractive. If you plan to rise above 512MB, make sure and buy quality RAM to avoid any potential problems the VIA chipset may encounter with heavy memory loads. Unlike ABIT's previous offering, the KA7, the KT7 does not make use of Texas Instrument's CBT16292 multiplexing switches. This may be due to the inclusion of one less DIMM slot, or maybe the signal clarity gained by using the additional buffers on the KA7 was not significant enough to justify another implementation. Whatever the case may be , we encountered much better memory compatibility with the KT7-RAID than with the KA7.