The SD-11 is a full ATX form factor, which is free from the ‘clutter’ that most motherboards exhibit today. The IDE and floppy connectors are nicely located near the drive bays, allowing the cables to be shorter and kept out of the main airflow in the case (very important, as we discovered). There are no switches on the board and the only jumper is for clearing CMOS, which is located in an easily accessible spot – at the edge of the board in front of PCI slot 2, right next to the panel connectors and battery.
The board sports 5 PCI and 1 AGP slot, but has only a single ISA slot that shares the external case slot with the last PCI slot.. This is not as big a consideration today, since most expansion cards are available with PCI connectors but it could be an issue for those who still wish to use their ‘legacy’ cards. Due to the uncluttered layout, full length cards are not an issue at all. One item to note is that the board is equipped with only a single external serial port.
FIC chose to use the VIA VT82C686A Super South Bridge chip, which includes UDMA/66 and Advanced Power Management features. They did not implement the AC97 built-in Audio. One noticeable feature on the motherboard is the number of voltage regulator chips required to handle the power requirements of the Athlon processor. This is one of the reasons we recommend users carefully evaluate the power supply issue.
Though the official SD-11 spec sheet shows multiple bus frequencies from 50MHz to 133MHz, it is set automatically with no BIOS option for the user to play with. According to FIC this feature will be available in the next BIOS revision. There are no current plans to offer the board with the Award BIOS.
The PHD PCI card is a circuit level diagnostic, which communicates directly with the various controllers and checks out the entire PCI bus, DMA and interrupt controllers, memory controller, as well as the timer and RTC. This diagnostic did not show a single error even after 100 test runs.
The PCI Plus card is similar to the PHD PCI, but tests the ISA bus as well as the various controllers. The DMA channel tests did show some failures during data transfers, however this is not necessarily a serious issue. All tests are implemented by very strictly expecting adherence to industry standards, so even a small deviation in the signal will cause a failure, but would be unnoticeable to most expansion cards. What this failure does indicate is that there may be an increased possibility of an incompatibility with ISA cards that will not tolerate such a deviation (without using an oscilloscope, the amount of deviation is unknown). We did test several ISA cards, including modem, sound card and NIC without problems.
QuickTech Pro is a software based, system level diagnostic, which tests at a higher level than the PHD cards, and also includes a set of Y2K tests for the BIOS and RTC. Possible tests include processor (CPU and NPU), memory (cache and main memory), hard drives, floppy drives, video, keyboard, parallel and serial ports. Using a set of loopback plugs for the parallel and serial port, we ran the diagnostic for 2 hours (in ‘burn-in’ mode) with all tests selected. The only error reported was the EPP data ports test, which has been reported to FIC as a possible issue for the production board.
Operating System Support
In order to test the stability of the board, we decided to test under a variety of operating systems, and run the most stressful applications we had available. For Windows98, that meant Winstone99, 3Dmark99 Max and Quake2 in demo mode, running for at least 4 hours solid. For WindowsNT, we used Winstone99 business and high-end tests in demo mode. Linux was tested simply by running a kernel compile. We did not test OS/2 due to some installation issues that could not be resolved before our publication deadline.
We also wanted to do some compatibility testing with the limited number of video cards, SCSI adapters, NICs, etc. that we have available. The list includes the following:
- Video – Diamond Viper330 AGP and PCI (4MB), Matrox Millenium PCI (8MB), Matrox Millenium II PCI (4MB), Voodoo3 3000 AGP (16MB).
- SCSI – Adaptec AHA-2940UW PCI, Domex 3192U PCI
- NIC – NE2000 compatible ISA
- Modem – MaxTech 33.6 ISA
- Sound Card – ESS1868 generic ISA
- Memory – Crucial Technology PC100 and PC133, EMS PC133, Corsair PC133
During the compatibility testing, we encountered a lockup with the Viper330 AGP card trying to run 3Dmark99 Max. This was reported to FIC, and they confirmed the problem. The engineers are currently working on the solution, and in the meantime the card has been removed from the compatibility list.
Our biggest challenge was running the stability tests. 3Dmark99 Max and Quake2 seemed to have no problems even after 5 hours of continuous demo runs, however Winstone99 would blue-screen or lock up the system after 4 to 5 hours of operation. FIC confirmed that they had experienced some problems earlier, but felt they had solved the problem. It seemed that no matter what memory, video card or BIOS settings we used the lockups would occur after the same period of time. We finally checked the heatsink on the AMD 751 chip, and found it to be exceptionally hot when running the Winstone tests, so we applied a little extra cooling by blowing a fan across the heatsink. After this, the tests ran without any problems whatsoever.
Due to the limited time available, the number of tests we ran and the difficulty in identifying a few of the problems we did not get a chance to run our optimization tests. This will be provided in a separate article at a later time. This testing is quite involved, requires quite a bit of benchmarking, and needs to be carefully controlled for accurate results.
Our policy on performance information is simply to provide the best settings for the hardware we have tested, with some explanation of what effect each one has on overall performance. We do not believe that benchmark numbers have any usefulness in a stand-alone product review, as there is no way to compare the numbers to anything meaningful. As a comparison between different settings, the numbers are valid and will be provided, however.
We must strongly note, however, that any benchmarks given should not be used to compare with benchmarks on any other site, or with any other articles even on this site. Benchmark numbers appear to be one of the most misused and/or misunderstood tools in the industry today, and we will try our best to avoid adding to the problem.
Be the first to discuss this article!