The Trinity K7 is a single Slot A board, with 3 DIMM slots that will accept up to 768MB of either standard or Virtual Channel SDRAM. The clock chip used is the ICS 9248AF-64 , and jumpers are available for running the CPU clock at 180MHz, 200MHz, 240MHz or 266MHz. Unfortunately, the chipset does not appear to support FSB speeds above 200MHz, so the latter two settings do not work even with PC133 memory. Though the processor voltage can be detected automatically, there are also a set of four jumpers that allow manual selection in the range of 1.3v to 2.05v in .05v increments.
Tyan has traditionally catered to the ‘power user’, and the Trinity K7 follows this tradition. With 6 PCI slots, 1 ISA slot (shared with the last PCI slot) and the obligatory AGP slot, few should complain that the board does not have enough expansion capabilities. Smartly, Tyan decided to dispose of the almost useless AMR slot that has, for some unknown reason, become standard on so many motherboards, and is also in line with the ‘Performance Desktop’ market segment.
The standard PC99 compliant ATX connectors are present (parallel, serial, USB,PS/2 mouse and keyboard), as well as a game port and RCA connectors for AC97 audio provided by the on-board CODEC chip. Correction: The on-board audio can be disabled in the BIOS, under the ‘Advanced Chipset Setup’ selection by setting the ‘On-chip Sound’ option to ‘disabled’. Headers for Wake-on-LAN and Wake-on-Modem connectors are present, as well as Video-In (DVD) and CD-In (CDROM) for external audio input. In addition to the two USB ports at the back of the board, there are headers to connect the cables for two additional USB ports at the front.
In order to make sure that the motherboard had no low-level problems that would prevent the successful completion of the stability tests, I first run a burn-in test using Quick Tech Pro from Ultra-X. QuickTech Pro was run for 8 hours with all components being tested, including DMA/IRQ controllers and parallel/serial ports (with external loopbacks). No errors were recorded.
To ensure that there were no memory problems that might affect later tests, RAM Stress Test was run for 50 iterations (all 30 tests). With all DIMM slots populated with 128MB modules, the ATS test (Algorithmic Test Sequence) failed consistently. Removing one module, or replacing one with a 32MB module eliminated the error. Tyan has been notified, and I understand that they are making BIOS changes to correct the problem.
The next set of tests involved our standard circuit-level diagnostics using the PHD PCI diagnostic card, also from Ultra-X. This diagnostic card measures the signals for all circuits (IRQ and DMA controllers, PCI and memory busses, system timer, etc.) against reference timings, and if there is even a small deviation, a ‘failure’ is reported. These tests were run 100 times in succession to make sure there were no intermittent errors. The Trinity K7 passed all tests on the PHD PCI card without any failures reported, but did have errors reported from the PHD Plus card.
The PHD Plus card plugs into an ISA slot, and performs very stringent tests on the DMA and IRQ controllers and channels. Failures in these tests could be a warning that some ISA cards that cannot tolerate any deviation from the spec may not function properly. In my experience, however, the majority of motherboards experience some failures in this area so it generally isn’t a major issue. However, if you have an ISA card that you are planning to use it would be prudent to check for any reported problems before assuming it will function with this board.
Stability and Reliability
I first installed Windows 98 (second edition), and ran the Passmark Burn-in test. This is a multi-threaded burnin program that puts a heavy load on various aspects of the system simultaneously, including the processor, graphics and I/O subsystems. Burn-in Test was run for another 8 hours with all processes being active (except for the printer test), and set to 100% load. This test truly bogs down the machine, even with a 900MHz processor installed and 768MB of memory. There were no failures or lockups recorded.
I also installed and ran Business Winstone 99 and Content Creation 2000 tests in Demo mode for another 24 hours each. The test settings were set to stop on any error, and to reboot after every run. Unfortunately, with these tests it is virtually impossible to determine if errors are due to hardware, unless the errors occur very frequently. There were some occasional errors reported, but this is not unusual with continuous Winstone runs like this, and all recovered when the test was retried.
I next installed Windows 2000 and ran the Passmark burn-in test for over 72 hours with no errors. Winstone 99 (Business and High-End) and Content Creation 2000 were run in demo mode for 10 iterations without rebooting without any errors being detected, so the errors in Windows 98 were very likely operating system related. Note that Windows 2000 does not require the use of any drivers to be installed, other than the VIA USB drivers.
Be the first to discuss this article!