R581A motherboard update

The results that we are reporting were not based upon extensive testing, but were performed in a limited time frame. There are no guarantees that you can get these same results with a similar setup. We plan on additional testing with multiple motherboards and processors to validate these results

The R581A has been available now for several weeks. Most of the major hardware websites have posted their reviews, as have we. The general consensus has been that the R581A is a great motherboard that is both reliable and stable at speeds up to 90MHz. Everyone has also concluded that the board does not work at 100MHz, with reasons being given that include the speed of the Tag SRAM chip and the cache chips themselves. In fact, M Tech is currently searching for 5ns cache chips so they can determine if this will allow the motherboard to work at 100MHz.

In the original R581A report, we posted information that we believe the motherboard actually *does* run at 100MHz, and that the major limitation is that today’s processors will not function at that speed. This was based upon comments made on Tom’s Hardware Guide regarding tests done several months ago with a ‘custom made’ motherboard, some speculation posted on Anand’s Tech Guide, and from our own detailed tests. Most people who contacted us were convinced the problem lie in the cache chips rather than processors themselves. Even we were ‘swayed’ into believing that this was at least part of the reason.

Trying 100MHz

In running our compatibility tests with the K6-266, we decided to see if the processor would even try to boot at 100MHz. What we expected was that Windows would start, but would lock up before completing initialization. To our utter surprise, the system booted completely into Windows!

Not believing our success, we shut down the system and checked all of our BIOS settings. Both L1 and L2 cache were enabled, and the memory settings were on ‘auto’. Just to be sure, we cleared the CMOS and started again. Once again, Windows came up without any complaints.

Well, at this point, we figured it was time to do some benchmark testing. First, we had to get a reference benchmark, since all benchmarks are relative values. The system was not set up to get the fastest results, but was only intended to test for reliability. This means that some of the components were low-end, and caused the actual numbers to be low compared to what some people post. However, we did make sure that the *only* difference was the processor and related jumpers.

Tests and Results

The test system was composed of an R581A motherboard with all BIOS settings at default. The memory was 16MB of Crucial Technology 50ns EDO (two 8MB modules). We had a Trident TGUI9440 PCI card and a Seagate 1.7GB hard drive (PIO 4). These components were used consistently for all tests.

The first test was with a K6-233 at 3.5x66MHz. The Winstone97 results came up as 30.7, with no errors or timeouts. We then tested the K6-266 at 4x66MHz for comparison. This resulted in a Winstone97 mark of 31.7 – about 3.25% greater than the K6-233, and consistent with the results from other motherboards. The final test was the K6-266 at 3x100MHz. There was one timeout during this test, which we elected to retry. The test recovered and re-ran without any problems, and turned out to be the only error during the entire Winstone97 test run. The results of this test was 34.5 – almost 9% higher than running the processor at the standard 266MHz setting.

Conclusions

The only conclusion we can come to is that the R581A motherboard works fine at 100MHz even with the 8ns Tag chip, and the 6ns cache chips. The *real* limitation for running at 100MHz is the processor itself. We have come to the conclusion that the K6-266 is the *only* processor that will run reliably at 100MHz today, and that the R581A will run reliably at that bus speed with the proper processor.

The final tests will be when we receive a motherboard based upon the VIA MVP3 chipset, that is designed to run at 100MHz officially. At that point we will be able to take all of our processors and see if any others can run stabily at that speed. In addition, since the tests we ran were very limited, and only included one motherboard and two processors, we cannot guarantee the results are entirely valid. When we get the opportunity, we will need to test with multiple motherboards and processors to see if there are variations within the components that will prevent some people from running at 100MHz successfully even with this combination.