Scott, Thanks for the link, I'll certainly take a hard look at them before I buy.
It appears the problems occur on older revisions of the motherboard with Athlons at 1ghz or higher.
[Edit: It actually appears to be the other way around(!), with the problem occuring on newer revisions of the board]
This seems really odd to me. Athlons were 1GHz+, exclusively, when the first K7S5A board was released. ECS is the premier partner of SiS and they cooperate very well at tracking down problems before any launches are made. This is why ECS has such a good reputation for stability, although mainly among OEMs (which has previously been the primary target for ECS).
So, you'll understand if I am somewhat skeptical of any claims… however, it wouldn't be the first time that even a manufacturer with a good rep released a bad board (and fixed it in later revisions).
Running through the msg board threads, the following problems are listed:
[I had a long list, but they were all "bogus" problems, not directly linked to the motherboard with any significant certainty]
The real problem is identified by the Mr. Athlon guy you mention. He writes as follows:
Q: What exactly is wrong with the “bad” motherboards?
A: The exact problem is not known, but currently two things appear to be issues:
1) 99% of the problem: The “ZP issue” (Driver Impedance).
Testing has found that “bad” motherboards do not work correctly because of the way they “program” ZP, the AMD bus driver impedance for the CPU. These “bad” boards program the CPU to approximately 56 Ohms. Changing the motherboard’s ZP resistance to a value closer to 40 Ohms is thought to make the signals stronger and has fixed the problem in several dozen confirmed cases and in several dozen additional reported cases. So far, there have only been a few reported (and no confirmed) cases of changes to the ZP resistor that did not result in a complete correction of the problem. (Naturally, there are other possible problems and this fix doesn’t fix everything!)
In addition to the ZP resistance, the actual board impedance may vary, as does the actual driving characteristics of each CPU. This means there are at least three variables in this “ZP issue” equation. If you get a good match on all three, your setup is likely to work pretty well. If you have a bad match, you may have many errors.
Overall it appears that “bad” motherboards are right on the edge of working correctly. This is why the problem has been so hard to identify, and some boards actually work quite well, while others are really quite bad.
It appears that boards with 0, 1, or 2 (maybe also 3) on the sticker by the PCI slots usually have 40 Ohm ZP and ZN resistors on the board. Boards with 4+ usually have 40 Ohms for ZN and 56 Ohms for ZP. This change seems to approximately coincide with the approval for Athlon XP processors. Although the reasons are not known, an AMD document has been circulating that states motherboards designed for XP processors should use 56 Ohms for ZP as a starting resistance.
Most fixed board testing has been done with 200 Ohm resistors placed in parallel with the 56 Ohm ZP resistor on the board, bringing the effective resistance to about 44-46 Ohms. This seems to work well for both Athlon and Athlon XP processors.
2) 1% of the problem: Noise.
Although usually not a problem, the Power Switching FET’s and Diodes in the CPU power supply circuit generate large amounts of noise. The motherboard has a “leak path” that passes this noise directly onto the 12V power line. Typical boards have 3-4 volts of “bursting high-frequency” noise on the 12V line that might be the cause of occasional data corruption or other problems. This is still under investigation.
(Source: forum.ocworkbench.com )
This Mr. Athlon guy is kinda scary… take a look at all the tests he has run and the effort put into this whole thing. Whoa…
It should be noted that ECS completely denies the issue and issued the following statement:
In response to claims from a forum member of a certain hardware review site, ECS Elitegroup wishes to make the following statement:
ECS has conducted extensive re-testing of the K7S5A and has held discussions with AMD to investigate the claim that there is a problem with the ZP setting on the K7S5A. ECS tested the K7S5A, as it tests all motherboard models, with a wide range of industry-accepted benchmark tests, including Memtest (not Memtest86) and Winstone 2001, that are used by all tier one motherboard manufacturers. As none of these reputable tests identified any such problem, ECS and AMD are therefore of the view that the alleged problem does not exist. However, we do believe there could possibly be a utility problem with the Memtest86 program.
It should be noted that AMD issued a Technical Note on 30th August, stating that the ZP setting for its XP series of processors should be 56.0 ohms. Since this bulletin was issued, ECS has complied with the guideline to ensure that the ZP setting on the K7S5A is fixed at around 56 ohms, otherwise the XP processors do not work. There is no problem with Athlon processors.
The K7S5A has passed AMD’s stringent testing and appears on AMD’s list of recommended motherboards that are compatible with the full series of AMD XP processors.
In circumstances where ECS believes an unreasonable accusation has been made publicly against the company, ECS reserves the right to take legal action if deemed appropriate.
(source: forum.ocworkbench.com )
Personal conclusion: Thanks, Scott. Until I know more about these problems, I will not be buying an ECS K7S5A board.
-fyo |
