Isolating Where the Problem Occurs
OK, now I’ve had a chance to do a bit more testing, and I have also heard back from one mainboard manufacturer. As worded to me ‘… seems to be a VIA chipset issue, think we can fix with a BIOS update. Will keep you posted.’. In a sense, that answers the question of whether it is a VIA or Creative issue. But since they used the word ‘seems’, I’m still not 100% sure VIA is at fault here. Also, a very persistent reader (thank you very much) located the original official VIA announcement ( www.eurobizz.de/Via/Data_DOC/FAQs/Biosupdate.doc ), so even though it’s posted in an obscure location, that makes it official. I’ve read the text over and over, but nowhere do I see VIA admitting there is a chipset issue, however they do state there are issues with some of the BIOS’s used. This is an interesting site, as it looks to be a marketing / information service for the European market. This all seems very strange to me, and I could go on and on as to why it might be published there, but all would be pure speculation.
A couple of notes: Some have questioned the use of Winstone to look for any problems (including Dean Kent). If you look at how Winstone runs and the setup I’m using, I think you’ll see that it is a pretty darn good test. Each time you run Winstone it installs the actual programs and data, you can set it so if running the test in repetition it only installs the data once or each time it repeats. Since I’ve got Winstone installed on C: but running the test on D: every time it runs, the tests are installed from C: to D:, which means each drive is being access during the install (quite a bit of data). Also by running with Windows and it’s swap file on C: but the test on D:, both drives are being accessed at the same time at times during the test. So if the system is prone to errors while accessing two hard disks at the same time, the constant installing of the test programs and data (in essence reading from one disk, expanding and installing on the other), and running the tests, should be able to find any issues.
But Dean will have a utility available shortly specifically written to check for disk errors, which will hopefully be a great tool to nail down errors that are specific to disk activity. One problem with using Winstone the way I have is that I’ve been assuming the errors are purely disk related. I suspect we will be using this new program to double check any BIOS revisions issued to solve the disk errors, maybe also to double check some of the chipsets / mainboards that did not show any errors in my original tests.
Also, with all the mainboards tested I use PCI slot 2 for the sound card, PCI 3 for the NIC, and in all cases the SoundBlaster cards used had IRQ 5 assigned to them. Over the years I’ve found that if you can setup a SBL (or just about any sound card) this way you’ll have less problems. I’ve run a number of tests on the sound and had no issues (common complaint is ‘scratchy’ sound) with any of the mainboards tested using the SBLV. I’ve received a lot of E-Mail since the first article (thank you) from people who seem to be searching for a ‘cure’ for system stability or sound quality issues that may not be related to what I’ve been looking at, which is disk errors. This could be the content of a later article, but not within the narrow scope of these series of tests.
Remember, I’ve only really tested a very small sampling of the mainboards available with the 686 Southbridge using only one video card, NIC and hard disk, so results may vary from brand to brand, and with different components. And don’t think it’s a Soyo only issue, because it isn’t. I just happen to have easy access to Soyo products. The specific question I am researching is whether there is a disk data integrity issue when the VIA 686 Southbridge is used with a SoundBlaster Live. The answer is yes there is.
So, let’s see what happens when testing a few more chipset combos. Tests were run as before, repeated five times. All systems tested below were also run without the SBLV to check for a stable baseline.
- I re-ran the one test that gave me the most trouble. It would give me a system error very quickly: VIA KT133A (Soyo K7VTA Pro) and two WD 200BB ATA/100 set as masters on each port, but this time used both a SoundBlaster (SB) 16 PCI and a SB 128. Both are PCI 2.1 sound cards just like the SoundBlaster Live Value (SBLV) so if it’s an issue of PCI 2.0 vs. PCI 2.1 you’d expect to see issues with them also. No problems at all, did not get a single error.
- So next I ran an ‘older’ VIA set-up: KT133 and 686A (Soyo K7VTA). The results were the same as when running with the KT133A and 686B, even though it has the older Southbridge and only ATA/66 support. Run with two drives set as masters on each port with a SBLV installed and I did get system (disk) errors. Remove the card and there are no errors. Install the SB 16 PCI and there are no errors.
- So what about an ALi? If it’s a SBLV issue, then shouldn’t it show up with a different chipset? So I ran a Socket A DDR MAGiK1 (Soyo K7ADA) with a SBLV and two WD 200BB hard disk set up to run as masters on each port. No problems at all, not a single error.
That got me thinking about what the actual cause might be. VIA and their chipsets, or just one particular chip? Since the Apollo Pro 266 (Socket 370 DDR) uses a different Southbridge, the 8233, I thought that would be a good choice to try. The new KT266 uses the same 8233 Southbridge, but it in itself doesn’t seem all that stable yet (at least with the MSI K7T266 Pro). I didn’t think that would be a good test as it would be hard to tell what the cause of any system errors might exactly be. And what about the VIA 694A and 686B Socket 370 chipset? Would it show the same errors? I also gave a brief thought of testing a MVP3 Super 7 setup, but since that’s not really a current chipset anymore I did not persue it, even though some come with an ATA/66 Southbridge.
- Ran with a VIA Apollo Pro 266 & 8233 Southbridge equipped Soyo 7VDA, two WD 200BB hard disk connected as masters, one to each port. Installed a SBLV and ran the test. No errors at all…. Interesting.
- Next I ran a Soyo 7VCA2 with the VIA 694X Northbridge, but using the same 686B Southbridge as the Socket A mainboards I’ve seen disk problems with. Now I did get errors, but not as often as when using the KT133 series Northbridge.
- And what happens if we use the VIA 686B and an AMD 761 Northbridge? Using the Gigabyte GA-7DX mainboard I found no problems at all! Very strange….
- Just to see what would happen, I tried an Intel 815EP based Soyo 7IS2. No problems at all.
That seems to narrow the problem down to only the 686 series Southbridge when used with a VIA Northbridge, and only when a SBLV is installed. Whether Intel or AMD CPU, or different VIA Northbridge, it still happens. No 686 Southbridge and VIA Northbridge and there doesn’t appear to be any problems, even if a SBLV is installed.
Only time will tell if the VIA 686 drive error issue of running two IDE drives as masters, one on each port, and a SoundBlaster Live will be solved. It will be interesting to watch as new BIOS’s are issued. Will they solve the problem, or is it going to take a revision of the chipset? Is there anything Creative can (or should) do? Will there be any loss of performance or features (such as PCI 2.1 support)?
You will notice a few sites have ‘solutions’ posted. They may help, but really it’s going to take a ‘fix’ from the mainboard manufacturers (BIOS level) or VIA (chipset level), unless it does turn out to be an issue with the SBL. If it is a Creative issue, then is still may be up to the others to solve it. One thing I would do for now. If using a 686 South bridge and a SBL together, I would think twice about using two IDE drives, especially installing them on two different ports. But let’s also not forget that many users only have one hard disk installed, and this issue will likely only affect a small percentage of users.
Be the first to discuss this article!