I-Appliance BBS The Official Source for Internet Appliance Upgrades and Mods Click Here! BBS Main List | Sign In | Sign Up | Search | Help | Linux-Hacker.net Reply to Thread | Printer |

Home / I-Opener Areas / I-Opener Technical Stuff Help: Is "DMA enable" possible with Win98SE?

Help: Is "DMA enable" possible with Win98SE? (modified 0 times) vailr As topic says, can anyone with DMA enabled confirm that this is possible, and method used? The Hitachi 6.5 GB is capable of DMA (UDMA33) when installed on desktop machine, but not when installed in IO. Would lifting pins 37 & 39 on the SanDisk enable faster throughput? Or is the IO bios preventing DMA enable? 05-28-2000 10:01:20

RE:Help: Is (modified 0 times) klunk Profile I know this thread is old, but I have not been able to find an answer to the question "Is UDMA possible for an IDE hard drive under Windows 98 SE in the I-Opener?". It appears that my Wild_Pencil/Badflash bios V5.30 may not support UDMA since it only offers PIO4 and lower, but I have a DMA box that I can check in the Device Manager. Unfortunately, it won't stay checked. And, yes, I have selected "Primary IDE Port Only" in the Device Manager, and, no, I don't have any IRQ or memory conflicts. I have not disabled the Sandisk controller. Some posts claim UDMA is possible, and one claims not because of bios limitations. Comments? This seems like a pretty basic question that should have been answered definitively years ago. klunk 02-15-2002 08:11:51

RE:Help: Is (modified 0 times) Ragnar1 Profile Have you installed the Via drivers? If so, did you make sure during one of the setup screens you enabled DMA when it asked for it? 02-17-2002 23:36:56

RE:Help: Is (modified 0 times) klunk Profile Yes, I checked the DMA box when installing the 4-in-1 drivers. This works on my other VIA machines--a MPV3 machine with an AMD K6-III+ and a KT133a machine with an AMD Thunderbird. It does NOT work with the I-Opener. Has ANYONE here gotten DMA to work on the I-Opener? klunk 02-18-2002 06:43:59

RE:Help: Is (modified 0 times) *SF* Profile DMA won't stay on v5 help! 02-18-2002 14:57:15

RE:Help: Is (modified 0 times) klunk Profile Hey *SF*: Did you actually read the thread you linked for me??? Do you think I haven't??? Like ALL the other threads on this subject, it ends in confusion. Some report DMA can be enabled (with no real assurance), and one reader says that DMA can't be enabled because of bios limitations. Just like my original post, isn't it *SF*? Quit wasting our time. klunk 02-18-2002 18:35:24

RE:Help: Is (modified 0 times) *SF* Profile Yes, I Read The Threads I Link to. If You would have added to that thread or linked your research, I would have Know How Smart You Are !!! p.s. Don't Worry, I won't be Wasting Anymore of My Time on YOU. Good Luck, * StarFish * 02-18-2002 20:09:32

RE:Help: Is (modified 1 times) Wild_Pencil Profile | Email Y'know, klunk... for someone who's asking for help, you sure are sharp with the tongue when someone points you to a redundant lead. Without knowing what articles you've read, Starfish decided to be a Nice Guy and dropped a reply for you. Snapping at Starfish was uncalled for. As one of the local BIOS hackers, I'm too busy to field these kinds of questions. I rely on helpful people like Starfish to point newcomers in the right direction. Doing things this way lets power-hackers like me focus on the more interesting problems -- like creating a working TFT patch, or opening up a backdoor into the AnyISP Sandisk image. To answer your question. Some people have indeed been able to get DMA working, using the latest Via 4-in-1 drivers. It's speculation as to whether or not a Registry entry needs to be change to force it on. I can't help much on the Windows frontier, since my primary O/S is Linux. (DMA definitely works in Linux, provided your hard-drive can support it.) In my own experiments with the 4-in-1 drivers, I have to say that I absolutely HATE those drivers. Very inconsistent behavior. Removing IDE 2nd Channel always broke Win98, forcing entry into Safe Mode to try to salvage the operation. (And even then, I've only salvaged the system twice out of around 10 times...) In the end, I stuck with the default and never optimized things further for fear of over-tweaking. My advice to you. If you really want some hard-drive performance gains, look in the Tech Reference (Moderated) area for a post by Georgie. He has some instructions on manipulating the PCI registers directly to get a boost.. and I understand it works very well. It just takes a patient person to tweak it to the unique hardware involved for best performance. (It's not a one-size-fits-all thing, which is why I've no plans to create a BIOS patch to implement this..) -WP 02-18-2002 22:38:20

RE:Help: Is (modified 0 times) klunk Profile *SF* and Wild_Pencil: Sorry to you Starfish about my response. It was uncalled for, but I really have done my homework. I have read virtually every post on this board back to April, 2000. Wild_Pencil, if even the author of my bios hack doesn't have a simple answer to my question, then I guess it's hopeless. I don't believe in random performance on the IDE bus, with some people enabling DMA with little effort and some not, even with great effort. I have a clean install of Win 98SE and the latest VIA 4-in-1 drivers (ver 4.37). No go. Outside of some unconfirmed reports of sucessful DMA on the IDE bus, what suggests to you that DMA is possible with the 5.30 bios? Can you enable it in Linux? I have read all the posts on the PCI latency issues, especially as they concern the USB mess on the i-opener. Again, there was no consensus as to the effectiveness of monkeying with latency. As I recall, people were working on a new bios that would fix the USB problems. None was issued, to my knowledge. In any event, PCI latency optimization could hardly make up for the speed gains of DMA. But if that is all I have to deal with, then I'll give it a try. Latency optimization might be more effective speeding up the IDE bus than it is for preventing data transfer errors on USB. klunk 02-19-2002 05:56:56

RE:Help: Is (modified 0 times) Wild_Pencil Profile | Email If you compile Linux with the VIA-specific IDE drivers, then DMA is enabled by default. Some people have reported that disabling DMA for the Sandisk (hdb=nodma) speeds up Linux's boot time, since it would otherwise timeout on it. Of course, Linux is a bit of an exception, since it doesn't rely on the BIOS at all. Tbe stock IOpener BIOS doesn't have a setup option to toggle DMA mode. (Most BIOSes with VIA Southbridge do have such an option). Netpliance pretty much stripped out everything useful from their BIOS, leaving just the bare essentials to boot the IOpener up as intended -- as an Internet Appliance. It's a bummer to work with crippled firmware, but that's what we've got to live with if we want to keep the machine legal. Tweaking the Via 4-in-1 drivers is about the only way I know of to force DMA in Windows, since the BIOS won't do it for you. Have you tried setting "Drive D" to "None/None" in the BIOS setup to disable the Sandisk, leaving your DMA-capable drive as the only device on the IDE chain? -WP 02-19-2002 16:26:31

RE:Help: Is (modified 0 times) klunk Profile Well, I tried disabling the internal SanDisk in bios and DMA would still not enable. I then decided to go to the MS Technet site to scope out the DMA problem. MS, it turns out, has a paper on enabling DMA that shows how to use the "debug" program to probe IDE devices for "multiple-word" DMA compatibility. Guess what, the hard drive is compatible, as expected, but the internal SanDisk is NOT. I now think that having a slave drive on the primary IDE bus that does not support DMA kills it for the primary (hard) drive as well. Now it's time to disable the SanDisk controller electrically! klunk 02-20-2002 16:37:22

RE:Help: Is (modified 0 times) ckbone Profile I think you're wasting your time. I've got a cdrom as a slave on the primary channel. DMA still wouldn't enable....tried everything. Course maybe you'll get lucky.... 02-20-2002 18:11:05

RE:Help: Is (modified 0 times) Ragnar1 Profile To check if dma is enabled on 98se. Go to control panel, system, device mangler. Click on "view device by connection". Click the "+" beside Plug and Play bios. Third item down is Direct Memomry Access controller. Double click on it. Click on resources to see if there is any conflict. I have 98se installed on two 10gig IBM TravelStar hard drives. One with the Via drivers installed, the other without the Via drivers installed. Both use IRQ 4 and report no conflicts. It even has a tab called settings where you can select the size limit of a tranfer. One thing about the Via drivers. Without the Via drivers installed I can use my Badflash custom cable and access my 1402 cdrom. But I can't use any usb device. With the Via drivers installed I can use my usb cdrom and my usb camera, but the via drivers kicks the 1402 cdrom off of the ide chain. If I try and load any usb driver with the 1402 cdrom then 98se crashes. Oo la. 03-24-2002 21:31:14

Cleaning up the mess..... (modified 0 times) Georgie Profile klunk, DMA is possible, it stays checked for many people, including me. I actually have never had a problem with that one. I have tried it with the Sandisk enabled and disabled in the BIOS, it stays checked - and DMA truly works - for me either way. In one word, it is not the Sandisk. The reason for these tests is that someone contacted me who was having the same problem as you and ckbone. We haven't managed to solve his issue. But I am nice enough to give it another try ;) So, a few questions: What is your OS, IO version, BIOS version, 4in1 driver version at present? On another note: the HDD speed fix that you can read in the technical area, has nothing to do with PCI Latency. It correctly configures the data transfer timings of the IDE controller, which the IO BIOS completely forgets to do. As such the fix only does what all decent BIOSes do in all motherboards: configures your computer for optimal speed. It should be part of the BIOS. I will get back to this later. And yet another note: FYI, there is a completely new, perfectly working USB fix coming out. It really fixes all the USB troubles. So, please, come and have a look at it. Later, Georgie 04-13-2002 23:06:33

RE:Help: Is (modified 0 times) klunk Profile Georgie: I really appreciate your help! I have two I-O's both V5's with similar IDE DMA problems. One is a complicated machine with a hard drive and an IDE CDROM drive, which I won't describe since it is not a good base case. My second I-O is a V5 with the following: Bios: 5.31c, also tried ver 5.30 CPU: AMD K6-2+ @ 400MHz HDD: OEM IBM TravelStar 10 GB, also tried a Hitachi 15 GB drive VIA: ver 4.38 4-in-1, also tried ver 4.35 San: Sandisk is disabled in bios (D: drive is "none"), but active electrically PCI: pci.exe run in autoexec.bat with parameter set to 32, also tried w/o pci.exe Mem: 128 MB Op: Win 98 SE, clean install with all updates (IE 6.0, security patches, etc.) My Trials: With ver 5.30 bios and Hitachi drive, and ver 4.35 4-in-1's, DMA box would not stay checked, but machine would reboot OK with no errors. With ver 5.31c bios, 4.35 or 4.38 4-in-1's and Hitachi drive, same behavior was observed. With ver 5.31c bios, no 4-in-1's (ie Microsoft via bus master in Win 98) and Hitachi drive, same thing. The IBM drive is a different story! With ver 5.31c bios and Microsoft via bus master and IBM drive, checking DMA practically destroys the machine. Machine reboots in DOS compatibility mode on primary IDE and you can't do a thing to change this! You can't even install 4-in-1's. With ver 5.31c bios, ver 4.38 4-in-1's and IBM drive, same thing--DOS compatibility mode. At least in this case it is possible (with a lot of crashes) to get the 4-in-1's re-installed and out of DOS compatibility mode. Now, I have of course set the machine up with primary IDE only to avoid the IRQ 15 conflict with USB (I note that installing the 4-in-1's re-enables the secondary IDE port and you have to turn it off again). I have also looked at my DMA controller in the System devices. It is working properly, and happily assigns DMA 3 to the parallel port if I set that port up for ECP mode. The difference in behavior between the IBM and Hitachi drives suggests to me that the drive hardware microcode is a key to this mystery. By the way, pci.exe speeds up IDE transfers significantly, even w/o DMA--from 2.8 to 6.4 MB/sec on the Hitachi, and from 2.9 to 8.3 MB/sec on the IBM drive! Thank you for pci.exe! klunk 04-14-2002 06:53:38

More questions..... (modified 0 times) Georgie Profile First of all, I am happy to hear that the HDD speed fix works well for you too, but I have to point out that "pci.exe" is not my program, I just happen to know and like it. The register optimizations done with pci.exe are my work... As for your setup, it sounds good, almost identical to mine. I have the V1 BIOS, and v. 4.33 of the 4in1 drivers, as differences. You wrote that you use my HDD speed fix with a parameter of 32. What exactly is the pci.exe command line you are using now? Are you using the whole line? If not, please, try once using the whole line - you can find it here. There is one particular configuration problem which will be corrected by one of my PCI settings present in the whole line. It is worth a try. In case it still doesn't work, I will need you to send me data files created with pci.exe. Georgie 04-14-2002 22:22:48

RE:Help: Is (modified 0 times) klunk Profile Georgie: Some further news. I set up a mint V4 yesterday (AMD K6-2+, ver 5.31c bios, IDE CDROM drive as master, HDD as slave), and it allowed the DMA box for both the HDD and CDROM to stay checked! This was with the Windows via bus master, and without pci.exe. But with DMA enabled, I started getting screwey registry corruption errors and file corruption errors. USB starts to work on boot up, ie, the machine gets its DNS through DHCP, but suddenly shuts off when the desktop appears, and it cannot be restored. Sometimes the IDE CDROM drive disappears, as well. The VIA 4-in-1's cause severe problems with USB as well if DMA is enabled, and ultimately the machine crashes from registry corruption errors. The V4 i-opener is truly a different beast from the V5's, but still I believe DMA is broken on the i-opener. klunk 04-15-2002 05:01:42

CD-ROM.... (modified 0 times) Georgie Profile klunk, You always mention an IDE CD-ROM. Are you sure it is not your special IDE cable giving you a headache? Have you ever tried a bare IDE cable - with only 1 HDD connector, that is what I use, from BadFlash - or a IMOD2/3 HDD connection? Then again, many CD-ROMs have problems with DMA access, I mean even many desktop CD-ROMS. Some work, some don't, you would have to test it every time. As a matter of fact, due to the nature of a CD-ROM, DMA would never improve your performance on the CD-ROM, or only very insignificantly. Yes, I am suggesting that you shouldn't be obsessed with it. But when it comes to your HDD, you DO NEED DMA! There has to be something you ill-configure on all your machines. What type of IDE cable do you use? Home-made? 2 or 1 HDD connectors? Something is seriously wrong with your setups. I have only had a V3b and a V1 Iopener so far, but none had problems with DMA. The V3b is my personal machine and everything works on it. My USB works flawlessly with my IDE HDD working in DMA mode (UDMA-66). IDE transfer speeds are between 15 and 30MB/s with 2-3% CPU usage, thus it is DMA working at its best.... I suggest you look for a systematic error, a wrong step you always do on your IOs. Or a common hardware part: Cable?? One suggestion: try it once with your IOpener open, the CD-ROM disconnected and the IDE cable unfolded entirely. Keep the cable straight, and give it a try. It might just be your problem. The other thing can be the BIOS. Have you ever tried the V1 BIOS? Later, Georgie 04-15-2002 07:47:07

RE:Help: Is (modified 0 times) klunk Profile Georgie: After a lot of time and considerable expense, I can now report on my DMA experiences with two V5, one V4(a), one V3 and one V1 i-opener, all using the 5.31c bios from Badflash. To make a long story short, the V5's would not enable DMA. With IBM drives (I tried several ranging from 6.4 to 20 GB), the Primary IDE Controller would exhibit a yellow exclamation mark in Win 98 SE Device Manager and the DMA box would not stay checked. With Hitachi drives (10, 15 and 20 GB models), the box would simply stay unchecked with the Primary IDE Controller OK in Device Manager. The V4 machine will allow DMA with both types of drives. The problems reported by me earlier were thermal stability problems not related to DMA. The CDROM allowed enabling DMA as well, with no problems. The V3 and V1 machines both exhibited perfect DMA behavior with both the hard drive and the CDROM drive. In general, I found cable routing to be uncritical, except as it affects good air flow and temperatures in the box. I conclude that there is a basic problem with the V5 machine/ver 5.31c bios combination and DMA. I have not yet tried my V1 bios in a V5 machine. klunk 05-21-2002 08:43:28

V5 or 5.31 BIOS ??? (modified 0 times) Georgie Profile klunk, It seems we are getting closer to understanding this problem, thanks to your work. (It must have been a *lot* of work...) I suggest you try the V1 BIOS first on one of the V5s. If it still does not work, please, send me the PCI configuration of your Northbridge and the IDE controller. You can provide me with this using pci.exe. Execute the following command from a DOS window inside Windows: pci r?100 r,7,1?100 > klunk.cfg It will write a small text file in the current directory, that is what I would need from you. Please, send me this config file. You can put it up here, but email would be better. My email is: "my BBS ID" at chez.com Anyway, I would try the V1 BIOS first. One last note: when trying these things always make sure that in device manager you set "Primary IDE only" for the IDE controller, and if you happen to have a yellow exclamation mark next to the IDE controller check that you do not have a value named "NoIDE" under the key "HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\VxD\IOS" in your registry. In case you have it, erase it, and restart Windows. Good luck, Georgie 05-23-2002 13:31:40

RE:Help: Is (modified 0 times) jimmy Profile Georgie: Klunk is now "jimmy" because of problems with my password. Anyway, I will install a V1 bios in one of my V5's tomorrow. Darn, I hate tearing the unit apart again, since it works beautifully--apart from DMA. But if it helps the i-opener crowd, it's worth it. My V1 unit with an AMD K6-3+ running at 400MHz, 128 MB ram, 20 GB Hitachi drive, 24X CDROM and USB floppy (which installs as drive "A"!) is totally stable with DMA enabled. The Yamaha sound chip allows me to run my closetful of legacy DOS and early Windows games, as well as surf the net. It will be my "keeper"! 05-25-2002 15:20:41

RE:Help: Is (modified 0 times) salesman Profile Has anyone had any luck getting DMA to enable on a V5 with the V5 BIOS? Mine will not enable. Is this a BIOS issue or a controller issue? Thanks. 09-01-2002 06:44:28

RE:Help: Is (modified 0 times) BadFlash Profile | Email Several have reported that the problems with DMA is that they are folding the cables. The cables need to be split are routed outboard of the drive, not folded under it. Don't ask me why, but folding under the drive messes things up. Could be it is picking up noise from the board. 09-01-2002 12:06:07

Reply to Thread | Printer | All times are PST Powered by UltraBoard v1.62

I-Appliance BBS General - NEWS, Breaking NEWS - linux-hacker.net BBS on CD Questions - User Groups in Your Town - E-Bay, Yahoo Auction LINKS ONLY (MODERATED) - Jokes & Humor - usdtv box I-Opener Areas - I-Opener General Posts - I-Opener Technical Stuff - I-Opener Technical Info Reference (MODERATED) WebSurfer Areas - WebSurfer General Posts - WebSurfer Technical Stuff - WebSurfer Technical Info Reference (MODERATED) Other I-Appliances - Virgin Webplayer - WebPal - MailStation - 3Com Audrey - ThinkNIC / Oracle NIC - Gateway/AOL Connected Touch Pad - Philips/AOL-TV - ACER NT-150 STB - MSN Companion - Compaq IA - Emachines IA - IBM NetVista IA - Intel dot.station - MSNTV2 Network Computers - Fujitsu-Siemens Scovery Linux Net Computers - GallantWeb Server Consoles - Dreamcast - xbox hacking Web Pads - Epods One General Post - Panasonic CF-01 - Fujitsu Stylistic 1200 - Cybiko Thin Clients - Thin Clients DigitalVideoRecorders - DTV Recorders, Replay, TiVo, Etc... - Tivo Hacks Linux - Midori Linux General Board - ITX MotherBoards Single Board Computers - Earth Computers mARMalade Elite Member Area MISC Areas - The Swap Meet WE TAKE NO RESPONSIBILITY FOR ANY TRANSACTION HERE AND NO COMMERCIAL POST PERIOD !!! - Wireless Net - Cameras - Robotics - Routers/Gateways - Mattel JuiceBox