Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
Technology

Secrets Of BIOS Tweaking 172

Sivar writes "While most enthusiasts are familiar with some settings that yield significant performance benefits, many other BIOS settings remain poorly described and may unknowingly play a crucial role in system performance and stability. Ars Technica has an excellent article describing some of the most obscure settings, useful not only for performance, but for tweaking stability and hardware compatibility as well."
This discussion has been archived. No new comments can be posted.

Secrets Of BIOS Tweaking

Comments Filter:
  • by N8F8 ( 4562 ) on Thursday September 12, 2002 @09:14AM (#4244374)
    Just a tip: If you have trouble booting LILO with a USB keyboard try enabling Legacy USB support in the BIOS. It worked for me on a Dell GX240 Optiplex when all I would get is a Keyboard failure notice. You may also have to turn off "halt on error".
    • by Anonymous Coward
      Actually I've noticed on the Abit KT133 MB with a IOgear USB-PS/2 adapter and latest Bios.
      The keyboard will sometimes disappear. This happens under W2K & Linux. The only solution I've seen is to power-down and shut off power to the power-supply and restart from their. A simple warm-boot will not help.
  • PnP (Score:3, Informative)

    by Anonymous Coward on Thursday September 12, 2002 @09:19AM (#4244406)
    Disable that pesky PnP in the BIOS. Could've saved quite a bit of hair pulling on a FreeBSD 4.6.2 install.
    • Re:PnP (Score:4, Informative)

      by turgid ( 580780 ) on Thursday September 12, 2002 @09:54AM (#4244656) Journal
      That's good advice, since modern kernels like *BSD and Linux can and do prefer to take care of configuring PCI devices themselves. It's only users of DOS and DOS-based versions of Windows that need to have this feature enabled, AFAIK.
  • THG has had a good BIOS guide as well:

    http://www6.tomshardware.com/mainboard/97q1/970101 /index.html [tomshardware.com]

    and also a guide on BIOS tuning:
    http://www.tomshardware.com/howto/01q3/010725/inde x.html [tomshardware.com]
    • Those are some nice guides...

      for me to poop on.
    • by Anonymous Coward
      Or better yet, how about link back to the original, definative BIOS guide that Tom ripped off for that article?

      Adrian's Rojak Pot : http://www.rojakpot.com/Speed_Demonz/New_BIOS_Guid e/Index.htm

      It's in the middle of rev 7.0 now, but the 6.x series are pretty good.
    • by Wanker ( 17907 ) on Thursday September 12, 2002 @11:10AM (#4245175)
      Speaking of Tom's... After following the BIOS article (which is a very good read), the first forum I stumbled across was:

      http://arstechnica.infopop.net/OpenTopic/page?a=tp c&s=50009562&f=77909774&m=8400979235 [infopop.net]

      In this forum the poster makes a pretty convincing case that the photo of the P4 3.3GHz chip in the "Hot Contraband: P4 With 3.6 GHz" [tomshardware.com] article was forged. A subtraction analysis (described in detail in the forum) shows a nice little black box indicating they just copied a "3" to make the 3.3GHz photo.

      In my mind, this throws a lot of doubt on anything posted on Tom's hardware. Which is really too bad-- I liked that site a lot.
      • I can't believe it! I always trusted THG. I feel like my entire life has been a lie!!! aah!

        seriously though. why would they fake a pic like that? it makes me wonder if their articles are even true anymore. how many of their benchmarks are fake?
      • I think his claim that PCMark 2002 is an old program with no optimizations for intel or AMD throws it into doubt even more.
      • I haven't visited Tom's in ages... and it's been even longer since I was a regular reader. Fittingly, it was Ars that supplanted THG in my "holy trinity" of hardware sites to read when bored (Anandtech and SharkyExtreme being the other two - Slashdot is in a different category altogether).

        I was really hoping my last memories of Tom's would be good ones. Tom's used to be a great resource in the olden days of overclocking. I learned the basics from him, and overclocked my Pentium 166 to 200 and later my PIII 550 to 733, and both still run stably with "stock" fans and default voltage years later. Now I see deception, and I hear that this isn't the first time.

        This is like being 12 years old and being told to go visit your dying grandpa in a hospital room. Instead of just remembering him laughing and teaching you how to work with wood, you get to remember tubes everywhere and his hands shaking...
      • On the last page of Tom's advertizers section is a section entitled "Editorial content sponsorships"

        Thus the about face on Rambus and Intel in general.

        Tom "will never forget the deceit of Intel" revolving around the memory translator hub recall, the 1.13 Ghz .18 micron PIII recall and their support of Rambus when it was clear that Rambus was trying to assert a monopoly on ram.

        A few days after I noticed the "Editorial content sponsorships", he posted the editorial "proving" that Rambus was better than DDR because on a heavily overclocked system, DDR stops scaling.

        Tom is a heel. I once admired him. He brought Intel to it's knees by proving the 1.13 processor was factory overclocked.

        Now Tom is an Intel/Rambus/Bapco stool.

        At least Alex "Sharky" Ross had the moral spine to leave Sharky's when that site became an Intel ad.
        • Tom is a heel. I once admired him.

          Tom, is a FILTHY WHORE.

          He was in bed with 3Dfx, so proclaimed with supposed evidence that AGP gave no benefits over PCI. Then when he jumped out of their bed and into nVidia's, suddenly this was no longer an issue and he no longer ran benchmarks designed specifically to work within the SEVERE on board memory and PCI bandwidth constraints of 3Dfx products at the time (Voodoo2).

          His site, even showed a Matrox G200 running much faster than a dual Voodoo2 setup (during his 3Dfx days), with a benchmark that used textures too big for the Voodoo2's to handle without using the PCI bus. A MATROX G200 BEATING THE SHIT OUT OF DUAL VOODOO2'S?!?! If that's not AGP coming to the rescue, then what is it? Of course, he conveniently ignored that minor little detail.

          Anyone else know of any examples where Tom has made a sudden 180 degree turn in strong opinion immediately following a change in advertiser?

          Then again, could it just be that he is a complete moron? He certainly would appear to not grasp even the basics of CS, the good articles are written by others.

      • Tom wouldn't know computer science if he was drowning in a Cray.

        I remember back when the Voodoo2 reined supreme and he was in bed with 3Dfx, he claimed that AGP did not give any performance benefits over PCI, which is of course a great big pile of steaming horse shit.

        He had some article written by some supposed CS expert that tried to state just that, but without any technical merit at all.

        I thought it was pretty funny though, that on his own site, was a texture heavy benchmark showing a slow Matrox G200 AGP card running many many times faster than a system with dual Voodoo2's! This being due to the fact that the Voodoo2's could not cache enough of the textures locally and had to rely on the comparatively SLOW PCI bus.

        A performance hit so severe, that a slow G200 was even able to run much faster due to it's AGP bus.

        Of course, when Tom suddenly became an " Official nVidia review site! ", AGP was also suddenly much faster than PCI.

        Oh what power Tom wields! The power of ignorance.

    • I got system crc error on boot merely by changing a couple of memory options. This is not an universial guide for everybody and you should really experiment each option yourself.

      Just a little warning so that you dont' need to reset back to factory default. :)
  • I can't get to it right now, but I hope its all that and some. Whenever i'm in my BIOS theres always a few I don't get. Good find.
  • BIOS? (Score:5, Funny)

    by Lawst ( 242846 ) on Thursday September 12, 2002 @09:27AM (#4244459)
    Hmm, BIOS means Bandwidth Instantly Obliterated by Slashdot?
  • by The Moving Shadow ( 603653 ) on Thursday September 12, 2002 @09:31AM (#4244491)
    I always wondered about the fact that (almost?) no manufacturer supplies a manual describing their BIOS setup in detail. Most of them mumble something like: "you can press DEL to fiddle around with things you will never comprehend during your lifetime" and that's about as much help as you get. They of course also have included this neat *sarcasm;)* help function in most BIOS setups that displays the context sensitive help. I don't know how often i pressed F1 in vain just to see the message: "Help: Enable A-20 Gate. PG UP=on PG DN=off" Stuff like that...

    There sure has a reason to be for the lack of good documentation. The best manual uptill now was the one that came with my old ABIT KT7a RAID mobo, but maybe that's because back in those days it was considered a home "tweakers" board. So mr. Phoenix, Award, AMI, if you read this, please o please bundle nice manuals with your BIOS setups for us endusers to use, instead of hoping for great sites like Ars Technica and Tom's Hardware to help us out.
    • I can understand the lack of help within the BIOS; after all, you're trying to keep the BIOS footprint to a minimum and adding help text just increases footprint.

      However, that isn't really any excuse for not documenting it elsewhere, other than some ID10T getting ideas that tweaking the BIOS is a safe thing to do...

      • Why? AFAIK BIOS is in ROM and memory is really cheap these days.

        What I'd love to see in BIOS is a good disk partitioning tool and a memtest86 [memtest86.com] or something like it.
        • Why not just use LinuxBIOS [lanl.gov]? A minimal footprint system should fit onto a reasonable (32MB) flash part.

          Most BIOSes are designed to fit onto 128k or 256k parts. No real reason for this limit other than fact that these parts in volume are pretty cheap and they do the job required.

          A company I used to work for made some embedded devices where we put the BIOS, OS and applications all on a 32MB flash part. It was basically an i386 platform with some custom hardware and software. Heck it could even run Win9x off a disk drive.

        • A disk partitioning tool? OMG! That's a great idea. I wonder if PowerQuest has thought about an embedded version of their software.

          There are some problems if you think in the long term - a typical BIOS wouldn't necessarily get outdated, but one with support for specific file systems would. Or could.
          • Well, I didn't really have filesystems in mind. Even something as stupid as MS fdisk would work. The idea is that it'd be really handy when you mess with LILO too much and leave the computer unbootable. If the BIOS let me activate another partition I could just boot something and fix it instead of searching for a boot floppy or CD.

            Same for memtest86, if you have just one computer and the OS won't boot downloading it would be pretty hard. If some manufacturer made a decent board with embedded diagnostic and recovery tools I'd buy it as soon as I could

          • There are some problems if you think in the long term - a typical BIOS wouldn't necessarily get outdated, but one with support for specific file systems would. Or could.

            That's why BIOS is usually flashable, right? =)

            Seriously, yes, having a built-in set of better diagnostic and prep utilities would be great. Just think of all those Gigabyte motherboards out there with dualie BIOS on the board, and what you could put in all that extra space if you could somehow access it.

            On the other hand, while ROM does sound cheap, just remember that a $1 part undoubtedly costs more money to add to a board, probably more like $1.50 when you factor in the extra work in getting it on the board. Multiply $1.50 by 100,000 boards, and suddenly you're talking real money, and that's wholesale, way before retailer markup. Sure, hobbyists would be happy to pay the extra $1.50 (or even $5 or $10), but if they only buy 5-10% of the boards, the decision to include the part suddenly becomes much more difficult. Your average large VAR could care less about special BIOS options, and doesn't need to format drives in individual machines the way we do it because he has OEM licenses from MS, tapes of their latest OSes and software packs, and machines for batch-writing drives with the OS and software pre-installed.

            P.S. My dollar figures are used to show the issue of relative costs only, and are probably nothing like the real costs.
    • Those companies sell chips to motherboard manufacturers, along with 3-5 other companies. Motherboards are made in Asia, where English is not the primary language spoken. You will not be seeing what you are requesting any time soon. And if you do, they will not be very easy to understand either.
    • by Zocalo ( 252965 ) on Thursday September 12, 2002 @10:16AM (#4244817) Homepage
      I've always found Asus [asus.com] (they use Award's BIOS) to be another notable exception in this regard, and there are a few others I'm sure. It's definately not the the norm though, and BIOS help does indeed suck universally, except for some of those stupid BIOS-on-a-cutdown-Windows some Tier 1's used to use, but then, it's not hard to document the three available settings well, is it? :P

      Of course, since there are really only half a dozen or so BIOS vendors, the mobo vendor manuals are pretty much interchangable, and /. not withstanding, I've found several web sites post documentation for them besides this one via Google [google.com].

    • there is a simple reason for this and you may not like, system manufactures don't want people messing with the bios unless they know what they are doing and the easy way to do this is to make it as cryptic as possible, why? well many bios settings on some MB can easily burn out devices esp. ones about clock speed , they don't want some guy going "hey lets just crank up pci buss speed to 133mhz, shood make things faster!" and end up burning up a few pci devices. Its the same logic that MS has for the registry make it cryptic as possible so people don't mess with it
      • system manufactures don't want people messing with the bios unless they know what they are doing and the easy way to do this is to make it as cryptic as possible

        If they want it to be as cryptic as possible, they'd be using 0's and 1's only. :)

        they don't want some guy going "hey lets just crank up pci buss speed to 133mhz, shood make things faster!" and end up burning up a few pci devices.

        I guess I'm guilty. I can't remember how many times I looked at my 286's BIOS settings and thought: Why don't I just set my base memory to 2048K and extended to 0K and get past the 640K limit? (ok, it's not as bad as burning up a few PCI devices, but it does demonstrate the curiosity in all of us) It was back in the time when HIMEM.SYS causes my system to hang every now and then, so I was limited by software that knew about INT 15h and assigned all the extra memory to HD cache.
    • Bios "Help" (Score:3, Funny)

      by gnovos ( 447128 )
      I love the "helpful" BIOS messages...

      o ACPATI IEEE 9.0 compatability : ON | OFF
      Help -- This settiong turns on or off ACPATI IEEE 9.0 compatability.

      o CPU LIMIT PRIMER : FREE | POST | RETRAIN
      Help -- This setting sets the primer for the CPU LIMIT

      o DARNING PORT FLANGE : WITHIN | OVER | COMPLIANT
      Help -- This setting alters the darning on the PORT FLANGE
    • by pclminion ( 145572 ) on Thursday September 12, 2002 @11:56AM (#4245537)
      I don't know how often i pressed F1 in vain just to see the message: "Help: Enable A-20 Gate. PG UP=on PG DN=off"

      In case you are still mystified about A-20, it is an ancient holdover from when machines didn't have a high memory area. In 16 bit arithmetic, incrementing 65535 produces 0, and many programs took "advantage" of this fact. However on newer systems there was an extra address line, and if it was enabled this address "rollover" didn't happen correctly. Result: legacy programs crashing.

      A-20 is automatically enabled by all 32-bit operating systems anyway. The option in the BIOS is there to control HOW that enabling happens. Modern chipsets can enable A-20 directly. Historically, however, A-20 was logically AND-ed with a pin on the keyboard controller, so in order to enable A-20 you had to reprogram the keyboard controller.

      I love PCs, they are the only hardware I've encountered where you have to program the keyboard chip in order to enable high memory access. :)

    • I bought PC last year with a Grandmars i815 mobo, with a Phoenix BIOS. I put in my old hard disks, one for Win, one Linux, and powered on. The BIOS came up with a boot select screen to choose which OS to boot in... (not lilo, this was before that) I was pretty amazed and looked for some docs on this. Absolutely nothing on paper, on Grandmars or Phoenx's websites. (Lots on how to physically build it, though.) Nothing I could find on Google. I wondered what the point of putting in such a useful feature was if you didn't tell anyone about it. I never would have discoverd it if I hadn't had two bootable disks ready.

      So it's hard not to suspect something like what happened to Be, when they persuaded a manufacturor to install BeOS to dualboot with Windows, and MS perusaded them to remove nay mention of this feature from the manual.

    • So mr. Phoenix, Award, AMI,

      Note that Phoenix gobbled up Award about five years ago. They're now different product lines of the same company.
  • by stratjakt ( 596332 ) on Thursday September 12, 2002 @09:32AM (#4244506) Journal
    about the memory hole at 16M:

    "Sound Blaster Live cards like this to be enabled. It essentially removes 1MB of your RAM, so consider replacing the sound card instead."

    Yeah, it would suck to have only 511 megs available. I'm not giving up my SB Live any time soon, at least not till I decide to get Audigy. It does mention that this is for SB16 emulation, but doesnt clarify by saying you only need that if you want legacy DOS soundblaster support. It's actually wrong: SB16 emulation happens transparently, SB16 pseudo-emulated 'mode' requires this. (Booting into plain DOS rather than running in a Win2k/XP console)

    On the Video RAM Cache:

    "Disable this. You don't want to be wasting the L2 cache on fast video RAM when you have slow system RAM to deal with"

    Not every box has a sooper-dooper fast mega-card in it. I have boxes with old Cirrus Logic and Mach64 cards in 'em. And not every PC is equipped with AGP. Enabling this can yield a performance boost on some hardware, a little more detail here would help.

    I dont have time to analyse the whole thing.. It got slashdotted before I could make it through, and I'm not a know-it-all techie geek. I just have enough rope to hang myself with, as the saying goes.

    But like most 'BIOS' guides I've read, this gives alot of info on 'tweaks', with little mention of the damage that the wrong settings can do. I've seen RAM, PCI and AGP cards get fried because the user unwittingly 'overclocked' it.

    They always just tell you what the fastest possible setting is, but never mention "if your hardware doesn't support it, you'll wreck it". Personally I think sacrificing stability for the sake of a 1% theoretical boost in performance is bad mojo.

    There's also a disproportionate amount of Soundblaster-bashing going on here. Apparently my SB Lives have been crashing my systems and suffering poor sound latency for the last couple of years. Funny that I never noticed.
    • "Sound Blaster Live cards like this to be enabled. It essentially removes 1MB of your RAM, so consider replacing the sound card instead."

      Eh, I don't know about the Sound Blaster Live. The ISA spec (or common implementation) has some sort of a snag where you must do some sort of initialization routine (fuck, I'm not a driver programmer) in the first 16 megs of PC memory. As far as I know, this doesn't exist with PCI cards.

      I remember having a problem insmodding my SBPro for a very long time, but having no problems when it was compiled into my kernel. I finally got around to learning about this.

      Problems ceased when I got a PCI soundcard. (SBLive Value in fact)

      • yeah, thats true.. There's a backwards-compatible mode that SB Lives can use to act like an SB Pro for the odd guy who still runs pure DOS and needs to play Duke Nuke'm. Basically a 'pretend your an ISA card' thing, and IIRC it loads a driver (sys file) into this 1 meg hole to look like an SB Pro.

        I'm no driver progammer either, but I know that this setting doesn't affect SoundBlaster Live under linux, WinNT/2k/XP. Or even 95/98/Me for that matter.

        The article suggests that SB Live is somehow flawed in this respect, which to me just sounds like the author is making up a reason to hate SoundBlaster.

        (The SB Live does have legitimate problems which are beyond the scope of the article. It doesn't like to co-exist with a secondary sound card, for example.)
      • As far as I know, this doesn't exist with PCI cards.

        Problems ceased when I got a PCI soundcard. (SBLive Value in fact)


        I'll second this. I never had problems leaving the "16MB hole" option off when I was using a Creative ISA card. However, I've never needed it for my SBLive! 5.1, either. This is on Linux...I'm not sure about Windows, yet.
    • On the Video RAM Cache: Not every box has a sooper-dooper fast mega-card in it. I have boxes with old Cirrus Logic and Mach64 cards in 'em. And not every PC is equipped with AGP. Enabling this can yield a performance boost on some hardware, a little more detail here would help.

      I'd be surprised if you ever saw a speedup for normal usage. Most VRAM writes are done via an acceleration engine/blitter that has no ability to use the L2 cache. And, for normal writes, good MTRRs [tldp.org] will take care of the rest. So, they're correct: it's almost always wrong to use L2 to cache VRAM. Disable this option.

      Note that this is not the same as caching the video driver in system RAM. Back in the DOS days, the entire video driver was contained on the card in slow ROM. This option would cause that ROM to be copied to the system memory, offering a big speedup. However, all modern OSes use their own video drivers (assuming XFree86 is part of the OS). Caching the ROM driver in RAM only wastes RAM. Disable this option too.
    • by Zathrus ( 232140 ) on Thursday September 12, 2002 @10:47AM (#4245001) Homepage
      And not every PC is equipped with AGP

      Well, every PC built in the past 3 or 4 years is... and, frankly, if your main PCs are older than that you're not likely to be reading ArsTechnica or something about a BIOS tweaking guide.

      But like most 'BIOS' guides I've read, this gives alot of info on 'tweaks', with little mention of the damage that the wrong settings can do

      Obviously a problem... I haven't read the article yet (didn't feel like it this morning, and it's toast now), but they should really mark the settings that are potentially dangerous. Screwing around with your RAM timings, CPU clock, etc. can release the magic smoke awfully quickly.

      There's also a disproportionate amount of Soundblaster-bashing going on here

      Not really. Creative Labs has long made the worst hardware they could get away with, and did so thanks to having created the original standard for PC sound. They've never been high quality cards, and have often caused problems with other hardware and software. Go talk to someone who tried putting an SB Live in a dual processor NT4 system about it for example.

      Frankly, if you're looking for a new soundcard then there's little reason to buy Creative. For general use (games/music) both Hercules and Turtle Beach make better cards for less. For games alone, Hercules or Philips are better (Philips mentioned purely due to QSound). If you're talking about just playing music, doing a home theater PC, or mid to high end audio then a more expensive card that does real 24/96 or 24/192 audio is preferred - M-Audio and many others fit the bill here.

      I do think that the incompatibility bit is somewhat overstated (I don't have any problems with my SB Live or my much older SB64 ISA), although SB's are notoriously bad about sharing PCI IRQs and the like, but the poor sound quality and total lack of compliance to industry standards are not. The digital out on the Live series doesn't comply to any spec known to man -- its voltage is roughly 10x the allowed spec. Even the Audigy continues to resample everything to 48 KHz, which plays hell with CD Audio, and their claims of 96 KHz sampling rates are deceptive at best (only applies to the digital outputs, and only sometimes at that).

      If you want more details, I suggest either the PC AV Tech [pcavtech.com] or [H]ardOCP's Audio forum [hardforums.com]. If you're interested in HTPC's in particular, then take a look at AVS Forum's HTPC forum [avsforum.com].
      • Thats no joke!

        Creative Live/Audigy drivers have never worked properly on SMP machines, regardless of the operating system.

        I have gone through all the Windows' (NT/2000/XP) and the Linux driver on several dual machines with different hardware and none of them work. The sound cracks, pops and skips randomly. It seems to randomly spawn multiple instances of the driver, one on every processor, which then conflict with each other. On Windows-machines it helps if you turn down all acceleration features but its still not flawless.

        They are aware of the problem, have been for a loooong time, but just don't care. I wonder what happens when the "hyper-threading" Intel processors hit the market? Might get interesting :-)
    • My SBLive doesn't do any of that nasty latency-related stuff either... but normally you'll only see that on VIA chipsets (it's a well-known issue there) and I run pure Intel chips. And mine works fine so long as its lame-assed DOS emulation is disabled. When that's enabled, it NUKES Windows, and doesn't work anyway.

      As to the 15/16MB memory hole, IIRC (and IIUC) when it's enabled, it prevents any DOS program from using memory above that point, because DOS can't jump the gap like a fully protected-mode OS can. If you run big DOS games or databases in real DOS, this can be an issue, in which case the memory hole must be disabled or stuff won't work or will be really slow due to swapping to disk when it runs out of the first 15MB of RAM.

      So it doesn't exactly "remove" 1MB of RAM; it limits DOS on your system to a mere 16MB usable RAM, no matter how much physical RAM you have.

      Some older memory managers do the same thing.

    • There's also a disproportionate amount of Soundblaster-bashing going on here. Apparently my SB Lives have been crashing my systems and suffering poor sound latency for the last couple of years. Funny that I never noticed.

      Lucky you. The SB Live crashed my system. It instantly destroyed my previously rock-stable W2K installation as soon as I plugged in the card and rebooted. I'm not touching that crap anymore. Anyway, keep on laughing for now, but keep in mind who to blame the next time your system goes horribly wrong.

  • great! (Score:2, Funny)

    by jeffy124 ( 453342 )
    now if only Ars Technica would apply those BIOS settings to their servers, they wouldn't be slashdot effect victims, and I'd be able to read the article!
  • but I should have a pretty hefty server. I was copying the whole thing into a text file for myself. So, let's see if my server gets slashdotted. It's only 15k, so I hope not, besides, I need to use up my 40 gig of through put this month.
    ArsTechnicaBiosGuide.zip [bankholdup.com]
    • @time it still works hehe
  • by tangent3 ( 449222 ) on Thursday September 12, 2002 @09:44AM (#4244588)
    Optimizing your BIOS settings is not enough to prevent your server from being slashdotted...
  • Wow, I guess somebody figured that since ars [www.arstechnica] could surive one slashdotting you'd hit them with not one [slashdot.org] but two [slashdot.org] stories on /.'s front page at the same time.
  • Seeing as how the Linux kernel replaces most of the functionality of the BIOS [lanl.gov] will setting any of these options really make a difference?

    Any kernel developers out there care to chime in?
    • by Anonymous Coward
      Well, some of the options will, such as DRAM timings, but Linux reads a lot of BIOS settings at boot, then promptly re-configures them in a way which is optimal for Linux.

      To be honest, you are not going to see a massive improvement in performance by tweaking your BIOS settings anyway.

      The most effective way to increase performance is to use the most recent kernel and gcc versions, with relevant patches.
      • hmm well 15-20% is a lot to me, this is about what I got when going from factory default to tweaked settings on my current gaming box, or about the same as dropping another $100 on the next best cpu. Since it took me about 15 minutes to figure out the best settings thats a really nice hourly rate =)
  • by Anonymous Coward
    AFAIK, the BIOS is a piece of archaic legacy... why don't put it away and choose something better?
    Can't PCs use OpenFirmware or some other more flexible technologies?

    --
    I'm not an Anonymous Coward, here's my email: mosfet@ig.com.br. I'm just too lazy to register ;)
    Sorry for my poor english!
    • There is this effort OpenBIOS [linux.de] which tries to do just that. It's an independent implementation of Open Firmware, which is the boot code (i.e. BIOS) used on Apple and Sun machines (amongst others). What's really cool and geeky about it is that it's a FORTH system. There's a small nucleus of words coded in native assembly language and the rest is all portable FORTH!
    • I dont know, maybe couse we want to run an operating system on it, and the OS needs a Basic way to access the input and output on the System most modern OSs use it for start-up and for fall-back mode if a device doesnt have a driver, as far as im aware EVERY system has a bios including macintosh(you just cant boot into it to change the settings), once the OS has loaded and started its own method of accessing the hardware the bios isnt needed anymore(unless the device doesnt have a driver and can beaccess thrue the bios)
  • by LuxuryYacht ( 229372 ) on Thursday September 12, 2002 @10:00AM (#4244700) Homepage
    Replacing your legacy BIOS with LinuxBIOS [onelabs.com] yeilds the best overall system performance gains.

    SPD, ACPI and PCI init and config is still quite a mess these days. Using an open source BIOS allows system performance to be tuned and maximized beyond what the usual legacy BIOS setup screens offer.

  • by Spencerian ( 465343 ) on Thursday September 12, 2002 @10:06AM (#4244747) Homepage Journal
    The BIOS of the x86 world, in my opinion, is one of the reasons why we struggle but never quite reach a integrated architecture for PCs. Lord knows I've fought with quite a few of them, and hated having to remember to disable this in order to use that, with no guarantee that my change would work all the time.

    Shouldn't our computers know what hardware it holds and configure itself automatically nowandays, with little to no user interaction? It would make all that "plug-and-play" stuff that's taken for granted on Macintosh systems, to site an example, true for my PC game box as well.

    The technology is already here in the form of Open Firmware [openfirmware.org], which Apple uses as well as Sun. There is at least one company that has OF implementations for x86 [firmworks.com], but so long as Intel has a vendor lock on how motherboards are designed for their chips, I don't see this annoying and archaic method of maintaining a board going away any time soon.

    OF is configurable enough for crazy whiz kids, if necessary. A better BIOS would make things a lot better for the OS and bring a better experience. Why can't we break out of the BIOS hell? Hadn't we learned the lessons from the Y2k-incompatibilities that some BIOS had, among other headaches?
    • There's a free implementation being worked on here
    • that was kinda true back in the 486 days, but most modern pcs you dont have to touch the bios if you dont want to ,hell the only configuration I do to a bios is setting up the HD info so the system doesnt take 30 seconds to auto-detect, could I tweek my bios to get a little more speed, sure, but I like how stable my system is and dont want to mess with it, as for macintosh just couse it auto-configures for you does it meen its set for the fastest speed, it could be but I dought it, of course its faster to auto-config on a mac when you have full controll of the hardware and know what the hardware is its easy, but in the x86 world of open hardware and bios's being genaric they have to acount for all your hardware that the bios wasnt written for
    • OF has its own set of problems. Most of which are much harder to overcome then the little anoyance the PC BIOS's provides. For example OF is written in an old Forth dialect and doesn't support interrupts. This means that all your nifty new high speed adapters need to run in polled mode. Quite a problem if the adapter hasn't been designed with polled mode in mind (most of them nowdays).

      How about the fact that OF is really just a fancy PCI configuration and driver enviroment. This means that half of the work done by a modern bios/firmware (processor config/selftest, ram config/selftest, initial PCI bridge detection/selftest, etc) accually has to be done before the OF enviroment can even be started.

      Anyway, OF was a great idea that never really had the kinks worked out of it. Like java, its cross platform promises never took off except in a few cases because the manufactures (Sun, IBM, Apple, others) couldn't get the basic services to behave similarly enough to the spec that OF pci cards just worked between vendors. Try it, buy some sweet fiber channel cards from Sun and try plugging them into your mac. I will bet money they won't all work. Try plugging a mac video card into your AIX box, I'm sure you can't find a single video card that works in both the mac and the aix box.

      • I have never experienced any problems with PCI cards and installation with OF. They just work. I know that sounds like the typical Mac-guy crap, but hardware installs are really made very strong by OF. Your comments seem to me that OF still would have a problem in handling interrupts and other bugaboos of a BIOS, which makes it sound like the problems are more inherent in the motherboard or the OS than the BIOS.

        Sure, not everything works. Video cards are a good example since they usually require complex drivers. But interface cards such as SCSI work quite well, and there are numerous PCI cards that work in both platforms. I have a Fibre Channel card that is certified to work in both Mac as well as other systems. It's not that PCI devices can't be cross-platform--but most vendors don't write the drivers.

        Right--OF isn't the "boot" of a system. On a Mac, that's handled elsewhere. So maybe mobos need to look beyond the BIOS for a better boothandler.

        I have to disagree with you on one point: OF still IS a great idea...and is actively in use. But your points hold in that it may not be the best answer for this topic.
    • by Anonymous Coward

      ...to site an example...

      The word is cite, fucknut.

      Get it right.

    • The problem isn't Intel. If anything, I'd bet they'd love to drop the BIOS too. Do you have any idea how hard it is to boot strap a PC from bare metal? There are tons of legacy tricks and almost secrets, it's an amazing excercise to go through; if you don't have one already, I'm not sure where you can even go and buy a high quality 16bit x86 assembler any more.

      The problem is windows and microsoft. They can't boot without a BIOS. DOS, you know? Can Linux? It's pretty damn easy to boot Linux without one, pretty much put a kernel in memory and jump to the start of it. (No shit, it's about that easy) MS simply has never put a lot of effort in to getting away from the BIOS, they control the PCHEC specs, they can if they want.

      More importantly, you can spend some change on a BIOS, last x86 embedded project I worked on we used a LinuxBIOS type BIOS because it was going to be $4 to $12 a unit to buy one off the shelf, plus something like $30k to $50k up front for their developers docs. When you buy a new PC a BIOS can be an expensive part.

      I kind of suspect that MS likes it this way becuase it's all defined. The rules of engagement are pretty simple. There are only a dozen or so BIOS vendors. They work off of DOS, that's the benchmark for the most part. There are 2 bigger vendors who own the majority of the market. They all race to support new hardware and make DOS run on it. You take a step back and the only people that helps at all is MS. Nobody else needs DOS that much. You change the architecture and who knows who get's powerful, maybe somebody that's not friendly to MS. Expect it to become more of an issue as DRM becomes more central to home computers.

      • Good points. It really is a vicious symbiotic relationship with Microsoft and Intel. Some Mac zealots, for instance, see Intel as a "villain" as many see MS. I agree that MS has more pull in maintaining the status quo. Intel just makes chips, and I agree with that they would kick the BIOS out as well. In fact, haven't they tried in various concept motherboard specs?

        So it all goes back to the Linux/OSS "manifesto": Let MS burn, and the solutions will present themselves from elsewhere.
  • by xTK-421x ( 531992 ) on Thursday September 12, 2002 @10:13AM (#4244800) Homepage
    If I have an obscure Taiwan motherboard, this place almost always has a link to find the latest BIOS for it:

    http://www.wimsbios.com/ [wimsbios.com]

    I'm sure it's old hat for most people here, but some people will probably need it to find their latest bios to use this guide.
  • by sc00p18 ( 536811 ) on Thursday September 12, 2002 @10:34AM (#4244923)
    Go relax with some good ol' Adrian's Rojak Pot. [rojakpot.com]
  • by pepperino ( 608113 ) on Thursday September 12, 2002 @11:12AM (#4245190)
    Is it under File or Edit? I didnt see it under View or Favorites. I clicked on the Help one but it just laughed at me.
  • by Anonymous Coward on Thursday September 12, 2002 @11:26AM (#4245302)
    "Back then, we didn't have drivers"

    yes we did.

    "Long before the term "API" was born"

    I have a hardware book on the original 8088 IBM PC published in 1985 that includes specs on the bios and boot loader software, and talks of APIs.

    "You could call BIOS ROM routines to handle video, PC speaker beeps, the keyboard, and so forth."

    That is, if you only wanted your program to be able to run on that ONE model of PC. To get to hardware, nobody programmed to teh BIOS, they programmed using DOS interrupts.

    "Remember, in those days, people didn't build their own systems"

    Yes we did. I replaced the mobo on my XT with a 386 mobo. By then the only original components in that box were the case, power supply, and keyboard.

    The original 8088 mobo used jumpers and dip switches where we now make the adjustments in BIOS.

    Maybe he's a whiz on modern BIOSes, I don't know. But he's sure clueless about primitive PCs. Knowing this, I have to take everything he says with a grain (or whole damned shaker) of salt.
    • He may be an AC but he's right on, aside from a common typo or two.

    • That is, if you only wanted your program to be able to run on that ONE model of PC. To get to hardware, nobody programmed to teh BIOS, they programmed using DOS interrupts.

      Come on, BIOS interrupts were more just as standard as DOS interrupts. And you needed them if you were writting an OS, or bootloader, or boot virus or something... There were a few things you could only do with BIOS interrupts - change the screen resolution!!!!
  • BIOS == Built In Obsolete Software
  • I had massive bios issues on my shuttle AI61 board with my soundblaster card. it's all about choking shit off it seems.
  • But will my windows machine reboot faster?
  • A quote from the article: SDRAM Command Leadoff Options: 3, 4 Yet another setting that's faster lower. Set it to 3 if you have badass ninja RAM. Set it to 4 if 3 doesn't work. Come on now... I think most people reading this article expect and deserve at least a *little* more technical explanation than that! This article is chock full of non-explanations.
  • "Slashdotting" = disable.
  • I came across a page once that showed me how to change the EPA picture that's on the top right corner of the screen when you turn you pc on.
    It will not increase speed or stability, but it's a nice little trick.

    Personally I have a neat little picture of Tux, but for people that sell computers, I would believe that it's a nice way to advert their little company by having their logo there.

    Description more links can be found here:
    http://users.cybercity.dk/~dsl6178/bios/Bio slogo.h tml
    or
    http://www.biosmods.com/epas.php
  • by raytracer ( 51035 ) on Thursday September 12, 2002 @12:58PM (#4245967)
    Honestly, isn't your brain full of important things to think about? Wouldn't you like a machine which didn't needle you about details of its own operation that you don't understand? When you hop in your car, you don't expect to have to set fuel intermix ratios and timing to get out of your driveway, yet your computer manufacturer seems to think that you probably know better than they do what all these settings should be.


    Consider the memory options listed in the article. Do you know what the 15M-16M memory hole is? Autodetect DIMM/PCI Clock? Bank Interleave? The article says Bank Interleave gives you a massive performance benefit, why then have an option to turn it off? What's the point?
    Data Integrity Mode? Don't you think it would be nice if your computer knew whether it had ECC ram in it or not? Delay DRAM Read Latch? The article says that if you don't set it right, you can get crashes in your machine. Golly, don't you think manufacturers should just make the computer get it right?


    Memory options go on and on and on. The only thing I want to know is that my computer can read whatever memory is installed at the highest reliable speed. I shouldn't have to tweak the no less than two dozen different settings to
    get my machine working reliably, and those are just the memory related ones. A similar number awaits in the PCI and AGP configuration settings.


    Legitimate uses of the BIOS are perhaps to enable or disable peripherals and to choose boot devices. It might also be nice to have a mode which shows what peripherals are installed. Other than that, I'm perfectly willing to allow the computer to pick out its working parameters. If the resulting computer proves to be unreliable, then the manufacturer should be out of business for making crappy computers.


    Rant concluded.

    • God forbid some option might have both good effects and bad effects. And since when have you had to tweak two dozen BIOS settings to get your "machine working reliably?" What kind of machines are you buying that just don't work until you've done that?
    • Buy an IBM PC 300PL - it exactly meets your requirements,

      And it will run any P2 or P3 processor (with a suitable BIOS upgrade.) No need for those really hot P4s. And you can flash the sign-on screen to be your company's logo

      "Stolen from www.ourco.com"

      might be suitable text or maybe

      Linux stops here

    • What are you complaining about? You're saying new computers from all-in-one providers like Dell shouldn't need BIOS tweaking to start working? Well guess what, they don't. But the BIOS doesn't hurt anybody, and it's still necessary if you want to plug old non-plug-and-play peripherals into your PC.

      Yeah, yeah, if PCs had been properly designed for true plug-and-play from the start, BIOS configuration wouldn't be necessary. But they weren't, and it's no use blaming today's manufacturers, who are stuck with it. If we removed the BIOS, it wouldn't benefit ordinary users who never have to touch it anyway, and it would make it hard or impossible to

      1. use legacy hardware on PCs
      2. install alternative operating systems who need different settings than the default.

      No wonder Microsoft is pushing BIOS-less computers, like the new Toshiba laptops which are painful to install Linux on.

      In sum, BIOS access is harmless and necessary, and I can't see what you're complaining about. Do you think the hood of your car should be soldered shut too, so that you never have to look at your fearfully complex engine?

  • After reading this article, does anyone get old memory flashbacks to the days of Int-13 and Int-20? :-)

  • Just add it to your lilo menu :) (You know you've done WAY too much tweaking on your system when memtest86 option of your lilo menu is the first one.)

    Some advice for people on memory tweaking:

    Don't boot a real OS unless you like to reinstall often. Even XFS, EXT3, ReiserFS, and especially NTFS will corrupt if you can't trust your own memory. Instead, boot the Memtest86. Don't stop there!!! Boot Linux and compile your kernel while playing Quake3 :) If you can do that, there's no problems with memory timings.

    Don't forget that you can underclock your CPU and get better system performance overall by having faster ram. A lot of your computer's CPU cycles are wasted waiting for memory. Change the system performance option from "Optimal" to "Turbo" if you have it. Then keep trying tests until they all pass. Adjust your system bus speed down each time. Once you have a good setting, set it down a little more.

    Make sure you don't go too far out of the PCI 33Mhz standard, or don't use an intel video card or various other cards that depend on this heavily. I had an I740 video card that just wouldn't work at any other setting.

  • by fotoguzzi ( 230256 ) on Thursday September 12, 2002 @03:50PM (#4247194)
    I was poking around the Vcore settings on my ASUS board with an Athlon XP 1700 processor. The settings range from the expected 1.675V - 1.85V, but if you hold down the righthand Alt key and then Page Up and Page Down, you get values ranging from 500kV to 1.5MV in 50kV intervals.

    Does anyone know why these settings should exist?
  • I suggest that you don't change any of your RAM latency settings, and if you do, make sure you test your system very thouroughly before you trust it.

    The latencies (and a variety of other stuff) are spec'd by the RAM chip manufacturer (which means you can look them up for yourself if you read the chip number off of the RAM chips on your DIMM) and stored in a small ROM on your DIMM. Whoever designs the DIMM has to put the right info in the ROM. Then, during bootup, the BIOS is supposed to read these settings from the ROM using the SMBus protocol, and configure the chipset accordingly. This whole process is called Serial Presence Detect, or SPD. It is mentioned in the PC-100 RAM (and subsequent) specifications. In fact, I think it is now a JEDEC spec.

    While I don't know for sure, I would guess that most DIMM and BIOS designers did this right. (I know I did when I had to do it ;-)

    MM
    --

It's a naive, domestic operating system without any breeding, but I think you'll be amused by its presumption.

Working...