Home > In the News > Writing an end to the bio of BIOS Writing an end to the bio of BIOS Submitted by Josh 2003-12-30 In the News 39 Comments Intel and Microsoft are gearing up to move toward the first major overhaul of the innermost workings of the personal computer–the boundary where software and hardware meet–during 2004. About The Author Eugenia Loli Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker. Follow me on Twitter @EugeniaLoli 39 Comments 2003-12-30 6:00 pm the problem lies in that a few have decided that pci slots and motherboard componentes, WILL share IRQs, end of story. even if you disable com1, com2, lpt1, and usb. can you give those free IRQs to a device that needs it badly? Like that capture card in your last free PCI slot, slot4? nope. and what’s the deal with, for example, pci slot3 shares the irq with onboard mass storage, AT ALL TIMES. sigh. 2003-12-30 6:02 pm *Clean* 32-bit Operating Systems do have trouble wiht the current 16bit BIOSes. Ask BeOS and its troubles. So yes, there ARE reasons that are wrong with the BIOS. BIOS is a legacy thing, as long Intel can get rid of it without creating major problems, the sooner the better. 2003-12-30 6:35 pm The BIOS has hardly advanced at all since the first PCs. I’m glad that it is being replaced. I’ll be very pleased when I’ll finally be able to run hard disk utilities outside of an OS. It will be wonderful to be able to perform a defrag on ALL of my files, instead of the limited set I have to put up with, because Windows has so many of them in memory. Ah, it will be nice, just like the good ol’ DOS days. 2003-12-30 6:38 pm An evil MS or Intel invention? That’s the problem; all the companies out there want to impose their will on the PC market and close it as much as possible. I do not really know until what extent the bios is a problem, I never heard or read about many people complaining the bios being such a pain. On the other hand the desire of Intel and MS to dictate standards and control the market at their will is well known. 2003-12-30 6:48 pm The average joe that just messes with the configuration doesn’t really have a problem. They may complain that a certain feature is not present (i.e. 1MHz incrementS)But other than that, they don’t care so long as they can make the changes and it works. For others who have to program the BIOS’s or program for use with it (OS dev’s etc) it would make their life much easier. It would allow for modularity which in turn would allow for faster development of hardware compatibility. You think mobo’s are coming out fast now? If they had something that would decrease the engineering time of the mobo by 50% (guessing here) they could practically make a motherboards life last longer as the BIOS(EFI) would be able to be updated for new features easier. Mobo’s may be a bit more expensive because of this (supply and demand) but I probably wouldn’t have to drop coin as often. It would save them money in time, and that MAY trickle down to us.. MAY. The dynamics of the consequences of this are as yet, unlcear (at least to me). Change is good! 2003-12-30 7:03 pm Pro: – The BIOS *is* antique and needs a replacement. Although I’m not even close to an OS designer, I can imagine the old BIOS has become a load of patch upon patch upon patch upon workaround upon… well, you get the idea…. – If the OS could leave all the initial loading of bare bone drivers etc. to the EFI (?), you could create a blazing fast startup…:) – Coding for the BIOS is becoming easier, due to the absence of the need to do it in Assembler. – Now you *can* customize your EFI fully! Con: – I can imagine a machine that refuses to boot any other OS than the OS it is designed for (conspiracy theory engines get running ;-)…) – Much easier to trash the PC because of a bad guy that gets into your workstation and executes some code (or: ‘Please open this attachment, it contains your new update from Microsoft’. Joe Sixpack does clickety click and thus renders his WS useless. I see greeaat new horizons >:-) Just some thoughts… grtz & blessings, Joppes 2003-12-30 7:08 pm @Joppes. I don’t think the new EFI will be a win32 or .NET based environment. I don’t think any standards body would allow that as it’s too proprietary. 2003-12-30 7:12 pm I do agree that the EFI may allow a virus to do an incredible amount of damange. Though, I would imagine that Intel and Microsoft would try to prevent this from being a possibility, as everyone would point the finger at them immediately, if something went wrong. I doubt Intel would design EFI to only work with Microsoft OSes. Intel has spent a ton of money on Linux, and I believe is a major backer of the Linux organization that Linus Torvalds currently works for. They’ve also invested lot of money into Be, Inc, so I highly doubt their goal is to make it impossible for PCs to run anything other than Windows. 2003-12-30 7:25 pm I doubt this is quite what they have in mind but it just occured to me, wouldn’t it be great for alternate OS’s if drivers were written for bios not OS? 2003-12-30 7:32 pm The first itanium systems which were sold (those workstations which everyone was forced to sell by intel) had EFI on. It worked very well. –ralpht 2003-12-30 7:37 pm There are others (beside the Wintel folks) who strive to improve the BIOS. One example can be found at the following link : http://www.linuxbios.org I should mention that the linux BIOS is being developed at the Los Alamos Laboratory. So, those are neither script kiddies nor garden-variety coders. 2003-12-30 7:41 pm Actually that’s not a bad idea. Would it be safe to say that you would pretty much be making sure that all hardware was compatible with the EFI, and thus the OS (any OS) would only have to communicate with that? Wouldn’t that pretty much eliminate hardware compatibilty across OS’s? Would there be a HAL or would there be some other type of hardware layer involved. How would that effect performance. Asking a curious and legit question here. 2003-12-30 7:46 pm First: I, on purpose, used the term ‘the OS it is designed for’ instead of, for example, Windows. That’s because I simply could fill in any other OS-name (freeBSD, GNU/Linux, SkyOS and the rest of ’em). I was by no means trying to point to MS Windows (well, errrr, I *was* thinking about that ;-)…) That said, I would like to continue to the next point: Q. *Can* the EFI insist on booting just one, predefined, OS? A. I think it is possible. The thing is designed to make it possible to let 3d partys create additional tools and to incorporate ’em in the EFI (that’s what Extensible means, though?). Then, how much more difficult is it to create and load a tool that prevents loading any other OS than the predefined one? Is there a BIOS/EFI expert in the room? 2003-12-30 7:56 pm IMHO, an extra HAL would create some performance loss, but if that means that you can remove one ore more layers in the OS because you can leave tasks to the EFI, you get an overall better performance. 2003-12-30 8:10 pm Whatever they plan to use, better it to be an open and well documented standard, otherwise it won’t be nothing more than another Intel/MS strategy to collect royalties from everybody. And by the way do not even think that 50% of a motherboard complexity resides on the BIOS. 2003-12-30 8:22 pm time and complexity are two separate things and are typically directly related, I agree but I took a guess. A wild one I suppose but a guess none-the-less. In something like designing a mobo I would gather that the BIOS would have to coincide with the hardware features on the board. In which case this is a two part process, no? (I’m no expert and I won’t claim to be..just trying to be logical). Being a two part process wouldn’t it be safe to say that total time would be cut down dramatically (if not 50%) if one could build an EFI and only have to build onto it the required modular parts for supporting newer features or hardware. This would make thing much more interchangeable. EFI’s could be used with different mobo’s with the same chipset etc.. reference EFI’s could be developed. The possibilities are endless. Overall, I would say that the development process would be greatly reduced and I dare say that it’s not out of question that 50% could be possible. Anything’s possible. 2003-12-30 8:25 pm Apparently you guyes don’t fully read the setups. EFI uses a DOS partion on your harddrive in order to function right. You know the FAT type partion that MS is going after infringers for using. Also PowerPC has had irq’s gone for a long time now and they currently use Open Firmware(I think) for their hardware intialization. Yes microsoft would limit the OS to only their OS and then fix the ‘feature’ later after a lengthy legal battle. 2003-12-30 8:40 pm First of all I do not understand why everybody is bitching so much about this. Who else is going to take the initiative to go out and redisgn a major component of the hardware side that every PC on earth deals with. It is not the strugling Linux companies that are throwing away their desktop products so they can try and make a buck and it definitly isn’t the companies that don’t even make a revenue. Let me ask you this question and I hope somebody has the balls to answer it, honestly. Who in the world of computing has enough force behind them to make a major change like this? IBM, maybe, Compaq, maybe. But this isn’t like you are just adding a software layer on top of hardware. This is chaging the very essense of how our computer will function at it’s most basic level. I think there is only a couple power weights in the industry that can ever do this, Microsoft, IBM, Intel, Dell, and Compaq. Those are the only five companies that have the force and money to force this change uppon the market. 2003-12-30 8:55 pm looks as though you are correct. Compaq tried this (putting bios config on hdd) and it was horrible. I don’t really want to mess with making EFI partitions period. I want to be able to put a new drive in and forget about it. This kind of advancement should be on silicon.. not the hard drive. A great idea poorly implemented. NEXT! 2003-12-30 9:40 pm openboot firmware would have been much better. then all cards across all major os would have shared fcode drivers for startup! do we unify all cards across 1 io model? noooo lets create EFI!! i mean, its not like we tried with ACPI and made a mess of that with its interpreter… but.. well lets try again with an EFI interpreter!! honest, we wont repeat the mistakes we made with ACPI.. really… 🙁 macs, ibm bigirons, suns, etc all been working with openboot for years… but no, we cant embrace an existed _good_ ieee/ansi standard…. honestly.. it boggles the mind. some companies just MUST have control of things…. fingers in pies… 2003-12-30 9:44 pm Peragrin: Intel states that it will be OS neutral, I agree that it is likely that the Mobo’s will ship with the OS MS loader flashed as standard, afterall lets face it that would suit the majority, albeit not likely many of us here, but as long as it’s possible for an alternate os to flash it own OS loader(as it appears in Intel’s doc’s) I don’t see a problem with this. 2003-12-30 11:39 pm if my motherboard forces me to use Windows, that’s definitely the day I move to PPC. I might even move to it before that. seriously, the idea of loading the device drivers independantly of the OS from the BIOS-to-be is a great idea. Althought that brings another new set of possible problems. Do you really want average joe to flash his BIOS to install a new driver for his nifty new video card? Not really. Also, do you really want MS to be in charge of the BIOS? That’s a whole new set of remote security exploits that will be possible. No thank you! Fortunately for us, there’s still other architectures out there. 2003-12-31 12:50 am What’s going to happen is that the EFI bios will be able to load any OS, yes any OS Intel or MS dictate. Both companies controls 90% of the computer market soft/hard, both companies have a well justified reputation of monopolistic behavior. Honestly I do not trust them, they will manage to f**** the thing somehow. Most of the modern problems on a PC are the responsability of MS, the stupid mouse and keyboard protocol between others, that clumsy ACPI, that shity USB… MS has never, never fixed anything when they had the oportunity unless they have something to earn on the operation. Intel is one of the big backers of the “Trusting computer” concept which is shared by their fellow MS… This doesn’t smell good. 2003-12-31 1:48 am then all hardware would work on all computers from Sun stations to Macs to Wintels with no modifications. 2003-12-31 2:39 am The idea behind EFI is not so bad. But I dont trust the TCPA-members. And how open will EFI be if you need an license from Intel like Inside Software? Especialy I doubt that boot-time will degrease by running some kind of JIT in the BIOS. 2003-12-31 2:51 am I think the bios should be as light as possible and leave tasks to the OS where it can. I don’t want to have a bios-OS forced upon me. 2003-12-31 2:53 am Open Firmware seems to do ok. it lets the hardware vendor program the amount of functionality right into it. 2003-12-31 3:36 am i dont like the fact that with efi you can install only one os in a hard disk 2003-12-31 4:18 am With EFI in place the pc will be fully ready for MS’s Palladium… As far as I can see this will end up as the foundation for its hardware DRM system… As for having it on THE single most failiure prone componant on a computer… thats just asking for trouble. Openfirmware does look very nice and with this EFI stuff, I suspect that it will begin to look even nicer. Guess I had better start saving up for that Mac as this hardware won’t last forever and I doubt very much that this is anywhere near as benign as either IBM or Microsoft would have us believe. And yes, I’d bet good money on that. 2003-12-31 5:12 am DRM has detected you have ( X amount) of unpaid media on your Hdrive. Please enter credit card details now… Agree ( Payment deducted ) Disagree … Computer will fail to boot. Love that DRM. Glad I’ve already got a viable alternative to my x86 platform. Viva le AmigaOne .. Think Freedom !! 2003-12-31 10:15 am You could always pay for your media What right have you to criticise MS if you have ill gotten gains on your comp? Seriously though that is equally possible now, fortunately we have laws that prohibit this 2003-12-31 11:48 am I’d have chosen Open Firmware myself, not that it matters. If this does turn into the evil thing that so many of you think it will, I’ll switch to some other hardware, Macs for instance. 2003-12-31 2:05 pm The obvious way to do that is through an open (non proprietary ) specification that others can also implement. The reason why it won’t be is because MS and Intel want to lock down the PC so that they can charge music and movie companies for the protection technology. The problem has been around for a long time now. How come they suddenly want to fix it ? 2003-12-31 2:35 pm But what if you own that media ? As in, you created it. Lets remember that M$ doesn’t code for the pissants, so, You haven’t had your media D.R. Classified by M$.. So its unsigned …therefore, (in M$’s eyes) its unpaid for. So .. You want your creation to have M$ D.r classification … Are you in favour with M$ YES .. goody for you .. Join the fast track queue No .. Ah .. Join this eternal queue. Either way ..Remember to have your credit card details ready … But YOUR the creator of this media, Surely its Joe Public that should pay to use it ? (M$ reply) But of course they MUST pay, We want our 10% from them for that priveledge just as we want 25% from you to get classification. Either way .. M$ is rubbing its hands with glee.!! 2003-12-31 2:58 pm At least it is written in C and not Forth. This opens EFI to a broader range of developers and much of the existing software (e.g.: defragmenters) can be easily ported. This reduces the costs for many companies and they do not need to find Forth developers for their tasks. For those who fear EFI: You might find http://www.openbios.org interesting. It is an attempt at writing a GPL OpenFirmware implementation. This will probably be an alternative to EFI (if they get it finished fast enough). 2003-12-31 3:51 pm So there an open firmware specification. Why not use that ? I think DRM is the main motivation. Microsoft also want to check what software runs on computers, in an attempt to block pirated copies, but that could inavertently have as a side effect to stop any non-registered software to run, such as viruses and other worms but also open source software, especially if there is a fee to pay to be registered. 2003-12-31 11:19 pm Great, now all Microsoft operating systems will not work. Microsoft needs a standard BIOS to install. Only Windows CE and Windows XP Embedded can run without a BIOS as we know it. *nixes can run on anything provided it has electricity and transistors. 2004-01-01 4:04 am My understanding of the old BIOS is that it “standardizes” the motherboards hardware and peripherals (mouse, video etc) by providing an API to access them. This is great if you want to limit yourself to VGA (does BIOS support SVGA? Even if it did it would not be accelerated). Any advanced functions for hardware peripherals (like accelerated SVGA) need drivers that talk to the hardware directly, or at least a virtualized version of the hardware though the OS. To me, this seems to give the maximum amount of flexibility. BIOS starts the copmputer, OS takes over and manages everything the BIOS can’t handle (which is everything). This new EFI thing (sounds like the Intel guy that thought this up is into cars!) allows for the drivers (more cars!) to effectively move into the bios arena. This begins to look like a mini-OS. How does this reduce complexity? How does this allow for rapid adoption of new technologies? Coding in C/C++/Ada etc won’t help because the designer will need knowledge of operating system principles just to add a module that inter-operates with the rest of the system. In other words, imagine the levels of patching ten years from now if EFI is the norm! Why write or re-engineer a module if a quick patch will fix it? In conclusion, BIOS is a bios, EFI is a bios with operating system features built in. It seems that it won’t solve any complexity problems, but may solve (initially) some problems associated with the supposed lack of BIOS designers. It will also allow for the underlying system (interrupt allocations, hardware issues like pnp) to be re-engineered. Seems to me that can also be done with the current BIOS concepts. For what its worth, my opinion is that less is better. Less BIOS functions = less complexity. After all, the function of BIOS is to put the hardware into a known state and hand over control to an operating system. Why mess with that??? Thanks, Yohn ps: I deliberately avoided the whole DRM issue as it can also be implemented in too many ways that don’t rely on bios. 2004-01-01 6:41 am At least it is written in C and not Forth. This opens EFI to a broader range of developers and much of the existing software (e.g.: defragmenters) can be easily ported. This reduces the costs for many companies and they do not need to find Forth developers for their tasks. Yes, but it is a closed specification. Yes, Intel isn’t the root of all evil BUT there isn’t a safe guard in place which stops the exploitation of its market position. EFI on paper sounds nice, however, it is another poorly re-invention of the wheel. The fact remains that Intel made up a list of so-called “OpenBoot limitations” simply to make their “case” for their version. The fact remains that until they submit it to a specifications body and all development is done in the open for industry players FROM BOTH SIDES to develop it. For those who fear EFI: You might find http://www.openbios.org interesting. It is an attempt at writing a GPL OpenFirmware implementation. This will probably be an alternative to EFI (if they get it finished fast enough). Its good, but isn’t enough. What is required is for it to be changed to BSDL and for some large motherboard manufacturers to get behind its development so that they’re no longer held by the short and curlies by the likes of Pheonix.