Anything special need to be done before/after flashing your BIOS?

[Insane3D]

VirtualLarry-

First off, let me apologize for my last couple of posts. I was having a bad night at work, forgot my coffee, and an old knee injury was acting up, basically putting me in an unusually foul mood. You were trying to have a civilized discussion about differing viewpoints on bios flashing, and I overreacted. For that I apologize.

I don't want this to become a long drawn out quote by quote debate, so let me make clear some of the points I was trying to make and also respond to your points. I think after reading over this thread again, we are closer to agreeing on this topic than we may both realize.

Ok, first, you contend that a Windows flash is more dangerous than a DOS based flash. Without getting ultra-technical in the reasoning, I believe that essentially you saying it's more "risky" because their are more things that could affect the flash in windows. Giving this a lot of thought, I would have to say, generally, I agree with you. I think the difference in what we were arguing was largely semantics. When I say that windows flashing is not really dangerous if the OS enviroment is stable to begin with, I think you should know what I consider stable. Before I deem any OS install stable, I run a battery of stability tests, including, but not limited to, Prime 95 torture test, both tests, for a period of 48 full hours, to establish baseline stability on the CPU and memory level. I follow this with several looping 3D applications, while running a Prime 95 torture test in the background, and looping them for at least 24 hours. These applications include, but are not limited to, 3D Mark 2001 and 2003, Quake 3, UT2003. I usually run these test @ a high resolution and graphical settings to test the video subsystem integrity, then I follow them up with a low resolution, software rendering mode to test how the CPU and memory subsystem handles Prime and a 3D application in software mode. My last test is to run Memtestx86 for a 24 hour period to test for any memory errors that might not occur in the OS enviroment. This is what I consider a "stable Windows enviroment".

Now, that being said, I consider a windows based flasher, in these circumstances to be essentially no more or less dangerous than a DOS based flash. However, I will yield the point that even at this level of stability, problems can still occur with the windows based flash program itself, that none of the other stability tests would have picked up on. IMO though, speaking generally, the level of risk over a DOS based flash in these circumstances are negligable.

I will also agree with you that early windows based flashers were inherently buggy and unreliable, and I have a non-functioning Asus P4B266-C here that was killed in that very manner by a early, beta, and extremely buggy Asus windows bios flasher. It simply needs to have the bios chip flashed, which I can send to Asus to do for $5, but the board is essentially useless to me now, so I haven't bothered.

However, I still contend that the windows based bios flashers are much better these days, and will only improve, and generally they all work very well, and are easier for the casual user to work with. There are exceptions to this rule though. I would never used a windows flasher in a Win9X enviroment, including WinME, as these were inherently buggy OS's to begin with. I also would not use a beta windows flasher, as this tells me the mainboard manufacturers don't have enough faith in it's stability to deem it non beta.

Now, about the bios savior. For people who don't know what this is, let me explain. It's basically a bios socket with a spare bios chip in it. You take out your old bios chip, with the enclosed bios puller tool, put the savior into the boards bios socket, then put the boards bios chip into the savior's socket, and you can use a switch to make a backup of sorts of your bios chip. In the event of a catastrophic failure, you can just flip the switch, boot from the backup bios chip, and flash the bios correctly back onto the dead bios chip. I suggest these to any customers of mine that are likely to flash their bioses on their own. This greatly cuts down on customers calling back with a failed flash.

I really hope at some point, the board makers follow Gigabyte's lead with their dual bios setup, as it would essentially end the RMA's for bad bios flashes, and I can't imagine it adds much cost to the board.

Now before this reply gets too long winded (too late 😉 ), let me briefly address some of your other points. First, you contend that it's best to use a UPS when you are doing a bios flash. I both agree and disagree with that point. In most cases, I think a UPS is overkill just to flash a bios. If you already have one, by all means use it. UPS's are generally a good thing for any PC, regardless of bios flashing, but again, I don't think the risk you are taking by depending on the power to your house for that 30 seconds or so your are flashing the bios is such that you should not do it without a UPS.

Although, again, there are exceptions. If you don't run a UPS, and you often have power surges, or even rolling blackouts like in California, then by all means a UPS is a must. If you are in a area like me where I can't remember there ever being a single power surge or outage that wasn't storm related, in at least the past 5 years, than I think it's generally "safe" to flash the bios without a UPS involved.

Secondly, you mentioned setting bios optimized defaults. I disagree with that as well, but again, it's largely semantics. The procedure I generally use when flashing is this.

1. Bring all bios settings back to default, and undo any overclocking that has been done.
2. Flash bios (duh) 😛
3. Clear CMOS
4. Load optimized defaults
5. Reboot and set the bios back to the desired settings.

Now, you mentioned that the "optimized defaults" sets the strictest memory and chipset timings. I disagree. All the recent boards I've worked with, Intel and AMD nonwithstanding, when you load "Optimized Defaults", it leaves things like memory settings on "safe" settings. It usually sets all the memory timings to "By SPD", which is safe unless the board is reading the SPD data wrong. I generally load optimized defaults before, and then again after the flash. I suppose it wouldn't hurt if you loaded "Fail Safe Defaults" prior to the flash, but again, the added risk of setting "Optimized Defaults" is negligable. I'm sure there are exceptions to this as well, but I'm speaking genrally.


Well, after all that, I think we agree more than me disagree. To be honest, what sort of bothered me about your posts was that you went into a great amount of detail for a basically simple question. That may just be how you do things, and that's fine. It's really not my place to tell you how to post as this is a public forum, and we are both going for the same goal of helping out original poster. We just go about it in different ways, and again, that's fine. 🙂

My biggest issue with the whole "gist" of your posts was that I try to get users to not be so afraid of bios flashing. It's something that the new user/builder always fears because they hear all these horror stories. I try to help people to understand it's really not that dangerous if you take your time, and follow the correct steps. You obviously feel the need for a higher level of "safety" than me, and there really is nothing wrong with that. I think some of the precatutions you mention are a tad overboard, but I guess in reality, it won't do any harm, and might even save someone down the line. Generally though, I wish more people weren't so afraid to flash their bios, and I was concerned that a post like yours might scare someone away altogether who was considering flashing their boards.

I hope I made myself more clear, and I apologize again for my demeanor and tone. No hard feelings and I'll see you around the forums.

🙂

- Insane

:beer:

Edit: Formatting and syntax. And to add...

IMO

 

[VirtualLarry]

Originally posted by: Insane3D
VirtualLarry-

First off, let me apologize for my last couple of posts. I was having a bad night at work, forgot my coffee, and an old knee injury was acting up, basically putting me in an unusually foul mood. You were trying to have a civilized discussion about differing viewpoints on bios flashing, and I overreacted. For that I apologize.

No problem, I wasn't offended. I apologize too, for tending to become utterly pedantic when discussing technical things, and overly verbose. It's obvious that you know BIOS flashing pretty well too. I didn't mean to imply otherwise.

Originally posted by: Insane3D
I don't want this to become a long drawn out quote by quote debate, so let me make clear some of the points I was trying to make and also respond to your points. I think after reading over this thread again, we are closer to agreeing on this topic than we may both realize.

Ok, first, you contend that a Windows flash is more dangerous than a DOS based flash. Without getting ultra-technical in the reasoning, I believe that essentially you saying it's more "risky" because their are more things that could affect the flash in windows. Giving this a lot of thought, I would have to say, generally, I agree with you. I think the difference in what we were arguing was largely semantics. When I say that windows flashing is not really dangerous if the OS enviroment is stable to begin with, I think you should know what I consider stable. Before I deem any OS install stable, I run a battery of stability tests, including, but not limited to, Prime 95 torture test, both tests, for a period of 48 full hours, to establish baseline stability on the CPU and memory level. I follow this with several looping 3D applications, while running a Prime 95 torture test in the background, and looping them for at least 24 hours. These applications include, but are not limited to, 3D Mark 2001 and 2003, Quake 3, UT2003. I usually run these test @ a high resolution and graphical settings to test the video subsystem integrity, then I follow them up with a low resolution, software rendering mode to test how the CPU and memory subsystem handles Prime and a 3D application in software mode. My last test is to run Memtestx86 for a 24 hour period to test for any memory errors that might not occur in the OS enviroment. This is what I consider a "stable Windows enviroment".

If you take care of your Windows' OS installation, then yes, there probably is not an inordinant risk of flashing in Windows, but in the context of a newbie flashing something though - oftentimes, they think of flashing the BIOS, to fix some random system problem that they are having. Oftentimes, the problem is not the BIOS at all, but something else making the system unstable, and in that context, flashing the BIOS, especially inside of Windows, could have really bad consequences. That's one reason that I discourage Windows-based flashing.

The other reason, is a bit technical, and is becoming less of a valid issue than it once was, because of the newer technology being used in the flash chips themselves. If I may be pedantic for a moment:

Half of it is execution latencies of a user-mode program in Windows. Those latencies exist, simply because Windows' is a multi-tasking OS instead of a single-tasking OS like DOS. If anyone wants to see these latencies by themselves, try installing and running HDTach 2.61 (read test) on their HD. Try it with all your "normal" background programs running. Then try it with no additional background programs running. Try running a test while online, and actively using your web browser. Notice the "spikes" that occur in the graph, when background system activity, or user foreground activity happens? Those are drops in throughput, which are equivalent to an increase in latency.

Now consider how flash memory chips are actually programmed. (I tried to explain some of this in my prior posts.)

Factor 1: Older flash chips, oftentimes were not programmable with byte granularity, they had to have either a large chunk (sector-erase), or all (full-erase) of their data cleared out, before they could be re-programmed.

The problem with that was, if Windows' crashed, the flash chip would be stuck with a huge empty chunk somewhere, or worse, totally erased.

Factor 2: Older flash chips, had very specific timing requirements for proper programming. The host system code programming the chips, has to delay for so many milliseconds before sending new data, and then pulse certain pins for a certain time to program it, etc. Suffice it to say, that the process is very critically timing-dependent for these older chips.

When you factor in the issue of execution latencies in Windows' into the equation, along with the fact that the timing algorithm might read back the right data right away, but if the timing constraints for programming are not met, the data might "fade" at some unspecified point in the future, and the BIOS might corrupt itself unknowingly. It's kind of like a poorly-burnt CD on cheap-quality media. If you don't get a "good burn", or the dye layer is poor, the data can tend to fade away after only a year, even though data verified as correct, immediately after the burn process. (I've seen it happen personally with some cheap CMC-made CDs. Surprised the heck out of me.)

That is one of the precise risks of flashing in Windows', that I was trying to point out, that you might have been unaware of.

That warning said, most newer flash-memory chips on the market, have "self-timing" programming algorithms now, that are not affected by timing inconsistencies that can occur by flashing from inside of Windows. They do this by having an additional data register for programming, and their own little timing circuit. In fact, if you look really carefully, when some of the first Windows-based flashers came out, they had warnings in their documentation about only working on boards that had flash chips compatible with Windows'-flashing mounted on the boards. What they were really referring to, were these newer self-timing flash-programmable chips, rather than the older externally-timed programming ones.

So hopefully this is now a non-issue with most systems. Which is I guess what you were trying to say, that flashing in Windows, as long as your OS is stable and doesn't crash, isn't really any more dangerous than DOS. I was being overly-paranoid, in terms of safety, because it wasn't clear what kind or era of system that was being mentioned.

Originally posted by: Insane3D
Now, that being said, I consider a windows based flasher, in these circumstances to be essentially no more or less dangerous than a DOS based flash. However, I will yield the point that even at this level of stability, problems can still occur with the windows based flash program itself, that none of the other stability tests would have picked up on. IMO though, speaking generally, the level of risk over a DOS based flash in these circumstances are negligable.

I will also agree with you that early windows based flashers were inherently buggy and unreliable, and I have a non-functioning Asus P4B266-C here that was killed in that very manner by a early, beta, and extremely buggy Asus windows bios flasher. It simply needs to have the bios chip flashed, which I can send to Asus to do for $5, but the board is essentially useless to me now, so I haven't bothered.

I think that's why I am perhaps slightly prejudiced against ever suggesting to use a Windows-based flasher, because I know what they could do in the past (or rather, fail to do properly). I remember seeing countless posts about "dead" mobos in the MSI tech-support forums, after they released their Internet/Windows-based "LiveBIOS" update program. Knowning what I know about the complexities of modern systems and Windows, it does kind of terrify me to know of all of the things that could potentially go wrong during that operation.

Originally posted by: Insane3D
However, I still contend that the windows based bios flashers are much better these days, and will only improve, and generally they all work very well, and are easier for the casual user to work with. There are exceptions to this rule though. I would never used a windows flasher in a Win9X enviroment, including WinME, as these were inherently buggy OS's to begin with. I also would not use a beta windows flasher, as this tells me the mainboard manufacturers don't have enough faith in it's stability to deem it non beta.

Ok, I would agree with that. I guess most of them aren't quite as "deadly" to boards now, as they once were in the past, but I still personally feel much safer flashing in DOS.

Originally posted by: Insane3D
Now, about the bios savior. For people who don't know what this is, let me explain. It's basically a bios socket with a spare bios chip in it. You take out your old bios chip, with the enclosed bios puller tool, put the savior into the boards bios socket, then put the boards bios chip into the savior's socket, and you can use a switch to make a backup of sorts of your bios chip. In the event of a catastrophic failure, you can just flip the switch, boot from the backup bios chip, and flash the bios correctly back onto the dead bios chip. I suggest these to any customers of mine that are likely to flash their bioses on their own. This greatly cuts down on customers calling back with a failed flash.

I really hope at some point, the board makers follow Gigabyte's lead with their dual bios setup, as it would essentially end the RMA's for bad bios flashes, and I can't imagine it adds much cost to the board.

Good explaination, and I heartily agree about Gigabyte's dual-BIOS feature. Unfortunately, it still seems to increase cost more than it should, even though the cost of the flash chips themselves have gone done. It actually should be feasable to do basically the same thing, with a single chip of double the size. The "boot block BIOS", in that case, would just be an entire identical backup copy. All the user would have to do is change a jumper. This assumes that the most likely failure is a bad flash, and not the chip actually being physically-destroyed somehow. I think that's a safe assumption to make though, because otherwise the motherboard would probably also be damaged at that point.

Originally posted by: Insane3D
Now before this reply gets too long winded (too late 😉 ), let me briefly address some of your other points. First, you contend that it's best to use a UPS when you are doing a bios flash. I both agree and disagree with that point. In most cases, I think a UPS is overkill just to flash a bios. If you already have one, by all means use it. UPS's are generally a good thing for any PC, regardless of bios flashing, but again, I don't think the risk you are taking by depending on the power to your house for that 30 seconds or so your are flashing the bios is such that you should not do it without a UPS.

Like I admitted to, I've flashed other people's motherboard BIOSes on-location, without having a UPS available. It makes me nervous, but what can you do if they want the job done? Perhaps I overstated the importance of a UPS a bit too strongly, but my personal preference is to run with a UPS always, in any case. (And in fact, MS suggests that you do not enable write-caching to the HDs in Windows, unless you are running on a UPS. That is the "power protected" write-cache policy setting.)

The lack of a UPS won't automatically make the flash go bad, of course. Just eliminating a (rare, but possible) risk of the power going out during a flash operation. It's a good idea, but perhaps not strictly necessary.

Originally posted by: Insane3D
Although, again, there are exceptions. If you don't run a UPS, and you often have power surges, or even rolling blackouts like in California, then by all means a UPS is a must. If you are in a area like me where I can't remember there ever being a single power surge or outage that wasn't storm related, in at least the past 5 years, than I think it's generally "safe" to flash the bios without a UPS involved.

I'm in the northeastern US, we've been having thunder and lightning and occasional rain nearly every night this week. My UPS briefly alerts, often.

Originally posted by: Insane3D
Secondly, you mentioned setting bios optimized defaults. I disagree with that as well, but again, it's largely semantics. The procedure I generally use when flashing is this.

1. Bring all bios settings back to default, and undo any overclocking that has been done.
2. Flash bios (duh) 😛
3. Clear CMOS
4. Load optimized defaults
5. Reboot and set the bios back to the desired settings.

Ahh. Ok. I agree with that perfectly, with the caveat that I always manually check all of the settings, after first setting "Optimized defaults".

What I took issue with was MichaelD comment about setting "Optimized defaults" *before* flashing, as part of the de-overclocking process, not after. You always should reset CMOS settings to one of the default settings *after* a flash to clear them out and set them up again. (But should also use clear CMOS jumper first too, some bytes of the CMOS config don't get touched/reset by setting a default setting.)

The only reason I disagreed with that is (again, the "better safe than sorry" principle, combined with a lack of specific knowledge about the board in question), was that some motherboard BIOSes do set critical memory-timing parameters to their most-optimized settings, which in many cases (memory latencies at high FSB speeds), the hardware itself is not capable of (but the chipset may be). That could potentially de-stabilize a system's memory subsystem right before the flash operation, which is a Bad Thing.

In fact, assuming that the memory is stable already, don't change the settings at all, just un-overclock the FSB and CPU settings if necessary, and perhaps decrease voltages to "normal" levels, assuming that they had to be increased in order to overclock. (Actually lowering the voltages might not be necessary at all.)

Originally posted by: Insane3D
Now, you mentioned that the "optimized defaults" sets the strictest memory and chipset timings. I disagree. All the recent boards I've worked with, Intel and AMD nonwithstanding, when you load "Optimized Defaults", it leaves things like memory settings on "safe" settings. It usually sets all the memory timings to "By SPD", which is safe unless the board is reading the SPD data wrong. I generally load optimized defaults before, and then again after the flash. I suppose it wouldn't hurt if you loaded "Fail Safe Defaults" prior to the flash, but again, the added risk of setting "Optimized Defaults" is negligable. I'm sure there are exceptions to this as well, but I'm speaking genrally.

Ok, that's good news then, but I know that there are some problem boards/BIOSes "out there in the field" somewhere. They do exist. The majority may not have that problem though, so I guess my concern is only applicable to a minority of boards.

Originally posted by: Insane3D
Well, after all that, I think we agree more than me disagree. To be honest, what sort of bothered me about your posts was that you went into a great amount of detail for a basically simple question.

I agree about the first, and I apologize about the second. I guess that's just the way I am. I tend to launch into heavily-detailed technical discussions, without restraint, or apparent need to even do so, save that someone might read the posts in the future and learn something. (For example, Google's Usenet archive is a wealth of tech-support reference information. It actually saddens me, how much discussion occurs on web boards like this one, and isn't archived nor searchable. A lot of good knowledge goes to waste. Usenet actually had things pretty much "right", IMHO.)

Your earlier post actually answered the OP's question quite satisfactorily, so I probably didn't need to post at all, but I disagreed, as I mentioned, with one of MichaelD comments on technical reasons, and then posted some addtional (unnecessary, probably) material, basically as a sort of additonal FYI on BIOS flashing. You then took issue with one of my assertions, I disagreed with you, and the discussion chain-reaction kind of happened from there, I guess.

Originally posted by: Insane3D
My biggest issue with the whole "gist" of your posts was that I try to get users to not be so afraid of bios flashing. It's something that the new user/builder always fears because they hear all these horror stories. I try to help people to understand it's really not that dangerous if you take your time, and follow the correct steps. You obviously feel the need for a higher level of "safety" than me, and there really is nothing wrong with that. I think some of the precatutions you mention are a tad overboard, but I guess in reality, it won't do any harm, and might even save someone down the line. Generally though, I wish more people weren't so afraid to flash their bios, and I was concerned that a post like yours might scare someone away altogether who was considering flashing their boards.

You know, that's actually a very good point. I vaguely thought about that some time afterwards. I guess that was one of the points that you were trying to make, but not explicitly stating, with the comments towards me about being overly-safe, which I think I took at the time a bit more personally than I should have. What you really meant was, no one should fear flashing their BIOS. It won't cause a nuclear meltdown, it won't set the world on fire. You just need to take some (minor) precautions, in order to reduce the slight risk that your system could end up non-functional after the process.

I guess my personal estimations of the risks might have been a bit overstated, subsequently my risk-avoidance procedures a bit overblown as well. I still will follow them personally, but they may not be for everyone.

Originally posted by: Insane3D
I hope I made myself more clear, and I apologize again for my demeanor and tone. No hard feelings and I'll see you around the forums.

🙂

- Insane

:beer:

Edit: Formatting and syntax. And to add...

IMO

Same here, no hard feelings, I always enjoy "talking tech" with someone in the know. Cheers! :beer:

PS. Sorry, I just can't seem to write a short post, when talking tech either.