| Date: | April 12, 2005 / year-entry #91 |
| Tags: | other |
| Orig Link: | https://blogs.msdn.microsoft.com/oldnewthing/20050412-47/?p=35923 |
| Comments: | 76 |
| Summary: | A bunch of us were going through some Windows crashes that people sent in by clicking the "Send Error Report" button in the crash dialog. And there were huge numbers of them that made no sense whatsoever. For example, there would be code sequences like this: mov ecx, dword ptr [someValue] mov eax, dword ptr... |
A bunch of us were going through some Windows crashes that people sent in by clicking the "Send Error Report" button in the crash dialog. And there were huge numbers of them that made no sense whatsoever. For example, there would be code sequences like this: mov ecx, dword ptr [someValue] mov eax, dword ptr [otherValue] cmp ecx, eax jnz generateErrorReport
Yet when we looked at the error report, the We had to mark these crashes as "possibly hardware failure". Since the crash reports are sent anonymously, we have no way of contacting the submitter to ask them follow-up questions. (The ones that the group I was in was investigating were failures that were hit only once or twice, but were of the type that were deemed worthy of close investigation because the types of errors they uncovered—if valid—were serious.) One of my colleagues had a large collection of failures where the program crashed at the instruction xor eax, eax How can you crash on an instruction that simply sets a register to zero? And yet there were hundreds of people crashing in precisely this way. He went through all the published errata to see whether any of them would affect an "xor eax, eax" instruction. Nothing. He sent email to some Intel people he knew to see if they could think of anything. [Aside from overclocking, of course. - Added because people apparently take my stories hyperliterally and require me to spell out the tiniest detail, even the stuff that is so obvious that it should go without saying. I didn't want to give away the story's punch line too soon!] They said that the only [other] thing they could think of was that perhaps somebody had mis-paired RAM on their motherboard, but their description of what sorts of things go wrong when you mis-pair didn't match this scenario. Since the failure rate for this particular error was comparatively high (certainly higher than the one or two I was getting for the failures I was looking at), he requested that the next ten people to encounter this error be given the opportunity to leave their email address and telephone number so that he could call them and ask follow-up questions. Some time later, he got word that ten people took him up on this offer, and he sent each of them e-mail asking them various questions about their hardware configurations, including whether they were overclocking. [- Continuing from above aside: See? Obviously overclocking was considered as a possibility.] Five people responded saying, "Oh, yes, I'm overclocking. Is that a problem?" The other half said, "What's overclocking?" He called them and walked them through some configuration information and was able to conclude that they were indeed all overclocked. But these people were not overclocking on purpose. The computer was already overclocked when they bought it. These "stealth overclocked" computers came from small, independent "Bob's Computer Store"-type shops, not from one of the major computer manufacturers or retailers. For both groups, he suggested that they stop overclocking or at least not overclock as aggressively. And in all cases, the people reported that their computer that used to crash regularly now runs smoothly. Moral of the story: There's a lot of overclocking out there, and it makes Windows look bad. I wonder if it'd be possible to detect overclocking from software and put up a warning in the crash dialog, "It appears that your computer is overclocked. This may cause random crashes. Try running the CPU at its rated speed to improve stability." But it takes only one false positive to get people saying, "Oh, there goes Microsoft blaming other people for its buggy software again." |
Comments (76)
Comments are closed. |
Interesting… But how could overclocking cause those crashes?
Uhm… by running a CPU at a higher clock speed than it’s rated for?
It’s like playing a musical instrument with a faster tempo that you can play: If you’re really good you may only skip a beat (crash the computer) a couple of times a day.
Overclocking is trying to get more instructions through your pipleline/processor than it was built/tested for.
Think of it in terms of Speed Rating on tires. These Dunlop tires I’m using are speed rated for 130mph. I can go 140-145mph with them without much of a problem. But get into the 150-180mph range, and it’s anyones guess as to what’s going to happen. The tires are moving faster than the structure that was built inside of them allowed for, which is going to cause some anomolies and "Mis-shapes" of the tire. Eventually you’ll have a blowout. Assuming it doesn’t blowout and allows you to go that fast, chances are it’s not going to be a smooth ride.
CPUs overclocking is very similar, you’re trying to get the CPU to work faster than it should, or was built to, and while it will do it with the proper cooling, it does not mean that the integrity of the data being processed is guarenteed.
Why not have the crash dump include the clock speed info? CallNTPowerInformation and PROCESSOR_POWER_INFORMATION can tell you a lot of good stuff on most recent CPUs, including the current and max speed. Also, SYSTEM_POWER_INFORMATION lets you know if the system is throttling because of overheating such as a failed fan or too much fuzz in the heat sink. If it’s considered too risky to make that call during crash data collection, you can usually get the max speed by calling CPUID to get the CPU description string; the speed is usually in that string.
I don’t understand why you find it unusual. For someone that does computer repairs it is pretty obvious that if a CPU is allowed to overheat it starts to act erratically and programs fail "randomly".
I’m very surprised the Intel guys didn’t flag this instantly, particularly with that xor instruction.
Why does overclocking cause processors to fail? It’s not so much about the heat, although that is an issue. The problem is that various components exhibit charging curve characteristics when building up voltage as switching occurs. Each transistor takes a little time to build up enough charge and switch. If the clock is sped up too much, the circuits may not have had enough time to stabilise on the correct result for their inputs. The result is an adder that can’t add.
So why can you get away with it a lot of the time? Because the manufacturers are largely trying to cater for reliability, and so build in quite a bit of safety margin. Also, they’re improving the process. I suspect that the number of parts that meet the higher criteria is actually greater than the demand for those parts – as a result some of the parts are actually marked and sold below their actual capability.
Thanks Mike, that was the answers I hoping I’d find.
Something that is interesting to me here is just how the ‘error report’ process works at Msft. Like most(?) other software vendors there isn’t a nifty link in each application to report bugs. And I’m sure if there was, Raymond would get a lot of spamy reports. Does that mean that bugs are only fixed if found in QA (and how is a bug defined at Msft?), or are they fixed only when enough paid-support calls come in?
I often do find bugs in msft code, but they simply are not important enough for me to bother opening a support incident for. More on the level of annoyances.
For example find an audio CD set with two disks in it with the same title in the title database – media player gets very confused if you rip both CDs. Kinda ruins my ‘experience’ playing ripped my http://www.T.I.S.M.wanker.com tracks, but am I going to go to the pain of calling somebody about this? No. And I’m also sure that since there is no easy way to report this problem, it will never be addressed.
(I need to add this to the Suggestion Box page.)
You can report bugs or ask for new features via http://register.microsoft.com/mswish/suggestion.asp And yes every single one gets read. Including the death threats.
I remember reading a while back that some Intel engineer had come up with some interesting way to detect in hardware whether or not a CPU was overclocked. It consisted of implementing a constant rate clock (that can’t be changed) on the motherboard and timing against that with a value in the CPU that varies with the clock of the CPU vs. the "actual" clock.
I didn’t quite spend that much time reading it, so I don’t remember the details. If anyone knows the article I’m speaking of, a link would be greatly appericiated :).
I work in Product Support, and my team deals with debugging a lot of these sorts of crashes. Most of my customers are corporate users, so I thankfully don’t see too much in the way of overclocking, but in the last few years we’ve seen a noticeable increase in these kind of weird crashes as a result of CPU errata.
I’m not an expert on CPU design, but I’d guess that as CPUs become more complex, maybe it’s inevitable that bugs in their microcode become more common too. KB article 842465 describes one common crash we eventually tracked down to a CPU microcode bug: http://support.microsoft.com/?kbid=842465
It’s important to keep up to date with the latest BIOS revisions, since these will generally contain the latest CPU microcode updates which contain fixes for the published errata.
Maybe I should also mention the time that my own home machine, which had been working flawlessly for months (and is not overclocked), suddenly started bluescreening randomly, every few hours or so.
I looked at one of the memory dumps and found a single-bit error in one of the CPU registers. This lead me to open up the case and take a look. I pulled off the processor fan and noticed that the heat sink fins had gotten completely clogged up with dust. I used a few blasts of compressed air to clear them out, replaced the fan, and the system has worked fine ever since.
It’s generally a tough sell trying to convince people that their system instability is caused by a hardware problem, but it happens a lot more than many people realize, and it’s not always easy to prove.
You may find this interesting. I showed a local expert the "how can it crash?" question:
"xor eax, eax
How can you crash on an instruction that simply sets a register to zero?"
His (brusque) reply was:
> Would you like a real answer for his question?
> Page table invalidation failure, causing either an instruction prefetch fault or an incorrect instruction to be fetched, which the debugger wouldn’t notice because it creates a separate mapping to inspect it.
> "Been there, fixed that."
Windows programs crashing often? It might be because your machine is overclocked.
Riffing on Raymond, once again :)
Raymond’s post today reminded me of an email message sent…
I have regular random crashes.
I don’t have an overclocked machine, but it is a hardware issue.
I have a video card that needs a lot more power than it is "supposed" to.
Guys, just to be clear about something.
xor eax, eax
This instruction takes place entirely within the cpu, using only registers, not moving any data in and out of the cpu, to or from memory or otherwise. It happens in one cycle, cannot be preempted or in any way allow another instruction to sneak in during the middle of it.
In other words, if this instruction fails it is because the CPU failed. The cpu managed to flip a bit in the register and the resulting value in the eax register was something other than zero.
If Anyone out there is telling you otherwise they are a knucklehead or just plain making something up.
IIRC Intel a while ago had a program that checked for an overclock. I remember this because my computer was acting funky so I went to the Intel site and found it. It turns out that my computer wasn’t overclocked and I had a RAM error. I know that the current Intel® Processor Identification Utility (http://snipurl.com/dz11) does give current Clock speed as well as Maximum Tested Clock speed. Useful stuff, it’s how I found out my Laptop doesn’t have SpeedStep® Which would explain my lack of battery life.
On the issue of people not believing that Hardware can cause system instability, I think it’s bad even amoung "professionals" I have a warrenty on my laptop I got from Best Buy because I get a new battery every so often and it was well worth the cost for hardware replacement. I brought my laptop in 2 times telling them the harddrive was going bad (I got Unmountable_Boot_Volume blue screens once) and they kept saying it was a software problem (of course not covered under warrenty.) Finally the Drive crashed for good (Yesterday in fact) and they had to fork over the new drive, but not before they ran every test they had because surely it couldn’t be a bad drive casuing this.
To Joe Dietz: Problem solved! That’s not a Microsoft bug. That’s just a problem with IDing different CDs with the same ID. So it’s a design flaw in the CD database system. If it were my design, I’d assign every piece of every album its own GUID, so your two-disc set would be one album with two pieces, therefore two GUIDs, which means the database could keep things separate; it’d just have to have some sort of marker to indicate "I’m part of a set, and I’m disc number n of m." Or maybe even the same ID plus a -000x tot he end to identify which part of the track listing to grab.
Whine at the music industry for their problems, and MS for theirs.
Of course, the part that kills me that no one else seems to be noticing is that much of the overclocking is done by "Bob’s computer store" type shops.
People, these shops are fooling customers and being unethical. They obviously are buying cheaper computers, overclocking them, then selling them as high end machines with a high end price tag.
I find this to be outrageous – and totally bogus that MS has to once again deal with someone else’s crap.
James Summerlin
How to *detect* overclocking — that’s a good question for Intel engineers. They spent a lot of time to put in the processor the circuits to *prevent* the overclocking, but that was not so successful. Detecting must be simpler.
As far as I know, all the modern processors have the possibility to change their microcode, so it must be possible to store in the processor the desired clock frequency as some constant even after the clock decision is made. I guess AMD must have some similar possibilities too.
There’s various cpuid type of progs around, and the system properties seems to also say the running frequency and the original frequency.. Atleast on my P4.
Anyway there’s a lot of information available to give hints about overclocking etc if you can install and run the soft needed to dig it.
Also there are shops who sell different memory brands under same name and with bad luck you can get ones with different timings and combine that with crap mobo etc.. anyway people should run memtest86 for 2 hours without errors atleast before running windows. Single memtest pass doesn’t always reveal problems.
I bet all those security holes in IE are caused by over clockers too!
Kidding of course… I had my computer crash quite a bit due to dust build up in the heat sink — caused the CPU to overheat. It took a while to figure that one out.
The question posed to the Intel folks was "Can you think of anything *else* that can cause this?" Obviously overclocking was everybody’s top guess, but you want to have a backup plan in case somebody responds, "No, I’m not overclocking."
I forget what the consequences of mis-paired RAM were, but they didn’t apply in this case.
Given some of the symptoms discussed, and other things talked about I thought some of you might find the following link interesting.
http://www.osronline.com/article.cfm?id=380
Unfortunately I think registration is required, but it is an interesting story that shows just how non-intuitive some bugs can be.
To Robert Sharp: The CD unique identification issue is one of my pet peeves. I’ve been "planning" a fantasy time travel trip back in time to correct IMHO some of our technological shortcomings. One key stop is the CD Red Book standards committee meeting. Please, please put some bits about bits on the CD. At a minimum a unique ID that can be used later against a database. Better yet encode minimal track information. I can see the conversation now from that era: "Why would anyone want to put a music CD in a computer?" My next stop would be a visit to a young man named Bill Gates…
O poveste interesantă despre overclocking:
http://blogs.msdn.com/oldnewthing/archive/2005/04/12/407562.asp…
I’m kind of surprised the album databases go by album or album title at all. I’d always assumed they treated each CD as unique, and generated some sort of key based on number of tracks, track length, etc.
jstatz: it is, of course, worse than that.
http://www.jwz.org/doc/cddb.html
It’s a good idea to download a CPU temperature monitor utility and run it now and again. (Most BIOS’s will have one, but you’d have to reboot when the computer’s hot to use it.) The CPU temp rises over time as the heatsink/fan get clogged up. When it gets too high, you know to open up the box and clean out the dust. That way you should avoid heat-related crashes.
Most overclocking is done by boosting FSB clock speed… So I suppose the question is how to detect FSB speed (most manufacturers gives tools to change that live while running the computer but I can’t think of a simple and uniform way to get that information)
That’s the reason I build my computers from parts. (Okay, not the real reason.) I have my CPU slightly overclocked; Primes95 runs stable for hours though. I didn’t have Windows crash for the last three months or so; and Windows Explorer for at least a month.
BTW, my old computer (bought off-the-shelf, supposedly not overclocked) crashed quite a lot. It ran Windows ME, though I won’t blame it until I know if it wasn’t the fault of Hauppauge’s (notoriously crappy) tv-tuner drivers.
Switching to XP solved the problem though. *Maybe* Win ME was unstable. But that switching the OS fixes the problem doesn’t prove it wasn’t the hardware’s fault, right?
I believe it’s a really good idea to inform the user that his computer is overclocked, at least when the crash happens. We saw that at least for some processors it’s possible to get the information of the frequency declared by producer. So at least when the possibility exists to detect overclocking, it’s a good thing to inform the user. And of course also to put it in the "send to microsoft" info.
I don’t think that the false positives would be so bad. It’s overclocking — anybody who does it shouldn’t expect anything to work.
To have the information that the system is running under overclocking circumstances, you don’t have to wait for a crash, you can detect it at some convenient time, or even inform the user after he reboots after the crash.
The false positive I’m more worried about is reporting a CPU as overclocked when it really isn’t. I’m told by the error reporting folks that overclocking cannot be detected 100% reliably. They also drew my attention to this motherboard http://www.msi.com.tw/html/service/techexpress/mainboard/6728/page3.htm which gives strange military terms to their overclocking.
What I find aggervating is that there are some computer equiptment manufacturers out there who make motherboards and other components overclock right out of the box. When I worked for a fairly large channel OEM, we had many problems with motherboards (ASUS in partictular) that were built out of spec to begin with, or were set to OC in the BIOS by default. In one incident, an order of 200 computers needed to be rebuilt after it was found that the floppy drive would not work while the onboard ethernet adapter was running. After much head-pounding, we traced the source to the 533FSB P4s we were using, which were reported to be compatable with the board. Turns out that the chipset on the board was only compatable with 400FSB CPUs, and that it had been tweaked to accomidate the 533FSBs. When we asked ASUS about the issue, they replied (We tweaked it, but we still have bugs to work out of it.) This was a board in mass production.
We ran into many other problems associated with this, and while I’m sure they won all of the benchmarks like their advertizers say they did, I do not expect half of them to run stable for a considerable length of time.
"This instruction takes place entirely within the cpu, using only registers, not moving any data in and out of the cpu, to or from memory or otherwise" — but it’s a two byte instruction (33 C0), so what happens if (EIP&4095)==4095 and the next page is mapped out with nothing to back it?
(no I’m not sure myself ;)
Oh dear. I think my answer is that without the second byte, it won’t be "xor eax,eax" in the first place :)
Four and a half years ago I bought a motherboard made by a company called PC Partner (now hopefully defunct). I had little choice as the AMD CPU/Gigabyte motherboard I had originally chosen ran too hot for Saudi Arabia, and the motherboard was the only replacement one they had in stock for the Pentium 733Mhz I had exchanged the Athlon for.
The CPU temperature went down to normal but I kept on getting strange IO errors. I took the thing back to the shop half a dozen times, but nobody seemed capable of diagnozing the fault.
It was only when I ran Sandra that I realized what had happened. Sandra reported the 66Mhz AGP video card running AT 106mHZ and the PCI bus running at 53Mhz.
What had happened is that PC Partner had presumably bought a load of mobo chips designed to run at the 83Mhz speed used by the Cyrix CPU, sales of which had collapsed a year previously. They had stuck them on this motherboard and then said that you could adjust the FSB to either 66Mhz, 100Mhz or 133Mhz. Of course at 100Mhz you were already significantly overclocked, and at 133Mhz everything became totally unreliable, including the IDE bus which ran overclocked like the PCI bus.
I swapped out the motherboard for a Transcend (the only one around that would run at 133Mhz, and had an ISA slot needed for legacy hardware). The machine has only crashed once since, and is still my main machine.
What did surprise me though was seeing a lot of these PC Partner motherboards on sale, and lots of salesmen who through ignorance were prepared to sell 100MHz maximum motherboards with 133Mhz CPUs.
Mind you it was only a year before that that PC Chips had caused no end of grief by selling overclocked mother boards. People spent hours trying to fix software faults when the problems were hardware related. In general, of course, the rule is that if you can’t regularly replicate the fault it is hardware related, but there are still catchmes.
Incidentally when Intel decide on the clock spped ror a chip they make load of chips in one go and then give the higher clock speed to the ones that pass quality control. However, as they get better at making the chips there are more chips capable of running at the higher clock speed than they can sell at that price range, so they simply stamp a certain proportion of these chips with the lower clock speed. This explains why you could often safely ‘overclock’ chips near the end of the run when you would blue-screen doing the same only three months earlier. You weren’t actually overclocking the chip, just restoring it to the speed Intel’s marketing department wouldn’t let it be stamped at.
Are we going to update the automated tool and give the user guidance when these ‘impossible’ situations occur (ie. stop on setting eax to zero), such as 1)informing the user that the stop code they experienced could not have been caused by software, and could be caused by overclocking (or other causes like a non-functional CPU fan) and 2)give them a link to a tool to identify whether or not their CPU is overclocked? (http://www.intel.com/support/processors/sb/cs-015477.htm)
Tim Alexander wrote: "Please, please put some bits about bits on the CD. At a minimum a unique ID that can be used later against a database."
There is the option of labelling each track with its ISRC number (see http://www.ifpi.org/site-content/online/isrc_guide.html).
"Better yet encode minimal track information."
That’s what the CD Text extension is for. Unfortunately it’s not commonly supported. My DVD player can show the text but I haven’t yet noticed a disc that has it.
I right-click the "My Computer" icon on the desktop and get properties. It reports two CPU speeds, one of which matches the OEM’s spec and one of which is around 1% slower. If someone had overclocked the machine wouldn’t the second number be too high instead? In other words, isn’t there already enough information for the end user to know the answer?
As for mismatched RAM, I thought the situation was rather obvious. If the BIOS checks the speed of the first RAM card, and then a cursory check of the second RAM card thinks that the second card can run at the same speed, then maybe the computer’s going to do that. And then some more intensive operation comes along and the second RAM card produces nonsense results because it really can’t handle that speed.
There’s a certain non-Microsoft OS where, for a while, a video driver was running a video chip at 70Hz. The driver developer said there was no difference between running the chip at 60Hz or 70Hz because the LCD screen still only responded at an effective rate of 50Hz. Well sure there would be no effective difference between 50Hz and 60Hz, but 70Hz made a big difference because the chip wouldn’t run right at 70Hz. Fortunately that got fixed.
Ok OC’ing causes problems, most people that do it purpose know that. However it’s total unfair to blame "bob" and his small pc store when the major mobo manufactures do it, esp as MS can influence those far easier.
It should also be apparent that if you had 5 people who knew they were oc’ing also had no clue that that was the reason for the crash…so the feedback that the error routines are giving is not good enough to enable even technical users to solve the problem.
"What did surprise me though was seeing a lot of these PC Partner motherboards on sale, and lots of salesmen who through ignorance were prepared to sell 100MHz maximum motherboards with 133Mhz CPUs."
Actually… I think I remember the reason why people did that. Back in those days, Intel had a killer chipset known as 440BX. It was a good performer, reliable and had all the features required of a good chipset at the time. Except… 100MHz only.
Intel’s next chipset was the 815. Anyone remember that one? It actually performed worse than the 440BX, usually catered to the onboard graphics folks and was less stable as I recall. Nobody wanted it. However, the 440BX usually ran quite stable at 133MHz with the new generation of Pentium processors.
Most components (video, CPU, memory and hard drives for that matter) run better if you cool them down sufficiently. This usually gives you ample room to increase speed. Why not take advantage of it, as long as the system remains stable?
Certain components OTOH can turn out to be already clocked to the hilt. E.g. http://tomshardware.com/ ran a test a while back of various multi-display graphics cards and benchmarked their performance with Flight Simulator and various other games. While conducting their tests, atleast one card burned up and several displayed various anomalies as they activated the second display. The cards simply couldn’t handle 3D on both displays.
So… Even if the user is not overclocking, he/she must make sure the components can handle the load required. It’s not really about overclocking, but rather finding a good testsuite and pick decent hardware.
—
Rune
Dear Rune,
You’re grossly overestimating the technical capacities of these salesemen and their bosses. Even the guys who assembled their computers couldn’t choose the right EIDE cable for an HD.
You’re right about the chipset, but the reason was much more prosaic. They had a load of these 440BBX motherboards around, but everybody was buying 133Mhz chips and memory. So, rather than losing the money they’s already paid for the motherboards, they told people they could overclock them with no problem, even though the manual said they couldn’t.
Steven, I wouldn’t trust a salesperson even if he told me that the sun is extremely hot.
I’m pretty sure I’m correct concerning the low quality of the i815 chipset. I remember discussing this with several friends at the time, and we were all holding off our upgrades until something better than the 440BX came along. People were actively still buying 440BX motherboards and their enthusiasm was fueled by hardware reviews on the web, not salespersons. (e.g. http://www20.tomshardware.com/motherboard/20000611/i815-04.html)
Eventually the Athlon appeared and many of us haven’t looked back since. (Next week my Athlon XP2500+ at home will be replaced by dual Opterons! :D) Specially these days when nVidia are able to match the quality of Intel chipsets.
—
Rune
Microsoft’s Raymond Chen (whose epitaph will no doubt include the words, "developer of the original Tweak UI utility for Windows") put up a fascinating post earlier this week. It’s worth reading for two reasons. First, it details how Microsoft engineers really do use the data you submit when your Windows computer crashes. Second,…
What about underclocking? I just bring this up due to the fact that I have a machine that is suffering these types of errors, and it is currently underclocked because PC100 ram on a processor that usually uses a 133 FSB = Underclock. The thing that makes me think that this is an issue is that this system was orginally using a duron (100 mhz fsb), and I upgraded the CPU. Before the CPU upgrade the system was stable as hell. Now when certian combinations of apps load it likes to crash. The machine was formated after it got the new CPU, so that isnt an issue. I know I should upgrade the ram, but being a poor student doesnt help matters. I just figured that this is an amusing flipside to this blog post.
Norman Diamond: You said "I right-click the "My Computer" icon on the desktop and get properties. It reports two CPU speeds, one of which matches the OEM’s spec and one of which is around 1% slower. If someone had overclocked the machine wouldn’t the second number be too high instead? In other words, isn’t there already enough information for the end user to know the answer? "
No, there’s not. You are not getting two speeds, you are getting a processor name STRING in the first line and a processor speed in the second line. One of my systems says this under Computer:
AMD Athlon(tm) XP 2500+
1.85 GHz, 512 MB of RAM
The first line is the processor name string that is encoded in the processor, and the second is the actual clock speed. (As you know, AMD claims its XP running at 1.85 GHz is "comparable" to an Intel running at 2500. Hence the name. Benchmarks tend to agree. But that’s another topic.)
You can’t easily compare these.
David Walker
Norman: All of what I just said presumes that you’re on a uniprocessor machine.
For what it’s worth,
the Nvidia nTune(linked as my URL) will let you know what settings, if any might have caused your crash… that is IF you used it for your overclock, and of course IF you your using an Nvidia chipset(Nforce2, 3, or 4)
…and I impressed. It looks like some guys don’t understand the point and some other want to be just smarter.
My favourite line is: "I’m not an expert on CPU design, but I’d guess that as CPUs become more complex, maybe it’s inevitable that bugs in their microcode become more common too." Good guess, indeed!
A guy who "tunes up" his car messing with plugs and wires is considered to be a sad jackass. A guy who do the same with his PC is likely to be called an expert. How come?
Since my computers never crash, am I a lucky one or rather a boring sucker? Should I overclock something or at least say that Windows sucks and Macs are just better?
"A guy who "tunes up" his car messing with plugs and wires is considered to be a sad jackass. A guy who do the same with his PC is likely to be called an expert. How come?"
Look, some components are simply better than others. Some people spend money on good quality RAM chips and will be able to run those at a higher speed than the ones that comes with OEM kit.
And there’s little doubt that adding extra cooling to a CPU will in most cases increase its tolerance for overclocking. The manufacturer’s rating is based upon standard cooling solutions. If you use liquid cooling or buy a bunch of noisy fans, then there’s a good chance you’ll be able to overclock quite a bit and never encounter any problems at all. (but if you stuck with standard cooling and the rated speed, there’s still a chance something could go awry down the road — there’s no guarantee that avoiding overclocking somehow makes everything bulletproof)
This isn’t rocket science, so I don’t think any of the overclockers are trying to label themselves as "experts" just because they bumped up the speed of their rig a notch. That’s your label, not theirs.
In many cases, you get identical parts when buying two components of varying speed. So… If you overclock the slower part to match the faster part — how exactly do you propose that would cause any problems?
—
Rune (who don’t remember having overclocked anything, but might’ve tried a few years back — I didn’t inhale though)
Rumored many times, pushed so hard over the weekend that NPR had mentioned it this morning, it is now…
The Old New Thing : There’s an awful lot of overclocking out there The other half said, "What’s overclocking?" He called them and walked them through some configuration information and was able to conclude that they were indeed all overclocked….
Getting weird crashes that nobody seems able to fix? Crashes that don’t make sense to the professionals?…
It could be overheating or malware.
PingBack from http://kernelmustard.com/2006/10/31/get-ready-for-more-random-crash-bugs/
PingBack from http://www.ljseek.com/on-overclocking_33857659.html
Gateway joins the fray.
PingBack from http://kylescove.com/2007/04/15/technology-links-04152007/
PingBack from http://jesus.spydro.be/weblog/2007-04-16/links-for-2007-04-15/
PingBack from http://www.shiftsolutions.com.au/?p=359
PingBack from http://www.cloman.com/chad/ChadsNews/2007/04/19/microsoft-actually-reads-those-error-reports/
PingBack from http://www.ellislive.com/blog/2007/05/23/feed-reader-core-dump-day/
PingBack from http://blogs.zdnet.com/BTL/?p=4845
Problem Description: A customer called in, complained angrily. “A call to ShellExecute API, passing in
PingBack from http://www.b0b.kwix.info/my-blog/?p=4
PingBack from http://www.secnews.pl/2007/10/07/wplyw-podkrecenia-cpu-na-kod-aplikacji/