...it's probably not the graphics card.
Oh dear.
20051025
20051024
I say, I say, I say... my computer won't boot
Then how do you use it?
I can't
Further updates on the health of hughanchor the computer, since you seem to be so desperately interested. After a very healthy day on Saturday, things took a turn for a worse overnight. I got up this morning to find a new windows XP stop error -- something about a bad IRQ level, or something (actually, it was "IRQL_NOT_LESS_OR_EQUAL"). I started trying to see whether the memory was the cause, so I started moving the memory around. This only seemed to make things worse, so I put in the old memory I had before I upgraded this summer. No joy, in fact the machine would consistently reboot after startup almost immediately: just after beginning to boot windows, just before the "Windows loading" screen would appear. So it now seems unlikely that the memory is to blame, since I've tried several different lumps of memory in it, and unless they are all broken, then it should be fine.
In fact, I managed to boot from the install CD to a prompt and explore the hard drives quite happily. I also downloaded and burned a copy of "memtest86" on my laptop (the mugwump2, god bless all who sail in her), and got hughanchor to boot that and run memory tests for several hours with no errors detected.
OK, so where does this leave us? Certainly, the computer is in reasonable shape, modulo the fact that it can't boot up its regular operating system. After my initial wrong accusations, the power supply seems fine (I checked it with a multimeter and through CMOS): although VCORE seems rather low at 1.44V, I think this is just a feature of my machine. CMOS also tells me that the operating temperature of those components that are measured is not too high (about 36 degrees C, it seems). Memory seems to be OK. Disks had a reasonably thorough checkdisk run on them this morning before XP started refusing to boot. Of course, any or all of these assertions could be hopelessly wrong, and may be the cause after all.
But there is one possible suspect who is beginning to look more likely. Can you guess who it is? It is my once trusted graphics card. Consider the evidence: the generic stop errors I've been encountering can be caused by bad memory anywhere in the system -- including the graphics card. The problem currently manifests itself as windows is about to start up -- which is when it switches graphics modes to put a pretty logo up while it loads. Text only applications, such as the memtest86 programs, seem to run happily with no problems. And there's one other symptom I haven't mentioned: now, during the bootup process, my monitor flashes up a brief error message (from the monitor itself): graphics mode not supported: Horizontal 31.2KHz, Vertical 59.5Hz. Which has never appeared before.
So, the chase is closing in. It is starting to look like the graphics card is at fault. Or is it? Some or all of these errors could also be explained by a faulty motherboard, (maybe the dilithium timing crystals are off, which is causing various synchronization errors). Perhaps it could be the L1 or L2 cache on the processor? Or maybe something I've tried to rule out, like the memory, is actually the problem?
The scientific method here is to replace each possible component in turn, and see if things improve. I was fortunate to have some extra memory lying around so I could do that test, but I don't have a large stock of spare graphics cards, motherboards or processors lying around. If I did, I'd probably build another computer with them. Oh well. Thank goodness I have a laptop (try searching the internet for debugging information when your machine won't even boot...) and a collection of DVDs to catch up on. I'm too busy to give this my full attention, with paper deadlines and some brief international travel coming up in the next few weeks, so I doubt I'll have the full answer any time soon. Maybe I'll have to admit defeat and get it taken to a specialist.
I can't
Further updates on the health of hughanchor the computer, since you seem to be so desperately interested. After a very healthy day on Saturday, things took a turn for a worse overnight. I got up this morning to find a new windows XP stop error -- something about a bad IRQ level, or something (actually, it was "IRQL_NOT_LESS_OR_EQUAL"). I started trying to see whether the memory was the cause, so I started moving the memory around. This only seemed to make things worse, so I put in the old memory I had before I upgraded this summer. No joy, in fact the machine would consistently reboot after startup almost immediately: just after beginning to boot windows, just before the "Windows loading" screen would appear. So it now seems unlikely that the memory is to blame, since I've tried several different lumps of memory in it, and unless they are all broken, then it should be fine.
In fact, I managed to boot from the install CD to a prompt and explore the hard drives quite happily. I also downloaded and burned a copy of "memtest86" on my laptop (the mugwump2, god bless all who sail in her), and got hughanchor to boot that and run memory tests for several hours with no errors detected.
OK, so where does this leave us? Certainly, the computer is in reasonable shape, modulo the fact that it can't boot up its regular operating system. After my initial wrong accusations, the power supply seems fine (I checked it with a multimeter and through CMOS): although VCORE seems rather low at 1.44V, I think this is just a feature of my machine. CMOS also tells me that the operating temperature of those components that are measured is not too high (about 36 degrees C, it seems). Memory seems to be OK. Disks had a reasonably thorough checkdisk run on them this morning before XP started refusing to boot. Of course, any or all of these assertions could be hopelessly wrong, and may be the cause after all.
But there is one possible suspect who is beginning to look more likely. Can you guess who it is? It is my once trusted graphics card. Consider the evidence: the generic stop errors I've been encountering can be caused by bad memory anywhere in the system -- including the graphics card. The problem currently manifests itself as windows is about to start up -- which is when it switches graphics modes to put a pretty logo up while it loads. Text only applications, such as the memtest86 programs, seem to run happily with no problems. And there's one other symptom I haven't mentioned: now, during the bootup process, my monitor flashes up a brief error message (from the monitor itself): graphics mode not supported: Horizontal 31.2KHz, Vertical 59.5Hz. Which has never appeared before.
So, the chase is closing in. It is starting to look like the graphics card is at fault. Or is it? Some or all of these errors could also be explained by a faulty motherboard, (maybe the dilithium timing crystals are off, which is causing various synchronization errors). Perhaps it could be the L1 or L2 cache on the processor? Or maybe something I've tried to rule out, like the memory, is actually the problem?
The scientific method here is to replace each possible component in turn, and see if things improve. I was fortunate to have some extra memory lying around so I could do that test, but I don't have a large stock of spare graphics cards, motherboards or processors lying around. If I did, I'd probably build another computer with them. Oh well. Thank goodness I have a laptop (try searching the internet for debugging information when your machine won't even boot...) and a collection of DVDs to catch up on. I'm too busy to give this my full attention, with paper deadlines and some brief international travel coming up in the next few weeks, so I doubt I'll have the full answer any time soon. Maybe I'll have to admit defeat and get it taken to a specialist.
20051022
How Mysterious
I've been obsessively reading books of "Solve it Yourself" mysteries lately. These are short vignettes -- a page or two -- detailing a crime, which you (the reader) must solve. These puzzles are inevitably quite irritating, since they almost invariably follow the same pattern: in the description is some detail which, combined with some external knowledge shows a contradiction in the story of one of the characters, indicating them to be a liar and hence guilty. At the best of times these stretch credulity. My favourite of these is one in which the murder victim shot his precious diamond out of the window at the top of his narrow house attached to an arrow to avoid the murderer getting his hands on it. The investigating officer tells the chief suspect that the missing diamond is only "an arrow flight away", to which the suspect responds "Well, go outside and look for it then". This immediately proves his guilt. Because, apparently, any innocent person hearing the sentence would assume that the officer said "It's only a narrow flight [of stairs]" away, and respond "Well, go upstairs and look for it then." Clever, eh? And also possibly the most convoluted and forced set up you can possibly think of.
The principal author of this crimes against logic is Donald J. Sobol, the creator of Encylopedia Brown, a children's version of the same thing. For four dollars, I bought a copy of a collection of his "two minute mysteries", a 400 page book where each mystery occupies two pages. That's 2c (approximately 0.01GBP) per mystery, and in several cases I feel that I have been cheated of my two cents worth. I'm going to quote one mystery in its entirety, because I can, and because it's fair use, and because this one above all others seems simply wrong in its execution. Take a read and see what you think:
After this question, at the bottom of the same page and upside down is the answer. In a vain attempt to create a false sense of tension, I shall try to write the answer using a white font, which will therefore be invisible until you highlight it with your cursor, unless your browser ignores these kind of instructions, or you are using a text or speech based browser. But anyway:
Now here's my challenge to you. Read the story and the author's explanation to the answer. Why is the explanation complete and utter bollocks?
Post your responses in the comments, and if no one can work it out in a reasonable amount of time, I'll post the reason myself.
The principal author of this crimes against logic is Donald J. Sobol, the creator of Encylopedia Brown, a children's version of the same thing. For four dollars, I bought a copy of a collection of his "two minute mysteries", a 400 page book where each mystery occupies two pages. That's 2c (approximately 0.01GBP) per mystery, and in several cases I feel that I have been cheated of my two cents worth. I'm going to quote one mystery in its entirety, because I can, and because it's fair use, and because this one above all others seems simply wrong in its execution. Take a read and see what you think:
The body of Monroe Sheld, an impoverished sixty-year-old painter, lay over a table in his sweltering one-room apartment. A bullet had entered his right temple. On his right side, by the leg of his chair, was an old fashioned single-action revolver. The table was bare, except for a cracked plate, a saucerless cup, knife, fork, and spoon, and an overturned salt shaker.
"Sheld has been peddling his sketches for a dollar or two in the local bars," said Inspector Winters. "His doctor says he was stricken with heart failure two months ago and nearly died. You migh say he ate a last meal and committed suicide."
"Correction," said Dr. Haledjian. "There was a bit about him in today's paper. A small but fashionable art gallery announced a one-man showing of his paintings next month. He had everything to live for."
Haledjian opened a small cupboard layered with dust. One plate and cup, and a setting of cheap silverware, however, was quite clean. Next he opened a paper garbage bag and sniffed.
"This food was put in the bag with the past three hours, or about the time of Sheld's death," asserted Haledjian. "Longer than that and it would have begun to spoil and smell in this heat."
Haledjian shut the bag. "I shouldn't close the case quite yet, Inspector. Sheld had a dinner guest, who tried to cover up his presence here."
What made Haledjian so certain?
After this question, at the bottom of the same page and upside down is the answer. In a vain attempt to create a false sense of tension, I shall try to write the answer using a white font, which will therefore be invisible until you highlight it with your cursor, unless your browser ignores these kind of instructions, or you are using a text or speech based browser. But anyway:
Sheld, having suffered a heart failure would have watched his diet carefully. He never would have brought the salt to the table unless it was for another's use
Now here's my challenge to you. Read the story and the author's explanation to the answer. Why is the explanation complete and utter bollocks?
Post your responses in the comments, and if no one can work it out in a reasonable amount of time, I'll post the reason myself.
More computer based hijinks
Following up on the last post, I've made some interesting progress on the constant rebooting problem. It's still intermittent: sometimes the machine stays up for ages, sometimes it reboots over and over. During one of these moody phases, I noticed a flicker of blue and white on the screen before the system reset. Could this be a blue screen? I hadn't seen one of them on the machine for ages.
Digging around deep in the XP system options, I found a setting for "Restart after system error" which is set to true by default. I switched it off, and went back to work. After an hour or so, the machine flicked up a BSOD, with the irritatingly cryptic error "Page_fault_on_nonpaged area". That is, the reboots have been generating error messages. But those redmond based idiots have the default setting to reboot the instant the error is generated, so unless you have a photographic memory, you'll miss the error entirely.
This means that my initial conjecture about a duff power supply is probably barking up the wrong tree. Some rooting about on the internet suggests that the most likely causes are (a) a newly installed device (b) some duff memory (c) some disk problems. Since I haven't put anything new in the system in the last few weeks, that suggests (b) or (c) are the most likely culprits. Since I installed some new memory earlier this summer, it's possible that that is the culprit. Or could be some bad sectors (do they still exist? My knowledge of computer architecture is rapidly becoming obsolete). So, after the next hang up, I'll swap over the memory chips and run some serious hard disk scanning on the offchance it's a fixable bug. If it still breaks, then I can try removing one or both of the memory chips to see if that helps to fix it. No solution yet, but at least the suspects are a little narrowed down.
Digging around deep in the XP system options, I found a setting for "Restart after system error" which is set to true by default. I switched it off, and went back to work. After an hour or so, the machine flicked up a BSOD, with the irritatingly cryptic error "Page_fault_on_nonpaged area". That is, the reboots have been generating error messages. But those redmond based idiots have the default setting to reboot the instant the error is generated, so unless you have a photographic memory, you'll miss the error entirely.
This means that my initial conjecture about a duff power supply is probably barking up the wrong tree. Some rooting about on the internet suggests that the most likely causes are (a) a newly installed device (b) some duff memory (c) some disk problems. Since I haven't put anything new in the system in the last few weeks, that suggests (b) or (c) are the most likely culprits. Since I installed some new memory earlier this summer, it's possible that that is the culprit. Or could be some bad sectors (do they still exist? My knowledge of computer architecture is rapidly becoming obsolete). So, after the next hang up, I'll swap over the memory chips and run some serious hard disk scanning on the offchance it's a fixable bug. If it still breaks, then I can try removing one or both of the memory chips to see if that helps to fix it. No solution yet, but at least the suspects are a little narrowed down.
20051016
Mmm, biscuits
I was browsing the aisles of a nearby supermarket, and hidden in the ethnic foods section under, bizarrely "Irish", I found that they stocked Bourbon creams. Naturally I snapped up a couple of packets. It's such a struggle finding real food in this country that you have to take the opportunity when you can. Still it's odd to think that people view 'Irish' as a food style to beplaced alongside Indian, Chinese, Mexican and Spanish. The shelf also contained other random items, such as Ambrosia Rice Pudding, HP Sauce and "Irish" Tea (distinct from the shelf upon shelf of "regular" tea elsewhere in the store). Well, I won't complain if it means I can enjoy a few more home comforts.
Meanwhile, hughanchor has not been well. That's hughanchor, my home computer, rather than Hugh Anchor, the ill-considered fictious weblog author. hughanchor (the computer) has been rebooting repetitively, often when I click on a new file. That's quite annoying. It almost certainly isn't caused by software, since it sometimes reboots a few times in the middle of booting up. It's quite annoying, and after checking all the leads and blowing the dust out of the innards, the problem seems to remain. It's most likely a power problem, although difficult to know what to do about this: probably get a new power supply, which means getting a new case and attempting a live transplant which would be quite time consuming and fiddly. It might just be that my electrical supply is dodgy, although there's relatively little one can do about that, and not much one can do to diagnose it short of packing the machine into a vehicle and taking it somewhere else to see if it works any better. Of course, after snapping in and out of life repeatedly over lunch, it seems to be behaving this afternoon when it doesn't have anything to do, but I don't really want to trust it with anything too important for a while. And naturally this problem is exactly the kind of "heisenbug" that you are never entirely sure whether it is gone or whether it has just gone away for a bit.
Ah well, I'll just nibble on a bourbon and see if it gets better over the next week.
Meanwhile, hughanchor has not been well. That's hughanchor, my home computer, rather than Hugh Anchor, the ill-considered fictious weblog author. hughanchor (the computer) has been rebooting repetitively, often when I click on a new file. That's quite annoying. It almost certainly isn't caused by software, since it sometimes reboots a few times in the middle of booting up. It's quite annoying, and after checking all the leads and blowing the dust out of the innards, the problem seems to remain. It's most likely a power problem, although difficult to know what to do about this: probably get a new power supply, which means getting a new case and attempting a live transplant which would be quite time consuming and fiddly. It might just be that my electrical supply is dodgy, although there's relatively little one can do about that, and not much one can do to diagnose it short of packing the machine into a vehicle and taking it somewhere else to see if it works any better. Of course, after snapping in and out of life repeatedly over lunch, it seems to be behaving this afternoon when it doesn't have anything to do, but I don't really want to trust it with anything too important for a while. And naturally this problem is exactly the kind of "heisenbug" that you are never entirely sure whether it is gone or whether it has just gone away for a bit.
Ah well, I'll just nibble on a bourbon and see if it gets better over the next week.
20051015
Highly Amateur Electronics Repair
Fans of my life will recall the Great Aerolineas Argentinas Baggage Robbery of 2004. One of the victims of the thieving baggage handlers was a portable travel alarm clock. This was a small LCD display item, that served just the purpose I needed, so I had to hunt around for a bit to find a replacement, and eventually found this (honestly, must have taken about five minutes and $10 in WalMart).
This roughly worked, and spent most of the time living on my shelf in the living room, with brief trips to Canada, Japan nd parts of Europe. Last week it developed a fault. If you look at the picture in the link above you'll see that it has four settings: time set, alarm set, alarm on and alarm off. It should spend most of its time in the "Alarm off" mode, just displaying the time. But for some reason, although the switch was set to off, the alarm symbol on the display was on. No amount of fiddling with the switch, blowing on it, etc., would get the symobol off. I ignored it, but the next morning at the appointed time, the alarm started going off. I tried taking the battery our and other basic tricks, but to no avail: the clock was in permanent alartm mode.
So tonight I decided to have a go at fixing it. I am of course completely incompetent with electronics (that's hardware; I do software). I unscrewed the back 4 tiny screws, to find that the PCB was held on to the front with eight even tinier screws. Hopeing that this would not keep going in some infinite recursion, I took them off, and inspected the front of the PCB. I cleaned some dust off, wiggled the switch around and put the battery back in. The alarm indicator was still on, and the alarm still went off when the time came round. Should I just bin this and spend another $10 on a new alarm clock?
In what I can only describe as a stroke of genius, I saw a solution. Trailing from the back of the electronics was a pair of wire going into a mini speaker that made the alarm go off. I screwed the tinier screws back on, and checked that the thing still worked. Then, with the utmost of care, I pulled the wires out of the speaker. Now when the alarm "goes off", no sound can be heard. And that's what I wanted. Sure, it's now only a clock, instead of an alarm clock, and for some reason the "nighlight" feature that allows you to read the time in the dark no longer works, but I claim this as a concrete result. Besides, my cell phone and MP3 player and socks now all seem to have alarm clocks and lights built in as throwaway extra features.
So there you go. The message is in this era of disposable goods and it being cheaper to replace than repair, the well intentioned amateur can still take stuff apart and reassemble it with only slightly less functionality at the end than at the beginning. A triumph for the small consumer.
This roughly worked, and spent most of the time living on my shelf in the living room, with brief trips to Canada, Japan nd parts of Europe. Last week it developed a fault. If you look at the picture in the link above you'll see that it has four settings: time set, alarm set, alarm on and alarm off. It should spend most of its time in the "Alarm off" mode, just displaying the time. But for some reason, although the switch was set to off, the alarm symbol on the display was on. No amount of fiddling with the switch, blowing on it, etc., would get the symobol off. I ignored it, but the next morning at the appointed time, the alarm started going off. I tried taking the battery our and other basic tricks, but to no avail: the clock was in permanent alartm mode.
So tonight I decided to have a go at fixing it. I am of course completely incompetent with electronics (that's hardware; I do software). I unscrewed the back 4 tiny screws, to find that the PCB was held on to the front with eight even tinier screws. Hopeing that this would not keep going in some infinite recursion, I took them off, and inspected the front of the PCB. I cleaned some dust off, wiggled the switch around and put the battery back in. The alarm indicator was still on, and the alarm still went off when the time came round. Should I just bin this and spend another $10 on a new alarm clock?
In what I can only describe as a stroke of genius, I saw a solution. Trailing from the back of the electronics was a pair of wire going into a mini speaker that made the alarm go off. I screwed the tinier screws back on, and checked that the thing still worked. Then, with the utmost of care, I pulled the wires out of the speaker. Now when the alarm "goes off", no sound can be heard. And that's what I wanted. Sure, it's now only a clock, instead of an alarm clock, and for some reason the "nighlight" feature that allows you to read the time in the dark no longer works, but I claim this as a concrete result. Besides, my cell phone and MP3 player and socks now all seem to have alarm clocks and lights built in as throwaway extra features.
So there you go. The message is in this era of disposable goods and it being cheaper to replace than repair, the well intentioned amateur can still take stuff apart and reassemble it with only slightly less functionality at the end than at the beginning. A triumph for the small consumer.
20051014
Irritation of the day
Conferences that require you to submit the paper in the proceedings format, and then announce that the camera ready version must be two pages shorter than the page limit for submissions. Given that no one reads printed proceedings any more, and electrons are relatively cheap, this seems like a particularly irritating decision just to annoy authors.
If you are not a researcher in computer science this will not make much sense to you. So instead, have a surreal lolly stick joke that I sucked off yesterday:
Q: The Ivy League
A: Where do plants play football?
Either this is a strange new direction into "Jeopardy"-style joke telling, or they had the stick put into the machine the wrong way round.
If you are not a researcher in computer science this will not make much sense to you. So instead, have a surreal lolly stick joke that I sucked off yesterday:
Q: The Ivy League
A: Where do plants play football?
Either this is a strange new direction into "Jeopardy"-style joke telling, or they had the stick put into the machine the wrong way round.
20051010
Wikipedants
I've had this link open in my browser for a few weeks: Clifton College.
I found the page when searching for the words to that trite "hush in the close tonight" poem, and on first glance I wondered if the whole thing was an elaborate hoax. The School Customs, and the Marshal who is banned from all local drinking establishments struck me as particularly suspect. But, on reflection, the whole thing makes perfect sense. I think it's fair to say that almost any British public school of sufficient age will look like a parody to an outsider. They're all barmy.
Meanwhile, it is moderately diverting to pore through the revision history and spot the various vandalism attempts, presumably by pupils of the school.
I found the page when searching for the words to that trite "hush in the close tonight" poem, and on first glance I wondered if the whole thing was an elaborate hoax. The School Customs, and the Marshal who is banned from all local drinking establishments struck me as particularly suspect. But, on reflection, the whole thing makes perfect sense. I think it's fair to say that almost any British public school of sufficient age will look like a parody to an outsider. They're all barmy.
Meanwhile, it is moderately diverting to pore through the revision history and spot the various vandalism attempts, presumably by pupils of the school.
20051002
Non-existent word
"Paleozooogy" is not a real word.
Type it into a search engine, and you'll get about 700 hundred hits. It must be a real word, right? Look, it's defined here: "The science of extinct animals, a branch of paleontology.". Definitely a real word.
Check the other pages: they all have the exact same definition, attributed to Webster's 1913, an American dictionary that is conveniently out of print and out of copyright. Somewhere in the process of converting this dictionary to electronic form, a typo entered the process. Clearly, this should be 'paleozoology', but someone forgot to press the 'l', or it never got scanned properly. To check this, I got my more recent (1962) edition of Webster's and looked it up: no 'paleozooogy', but definitely 'paleozoology'. The electronic form of this type has now been copied round a few hundred times, but all from one flawed source.
So, paleozooogy. Not a real word.
Which is a shame, since I could really have done with a real non-onomatopoeic word that has three o's in a row next to each other. But alas, it seems destined not to be. Booooo!
Type it into a search engine, and you'll get about 700 hundred hits. It must be a real word, right? Look, it's defined here: "The science of extinct animals, a branch of paleontology.". Definitely a real word.
Check the other pages: they all have the exact same definition, attributed to Webster's 1913, an American dictionary that is conveniently out of print and out of copyright. Somewhere in the process of converting this dictionary to electronic form, a typo entered the process. Clearly, this should be 'paleozoology', but someone forgot to press the 'l', or it never got scanned properly. To check this, I got my more recent (1962) edition of Webster's and looked it up: no 'paleozooogy', but definitely 'paleozoology'. The electronic form of this type has now been copied round a few hundred times, but all from one flawed source.
So, paleozooogy. Not a real word.
Which is a shame, since I could really have done with a real non-onomatopoeic word that has three o's in a row next to each other. But alas, it seems destined not to be. Booooo!
Subscribe to:
Posts (Atom)
