Seen this problem before? Help me save some time. Thanks

[Rick Law]

Anyone know off hand what causes a problem like this?   (See attach image)

Boots with display messed up, letter doubling up and/or random characters, doubling up like turning "System Information" may become "SyytemmInfo mmtiin".  It will boot then some random color blocks resulting in blue screen or just hung at color blocks.

This is from an old Acer motherboard, running Intel chip set with on board Intel graphics.  Old machine that I use for backup which I occasionally boot to get to some old stuff.  This problem happened twice before but self-healed.

I am leaning that it may be either the graphic chipset on board, or the ram chips on board (which is also used by the graphis processor).  But if you've seen this before and know what it is right the way, it will save me some time.   I needed some stuff on this system rather soon, I rather not tear the machine apart to debug right now.  So, if you have some tips that can help be ID the problem quickly, much appreciated.

Thanks!

[helius]

bad address lines or buffers

[magic]

How does this machine even boot with that kind of data corruption

Maybe it is the video adapter Or RAM, or the root of all evil - electrolytic capacitors

[greenpossum]

You're in an alternate universe. 😉

[Rick Law]

bad address lines or buffers

Yeah, I was leaning in that direction also.  It seems the timing between writing the byte to the video buffer, and advancing to the next address (next character) is some what off.

No, but I have a (well validated) prejudice against RAM for anything I can't explain.

Only after the RAM passed memtest AND the gold fingers being rubbed and wiped, will I start suspecting other components.

The motherboard GPU uses shared memory 1,8,or 16mb.  It has been a long time since I last enabled BIOS ram test.  I am not sure what the proper behavior for this BIOS when it hits bad RAM.  When I enable long test now, it seem to go forever and never gave me an error message.  I usually got impatient by about 5 minutes and I pulled the plug.  I should let it run a bit longer and see if it give me an error message eventually.

How does this machine even boot with that kind of data corruption

Maybe it is the video adapter Or RAM, or the root of all evil - electrolytic capacitors

It boots (shows me the OS choices on the multi-boot menu) and shows me the "Windows did not boot last time"  in broken English, of course.  Well, more like stuttering English "Winnows dii nnttlaas..."  It then blue screens somewhere into the booting typically after a screen redraw that occur during normal boot presumably to tries to switch the screen resolution.  But now this guy typically blue screen out or hung now.

That it went so far is what me think of it is either video's share-ram hits some ram problem, or hits some timing problem or something like that.

This PC is tuck into a spot that is very difficult to work on.  I suppose I'll wait a bit for more insights from you the more experienced folks.  Hoping against hope that I might see some short cut some how.

All else fail, I will swap the memory out and see if it boots with the old memory I saved (2x 256MB) before I replaced them with the 1GB (2x 512MB) -- eons ago, when Dinosaurs roam America.  That should tell me if I have ram problem.  Getting the bloody thing out is going to be hell.  I also have an old PCI VGA card that I could use to see if it is VGA problem -- but I am not sure if that VGA card still works.

[magic]

It obviously has to be some highly localized failure because there is no way the machine would be able to even load an OS otherwise.

I'm not entirely convinced that it's RAM because how could RAM show errors at so many addresses within the VGA text buffer and zero or almost zero errors elsewhere? What would be the mechanism behind RAM duplicating bytes? That being said, if you have a possibility to boot from pendrives/CD, download and try to run memtest86+.

If you can, try to boot Linux in VGA text mode. If this loads and works fine (except for garbled text), then

Wild guess: dried up capacitor in the northbridge Vgpu buck converter.

BTW, how old is "old" and does it have the DRAM controller in the CPU or in the NB?

[Rick Law]

It obviously has to be some highly localized failure because there is no way the machine would be able to even load an OS otherwise.

I'm not entirely convinced that it's RAM because how could RAM show errors at so many addresses within the VGA text buffer and zero or almost zero errors elsewhere? What would be the mechanism behind RAM duplicating bytes? That being said, if you have a possibility to boot from pendrives/CD, download and try to run memtest86+.

If you can, try to boot Linux in VGA text mode. If this loads and works fine (except for garbled text), then

Wild guess: dried up capacitor in the northbridge Vgpu buck converter.

BTW, how old is "old" and does it have the DRAM controller in the CPU or in the NB?

The Acer F2 is a bit over 12 years old.  Good that at one time I used my laptop to locate a drivers for the F2's chipset, so I had a copy of the F2's driver manager screen print saved on the laptop. The device manager has the graphic chipset as Intel 82865G/PE/P/82848P.

Since I don't have Linux (I have a Linux install disk on ISO on that F2), so, I tried some other disks I do have that is display-heavy...  This trying Linux a good idea, I found some interesting clues I didn't see before. 

DOS running gdisk (from Norton's ghost), as expected it read the HDD just fine.  I've encountered HDD power issues and except for boot-disk, it rarely drops to blue-screening out.  Running DOS, it occasional missing a letter or two on the display.  I didn't want to run Ghost, that one is easier to make a mistaken and write something to the disk when ran in my machine's "semi-blind" mode.

A home made PXE (I use Bart PE's instruction to create it, and same GUI tools as Bart PE, but I use Server 2003 as the base OS), it fails similar to my normal WinXP boot.  This was tried before, so I have to find another display-heavy thing to try...

Seagate HDD Utility Disk - this one gave me some insights.  I use V1.1 (no GUI, but VGA), it use > CGA resolution to display 43 lines by 80.  This utility doesn't let you do a thing until you accept the user agreement, so it is safe even in "semi-blind" mode.

The user-agreement is long with multiple pages of disclaimer text that is scroll-able by page.  When I press random keys, it does a full screen erase and rewrite the exact same text; and if the key is held down, the full-screen erase and text rewrite repeats -- rapidly.  So, any change to the display stands out right the way!

Holding the key down, I notice some positions were failing to erase frequently, while some other position also fails but at a much lower frequency.  Some position doesn't seem to fail at all.  It is not what is displayed but the physical row/column position of letter.  The positions stay consistent after a reboot, and after power-cycle.  That to me points to specific bytes of memory somewhere, be it RAM or buffers on the GPU.

If it is a dry-out capacitor causing power issues, the failure location-consistency kind of say no to that, I think (or does it really?).  If it is a buffer inside the GPU, not much I can do there.  If it is a RAM failure, it has to be the page(s) where it was memory-mapped to the GPU or the display.

What do you guys think?

Thanks

[magic]

Well, this machine has DRAM controller in the chipset so if chipset VCC were bad, chances are it would have more effect than just screwed text console. Or maybe not, if the GPU core has a separate rail?

Anyway, capacitors or not, I think it's something with the motherboard. I can't imagine a mechanism behind RAM chips duplicating bytes. RAM simply has a parallel input bus, each received byte goes to separate transistor logic and separate storage cells. And the memory bus works on 64 bit words so there is no such thing as misdecoding an address by one and returning the adjacent byte.

edit
What the heck, see if you get any improvement with underclocking or overvolting a bit (if it's even possible on this board). That being said, I think the writing is on the wall for this machine anyway.

[DrG]

I think you should try to isolate one issue and proceed. In your case, there are a lot of examples of this happening. It looks like one good explanation (if there is one) is a video card problem. e.g.,  https://superuser.com/questions/613702/what-explains-the-garbled-message-start-wandows-ngrmadly-in-text-mode

Do you have the MB manual? Is it a desktop or a laptop (I didn't see where you actually identified the machine other than an Acer.

You have on board video - is that right? Can you disable that video and install a card? You said you have a VGA card, that may or may not be working - have you attempted to substitute that card?

What I am saying is that bad video is a strong possibility, so why not do what you can along those lines?

You appear to be taking pot-shots (e.g., not even letting a RAM check complete).....with a very old machine.

[Rick Law]

Well, this machine has DRAM controller in the chipset so if chipset VCC were bad, chances are it would have more effect than just screwed text console. Or maybe not, if the GPU core has a separate rail?

Anyway, capacitors or not, I think it's something with the motherboard. I can't imagine a mechanism behind RAM chips duplicating bytes. RAM simply has a parallel input bus, each received byte goes to separate transistor logic and separate storage cells. And the memory bus works on 64 bit words so there is no such thing as misdecoding an address by one and returning the adjacent byte.

edit
What the heck, see if you get any improvement with underclocking or overvolting a bit (if it's even possible on this board). That being said, I think the writing is on the wall for this machine anyway.

That old mother board is not a high-end one; it doesn't support over-clocking.  I can't for the world of me figure how it duplicates a write.  The ghost character scrolls with the line, so something got into display buffer ram.

Looks like I will have to dig out a similar motherboard that I think I got.  I hope I didn't remember wrong and really has it somewhere.

Man, I have over 100 storage boxes in my basement, on racks 7 boxes high and 4 deep.  The only way to get to the one behind or below is to remove the one in front (or above).  The price one pays for being a pack-rat.

Storage is like the inside of a drainage pipe.  In time, crud build up to a point when water can hardly flow anymore. 

I think you should try to isolate one issue and proceed. In your case, there are a lot of examples of this happening. It looks like one good explanation (if there is one) is a video card problem. e.g.,  https://superuser.com/questions/613702/what-explains-the-garbled-message-start-wandows-ngrmadly-in-text-mode

Do you have the MB manual? Is it a desktop or a laptop (I didn't see where you actually identified the machine other than an Acer.

You have on board video - is that right? Can you disable that video and install a card? You said you have a VGA card, that may or may not be working - have you attempted to substitute that card?

What I am saying is that bad video is a strong possibility, so why not do what you can along those lines?

You appear to be taking pot-shots (e.g., not even letting a RAM check complete).....with a very old machine.

Yeah, I am "taking pot-shots", hoping wishfully.

That PC is a "look at old stuff" machine.  Since I installed it, many things have piled on top, in front, and on both side of it.  It is already in a very difficult to access spot, added to that the extra load of stuff around it...  So I was looking for clues that may help me solve it easier without having to move the entire mountain range of old things -- may be just move a hill or two and solve the problem.  I am just being a "wishful thinker" perhaps.

I have the manuals in storage, but they were written for users who may have problem telling the difference between a CD-tray verses a self-retracting coffee cup holder on your PC.  So it wont be helpful.

I was doing something in Excel, I recall doing exactly the same eons ago, so I was trying to look it up to see how I did it...  That's when I discovered that this very old machine is probably too old now...

Writing (and reading) this thread is helpful, I've a two step game plan now:

1. I am going to locate the old ram, move things around minimally and see if the "old-new-ram" makes a difference.  If it works, I'll figure if I do a next step for that.

2. If #1 doesn't work, I will be needing to change mother board.  I'll just hope that I can find the motherboard, and it works.  Doing all that would be a lot of physical exercise (moving stuff) which I am ill equip to handle now, but it is worth a try.

[MrMobodies]

I have seen that sort of thing happen when the dimms are loose in the slots and not making contact to the pins, rarely if a secondary module was added on and the dimm slot becomes dirty/dusty over the time it was unpopulated and cracked boards and sometimes the artifacts on the screen changes when pressure is applied the area causing it.

[DrG]

I have seen that sort of thing happen when the dimms are loose in the slots and not making contact to the pins, rarely if a secondary module was added on and the dimm slot becomes dirty/dusty over the time it was unpopulated and cracked boards and sometimes the artifacts on the screen changes when pressure is applied the area causing it.

Not a bad suggestion. I remember this happening with an old Dell more years ago than I care to admit. It would give POST RAM errors and cleaning the contacts and re-seating was all it took. Ahh the good old bad days Speaking of which and not to thread drift too much...I had an XP machine on just a couple of days ago and for just the same reason as the OP - getting some old stuff. Then I had a two quick power outages - of course it was in the middle of re-booting when the second hit, and no, I don't have that dinosaur on a UPS.

Upon rebooting it was stuck in infinite chkdsk hell with the dirty bit on perma-set. This immediately gave me all sorts of nightmare re-experiencing...fortunately my memory and some searching provided the fix [letting chkdsk fix it in safemode (which takes about an hour) even though it doesn't tell you it is being run - just looks like you are stuck on loading mups]. I am a little embarrassed but thankful that I remember that - well after reading for a bit.

Yeah, good times 

[magic]

I have seen that sort of thing happen when the dimms are loose in the slots and not making contact to the pins

Okay, but have you seen duplicated letters or just some other kind of mangled text output?

[MrMobodies]

I have seen that sort of thing happen when the dimms are loose in the slots and not making contact to the pins

Okay, but have you seen duplicated letters or just some other kind of mangled text output?

Yes, I have come across all sorts of artfiacts.

I just remembered this happening in December 2017 but it was a failing power supply in an old Acer quad core. The text was similar to that and in the bios. It would hang at boot when it did that but that recovered after a couple of minutes when I reset it and then the USB started to do drop out too. Then I heard a high pitched noise coming from the power supply. Despite it meausing in the bios close to 12v and 5v when I pulled the power supply out the 12v was a little over 10v and the 5v was about 4.5v and it would slowly increase a bit with load using a Doctor Power 2 tester.


I also had overheated graphics cards in the past do that but that was mostly graphical artifacts and as well as text.

Also some undercooled laptops too that got too hot but I did find a few that had wonky dimm slots from Acer (some had plastic arms locking the dimms that snapped) and rarely dirty ram slots that got populated even though the first thing I suspected was the graphics controller.

According to this thread:
https://www.techsupportforum.com/threads/solved-graphic-artifacts-corrupted-post-characters.404879/

http://fc00.deviantart.com/fs49/f/2009/226/3/b/Screen___7_by_RedTalonTPF.jpg

That one turned out to be the graphics cards.

[mrflibble]

Based on past experience: suspect #1: vga card. suspect #2: bios eeprom flash. suspect #3: dram. suspect #4: power supply. suspect #5: your cat.

[Rick Law]

Based on past experience: suspect #1: vga card. suspect #2: bios eeprom flash. suspect #3: dram. suspect #4: power supply. suspect #5: your cat.

Funny thing, I just finished cleaning out some surrounding junk, enough to get the PC's case opened, and was trying to get my new PSU (which I have on hand)...  The cat was in my way so I decided to take a break and read some email/forums.  The second post I read is your...  suspect #5: your cat.

I do have suspicion that cat hair might have interfered with airflow.  Now I can get least look inside, we shall see.  Once I get the permission from the cat to do it...

[mrflibble]

Funny thing, I just finished cleaning out some surrounding junk, enough to get the PC's case opened, and was trying to get my new PSU (which I have on hand)...  The cat was in my way so I decided to take a break and read some email/forums.  The second post I read is your...  suspect #5: your cat.

I do have suspicion that cat hair might have interfered with airflow.  Now I can get least look inside, we shall see.  Once I get the permission from the cat to do it...

Impeded airflow is indeed "one of those things". Once after fixing a randomly hanging/crashing computer from a family member.. Q: "Hey, what did you do? It is working fine now." A: "Oh, I just opened the computer case, removed half a cat, and closed it again." That one had some serious cat hair buildup. Actually it was a small miracle that it would even work for as long as it did.

The advice you got in another post about reseating components is also worth doing. In the past many a problem has been solved by removing PCI card, inserting PCI card, job done. So in your particular case it might be worth it to open it up, remove mem + cards, vacuum clean it, and reinsert mem + cards. Or if you have a compressor, use that for some extra "holy shit, look at all that dust!" fun moments. ;->

All this is of course after getting permission to do so by His/Her Imperial Majesty.

[Halcyon]

Usually bad (or badly soldered) VRAM or some other connector issue.

[tpowell1830]

Anyone know off hand what causes a problem like this?   (See attach image)

Boots with display messed up, letter doubling up and/or random characters, doubling up like turning "System Information" may become "SyytemmInfo mmtiin".  It will boot then some random color blocks resulting in blue screen or just hung at color blocks.

This is from an old Acer motherboard, running Intel chip set with on board Intel graphics.  Old machine that I use for backup which I occasionally boot to get to some old stuff.  This problem happened twice before but self-healed.

I am leaning that it may be either the graphic chipset on board, or the ram chips on board (which is also used by the graphis processor).  But if you've seen this before and know what it is right the way, it will save me some time.   I needed some stuff on this system rather soon, I rather not tear the machine apart to debug right now.  So, if you have some tips that can help be ID the problem quickly, much appreciated.

Thanks!

This appears to be the BIOS screen. I have seen this happen when the BIOS battery gets low from being shutdown for a long time. Leaving the machine on this screen for awhile will recharge battery to some extent. Then reboot, may work normally.
I have also seen this happen when RAM connection gets bad. The fix is to unplug and replug the RAM chips multiple times to clean the contacts. Works sometimes, no guarantees.
This is the type of thing that happens when a computer sits for a long time unpowered.

Hope this helps...

[james_s]

I'd start by re-seating anything socketed, I've seen a lot of weird problems from oxidized pins or cheap sockets. Also while I don't recall seeing them on any PC motherboards, look closely for leaky surface mount electrolytic capacitors, old Macs are notorious for having problems with that, I just re-capped a IIci for somebody yesterday.

[Rick Law]

Holy cow, that was helpful.

I've seen a few dozens of machines on a factory floor with interior a lot dirtier yet motherboard/ram running just fine; clot up fans, yeah, but not bad contacts with dimms.

Yesterday, when I had moved the PC out just enough to open the cover, the cat interrupted me before I can try the new PSU.  Today, recalling the replies about reseating the DIMMs and I thought yeah, I have not found my old dimms yet, but with the inside accessible, I can try just cleaning it out first instead of waiting till I found the old DIMMs.

So, I took out the 2 dimms, cleaned the contacts, vacuum the empty dimm slots opening, vacuumed the inside fans.  I was expecting this not to do anything.  I was expecting I need DIMM or motherboard swap.  Cleaning is always a good thing anyhow...

With the dimm contacts sparkling like an actress's teeth, I put one dimm back - boot, BIOS screen AOK with single dimm.  Not a single dup characters or bad thing seen on the screen.  Put the other one in , boot, no problem!
 
I was not expected dirt alone would cause the problem.  I was convinced it was either bad DIMM or bad GPU...

I actually didn't need to look for any old parts at all but just my vacuum cleaner.  All the while, I have been waiting until I found everything I think I may need...

Thanks guys.

Good to see this old fellow running...  I am first copying some needed files - copying for just over 30 minutes now with another 30 minutes to go.  Once I have another copy, I will do further testing...

[LeonR]

Either isopropylic alcohol or electronic contact cleaners works wonders for computers connectors, slots and sockets.

I recommend memtest86+ ( http://memtest.org/ ) for at least 6 hours to rule out any more memory issues.

[james_s]

I had a similar problem with an arcade board I fixed some time back. It turned out to be a bad RAM that holds the background character assignments. One stuck bit meant that it was loading the wrong texture into cells calling for a character that had that bit in the opposite state. In this game the background image is made of tiles that are stored the same as text characters.

Ignore the skew at the top of the image, the monitor I was testing with didn't like the sync signal from that board.