These exposed dies can take a surprising amount of abuse since most of the die is bulk silicon. The only part carrying actual transistors or metal layers are just a couple of micrometers at the bottom of the die.
About the broken die cpu, it might still be broken. As someone who assembled hundreds of Athlons back in the early 2000's I still remember the problems that those caused, usually the broken corners caused errors in extended instructions, we did a video processing test to check if they were ok or not, if they are broken usually the compression ended unexpectedly or the video contained green artifacts. If I recall it right we transcoded from uncompressed AVI to MPEG 2.
Glad that those finally work! Back then we used copper spacers to sort of protect cpus, although I've never chip a corner when mounting heatsinks without spacers, it's always been a concern for me. Also note that Socket A CPUs have 4 foam resting pads for the heatsink, in order to have the heatsink mounted as even as possible, but some heatsink mounting clips were still prone to die chipping (I'm looking at your, Volcano 7+). Tony, there's more have fun to be had with those ceramic Athlon and Duron: the pencil trick. That'll allow you to unlock the multiplier. You can overclock or simply run at the same frequency but at a faster FSB.
Great catch by the commenters that caught the jumper on this board for the FSB setting, and it's awesome that those two AMD processors work. Especially that "Morgan" core Duron processor, I feared that the chipped corners were going to have done that processor in... and yet it's a trooper! (Yeah, I've always thought it was cool that the code name for a processor shares my last name, even if it was only a Duron part, and even if it's actually named after the breed of horse.)
Very nice I was sure those CPUs were fine! The mustard color Athlon that has the X mark in the tray and you did not test it now with JP1 closed knowledge what CPU was?
Didn´t expeted the Duro too work because of the mangled die. My first Duron had 700MHz and just a tiny bit on the left side is missing and its dead like a doornail. So be happy that you have two survivors!
I used to have a similar AMD CPU probably a decade ago at this point which had chipped corners and while it worked it was very very unstable the computer acted very weird etc. It would be interesting to run a stress test on that to see if it’s stable or not.
ah, interesting. To be honest I feel that if the die is even slightly compromised, the CPU wouldn't work at all. But who knows. Thanks for mentioning, I'll bear that in mind!
The video issue in PC Player 640x480 benchmark is due to an issue between the software and the GPU and not an issue with the CPU. I've seen more issues on newer graphics cards and less on older. I enjoyed this video. Glad it popped up on my recommended.
Glad viewers caught this. I've built more socket 462 gaming rigs than I can count, and still miss the jumper occasionally. They're my go to for an 98 SE challenge. Look out for these low wattage models, as they're excellent dies: axmh2400fqq4c, axmh2500fqq4c, axmd2400fjq4c, axmf2500fvq4c.
i have a Duron that's chipped too, and still works! wow, that was so close :)) what material did you use for the CPU tray, Tony? it looks very nice and smooth. do you have a code, or something, for pcbway? wanna order some cpu trays and also found an SD adapter to the 3.5 bay
I added the link on the main video: ua-cam.com/video/xAcd3lh_MfE/v-deo.html Ideally you want ESD plastic but that's more expensive... I love those trays, finally no box anymore :)
Look for a used ender 3 printer, they are like 100 bucks these days. Cheaper then let pcbway print stuff like that. Once you have a printer you start printing more and more tools, trays and such.
@@tony359 damn it, i ordered PETG cause i did watercooling long ago and it seemed smooth like that. anyway, o rdered one piece of sd card adapter and i'll see how it goes with petg. 8 bucks is not that much
A 1200B, great for a 100FSB board... I had a GA-7IXE4 and had to change CPU, as I got a 1000C which only ran at 750 - got it replaced with a 1000B The board was a desperation choice when the 686B bug was news, so I didn't want VIA (or the half and half AMD760/VIA 686B) and also it had 2 ISA slots when I really would have liked 3 and most others had just 1. Trying to remember my full ISA loadout that came from a Tyan Tomcat Lasat Safire 56k modem - and it BLASTED a good V90 speed, V90 after I firmware updated from K56 SB16 ASP with midi DB Iomega Ditto Dash tape controller Additional Parallel port controller to avoid chaining printer and scanner. I ended up replacing the modem with a Cirrus Logic HaM chipset (controllerless, rather than a total softmodem) - which was also a solid V90 performer, and the sound with a PCI128
Awesome! I think only very late S462 motherboards were fully jumper less (nforce2 & VIA KT600 for example) and run by default at 100MHz, it's pretty common to find these with batteries that gad run out showing CPU speeds and models that never existed, you can leave the jumper at 100 as a safety measure, it will just under lock any compatible CPU.
And that’s why one should read the motherboard manual. I check all jumpers and their meanings. You mentioned the CPU spec, and that is part of the process, before even trying on a motherboard: you can fry the CPU depending on the motherboard/chipset generation. CPU Vcore compatibility comes first, and the FSB as you noticed. With Retroweb and Cpu-world at everyone’s disposal, it should be a mandatory sequence
@@tony359 we all make mistakes. Believe it or not, I fried a DVD Rom drive by inserting the molex power conector inverted: it shouldn’t be possible, but one can do that with enough force (doesn’t need to be extreme). My eyesight is not what it used to be, and now I use a flashlight and sometimes magnifying glasses to avoid such silly mistakes. That was a wakeup call, and I’m glad it was not a critical component. Thanks for sharing
My Duron had also cracked edges, but that wont harm them really because the logic is on the underside of the silicon. Using a higher FSB ist an easy way to overclock those CPUs, +33% should do almost all cpu's, some can even do +66%, my Duron back in the day could even do +100%, but had to crank up the VCore a bit. Imho it was a Duron 750 running at 1.5Ghz. I don't really remember the cooling, either a watercooler or compressor. That were the best times for overclockers. :)
yes I remember all the overclocking techniques and you could get quite a lot out of it! Nowadays it's less critical, CPUs are basically overclocked by the manufacturer :D
Put those cpus in extensive stress test, especially involving video encoding/decoding. I had Athlon with Barton core with damaged die, which seemed to be fine on a first glance, but overheated like hell under stress.
@tony359 Impossible, I double checked everything, including testing other cpu, which was fine. Plus that faulty Barton overheated in other motherboard as well.
@@tony359 It is called Socket 462, also known as Socket A. Not to be confused with slot A. I am a loser but at least I know my wikipedia articles by heart.
@@tony359 Yes, their Slot was Slot A (if I recall, same connector as Slot 1, but reversed) and the socket was Socket A... but there were never slotkets for AMD!
I have a small graveyard of Motherboards and Graphics cards that have failed in the past. Mostly from the early Athlon era, but all sorts of odds and sods. I kept them to harvest spare connectors and capacitors, so some parts may be missing. Would this be of any interest to you?
Hi - thanks for the offer. Athlon MoBos might be already "too modern" for me, if you happened to have something older than that, let me know. My email is on my profile!
if the cache breaks ( the big longated square) disable cache with a another cpu in the bios and you got a pentium 1 but superfast one but still slow as a duron or athlon -imagine like k2-500 speeds or even slower
@@tony359 Just wondering because I have this old Intel Celeron Q038A171-08423 sitting in an old laptop. I know the CPU works because I've used it before.
@@tony359 Oh the great days of pencil or conductive paint... some boards could set multiplier (dip switch or BIOS) but the multiplier bridges were cut on later CPUs - equally, the CPU also had bridges to set the multiplier itself and those could be reworked
Be carreful with those old AMD cpus, make sure to use some thermal paste since it is bare silicon and the overheat protection might not be fast enough to save it. Perhaps it was not shown in the video if you used thermal paste or not. 🙂
I did! :) But thanks, I'm aware those CPUs are very delicate. In fact, it's why I didn't want to install/remove the heatsink every time, the spring is very stiff and I feel I am about to destroy everything every time! :) But yes, if the heatsink tilts for a moment, bye-bye CPU.
These exposed dies can take a surprising amount of abuse since most of the die is bulk silicon. The only part carrying actual transistors or metal layers are just a couple of micrometers at the bottom of the die.
About the broken die cpu, it might still be broken. As someone who assembled hundreds of Athlons back in the early 2000's I still remember the problems that those caused, usually the broken corners caused errors in extended instructions, we did a video processing test to check if they were ok or not, if they are broken usually the compression ended unexpectedly or the video contained green artifacts. If I recall it right we transcoded from uncompressed AVI to MPEG 2.
interesting. Maybe that was the reason for the video corruption?
@@tony359 Yes, it's possible. If with other Duron in the same mainboard you don't get the corruption probably the CPU is damaged.
I might change the green check to a yellow question mark? :)
Good that you figured it out. These CPUs can be harder to work with. The die can chip easily and coolers can be tricky.
yes, I remember the fear of attaching the heatsinks on those CPUs - hence using the "temp" one on these two.
Glad that those finally work! Back then we used copper spacers to sort of protect cpus, although I've never chip a corner when mounting heatsinks without spacers, it's always been a concern for me. Also note that Socket A CPUs have 4 foam resting pads for the heatsink, in order to have the heatsink mounted as even as possible, but some heatsink mounting clips were still prone to die chipping (I'm looking at your, Volcano 7+).
Tony, there's more have fun to be had with those ceramic Athlon and Duron: the pencil trick. That'll allow you to unlock the multiplier. You can overclock or simply run at the same frequency but at a faster FSB.
yes, I know there is a lot to do with those CPUs but I'm happy to leave the overclocking to others! :) I'll bear that in mind though :)
Great catch by the commenters that caught the jumper on this board for the FSB setting, and it's awesome that those two AMD processors work. Especially that "Morgan" core Duron processor, I feared that the chipped corners were going to have done that processor in... and yet it's a trooper! (Yeah, I've always thought it was cool that the code name for a processor shares my last name, even if it was only a Duron part, and even if it's actually named after the breed of horse.)
Very nice I was sure those CPUs were fine! The mustard color Athlon that has the X mark in the tray and you did not test it now with JP1 closed knowledge what CPU was?
I did but that it's a 133MHz one, it's toasted at the back. I tested it anyways just for fun and still dead :)
Didn´t expeted the Duro too work because of the mangled die. My first Duron had 700MHz and just a tiny bit on the left side is missing and its dead like a doornail. So be happy that you have two survivors!
My duron (950 mhz I believe) with chips in both corners makes the HDD screech. Weird...
And its not the mobo beeper!
Have a nice weekend Tony!
Thanks, you too!
I used to have a similar AMD CPU probably a decade ago at this point which had chipped corners and while it worked it was very very unstable the computer acted very weird etc. It would be interesting to run a stress test on that to see if it’s stable or not.
ah, interesting. To be honest I feel that if the die is even slightly compromised, the CPU wouldn't work at all. But who knows. Thanks for mentioning, I'll bear that in mind!
Thank's for the folow up video, this made my evening! :)
Glad you enjoyed it!
This is half the fun with old computers.
ahah yes, those newer boards were "boring" compared to the, say, 486 boards with more jumpers than components!
The video issue in PC Player 640x480 benchmark is due to an issue between the software and the GPU and not an issue with the CPU. I've seen more issues on newer graphics cards and less on older. I enjoyed this video. Glad it popped up on my recommended.
I'd imagined that, the video adaptor is in the chipset on this board. Thank you!
Glad viewers caught this. I've built more socket 462 gaming rigs than I can count, and still miss the jumper occasionally. They're my go to for an 98 SE challenge. Look out for these low wattage models, as they're excellent dies: axmh2400fqq4c, axmh2500fqq4c, axmd2400fjq4c, axmf2500fvq4c.
Thank you!
YES!
i have a Duron that's chipped too, and still works! wow, that was so close :)) what material did you use for the CPU tray, Tony? it looks very nice and smooth. do you have a code, or something, for pcbway? wanna order some cpu trays and also found an SD adapter to the 3.5 bay
I added the link on the main video: ua-cam.com/video/xAcd3lh_MfE/v-deo.html
Ideally you want ESD plastic but that's more expensive... I love those trays, finally no box anymore :)
Look for a used ender 3 printer, they are like 100 bucks these days.
Cheaper then let pcbway print stuff like that.
Once you have a printer you start printing more and more tools, trays and such.
Yes, I can imagine once there is a 3D printer, then the sky is the limit! :)
What I don't have is the space....
@@tony359 damn it, i ordered PETG cause i did watercooling long ago and it seemed smooth like that. anyway, o rdered one piece of sd card adapter and i'll see how it goes with petg. 8 bucks is not that much
A 1200B, great for a 100FSB board... I had a GA-7IXE4 and had to change CPU, as I got a 1000C which only ran at 750 - got it replaced with a 1000B
The board was a desperation choice when the 686B bug was news, so I didn't want VIA (or the half and half AMD760/VIA 686B) and also it had 2 ISA slots when I really would have liked 3 and most others had just 1.
Trying to remember my full ISA loadout that came from a Tyan Tomcat
Lasat Safire 56k modem - and it BLASTED a good V90 speed, V90 after I firmware updated from K56
SB16 ASP with midi DB
Iomega Ditto Dash tape controller
Additional Parallel port controller to avoid chaining printer and scanner.
I ended up replacing the modem with a Cirrus Logic HaM chipset (controllerless, rather than a total softmodem) - which was also a solid V90 performer, and the sound with a PCI128
Awesome! I think only very late S462 motherboards were fully jumper less (nforce2 & VIA KT600 for example) and run by default at 100MHz, it's pretty common to find these with batteries that gad run out showing CPU speeds and models that never existed, you can leave the jumper at 100 as a safety measure, it will just under lock any compatible CPU.
I'll bear that in mind - I should have checked on the manual! :)
And that’s why one should read the motherboard manual. I check all jumpers and their meanings. You mentioned the CPU spec, and that is part of the process, before even trying on a motherboard: you can fry the CPU depending on the motherboard/chipset generation. CPU Vcore compatibility comes first, and the FSB as you noticed. With Retroweb and Cpu-world at everyone’s disposal, it should be a mandatory sequence
of course - I just assumed and I made a mistake :)
@@tony359 we all make mistakes. Believe it or not, I fried a DVD Rom drive by inserting the molex power conector inverted: it shouldn’t be possible, but one can do that with enough force (doesn’t need to be extreme). My eyesight is not what it used to be, and now I use a flashlight and sometimes magnifying glasses to avoid such silly mistakes. That was a wakeup call, and I’m glad it was not a critical component. Thanks for sharing
My Duron had also cracked edges, but that wont harm them really because the logic is on the underside of the silicon. Using a higher FSB ist an easy way to overclock those CPUs, +33% should do almost all cpu's, some can even do +66%, my Duron back in the day could even do +100%, but had to crank up the VCore a bit. Imho it was a Duron 750 running at 1.5Ghz. I don't really remember the cooling, either a watercooler or compressor. That were the best times for overclockers. :)
yes I remember all the overclocking techniques and you could get quite a lot out of it! Nowadays it's less critical, CPUs are basically overclocked by the manufacturer :D
Nice!
Thanks!
Put those cpus in extensive stress test, especially involving video encoding/decoding. I had Athlon with Barton core with damaged die, which seemed to be fine on a first glance, but overheated like hell under stress.
Good to hear that - so weird though. You think that maybe the damage might cause the heatsink not to adhere flat to the die?
@tony359 Impossible, I double checked everything, including testing other cpu, which was fine. Plus that faulty Barton overheated in other motherboard as well.
This must be socket A. Used to have one of those. I remember some of the processors could be overclocked using a pencil on those gold pads!
That's Socket 432. Never had a Socket A one.
@@tony359 It is called Socket 462, also known as Socket A. Not to be confused with slot A. I am a loser but at least I know my wikipedia articles by heart.
I couldn't be more inaccurate on my previous reply then :D
@@tony359 Yes, their Slot was Slot A (if I recall, same connector as Slot 1, but reversed) and the socket was Socket A... but there were never slotkets for AMD!
I have a small graveyard of Motherboards and Graphics cards that have failed in the past. Mostly from the early Athlon era, but all sorts of odds and sods. I kept them to harvest spare connectors and capacitors, so some parts may be missing. Would this be of any interest to you?
Hi - thanks for the offer. Athlon MoBos might be already "too modern" for me, if you happened to have something older than that, let me know. My email is on my profile!
The CPU die will heat up more at the broken corners because there is no contact with the cooler
But can we start the overclock party?
ahahah - Happy to leave that to other channels, I am not an overclocker person :)
Be careful, the foam pads will stop the cooler from touching the die when you just rest it on top like that. The cooler has to be pushed down!
Yes, I was pushing my luck there. But the cooler is heavy enough to compress those pads thankfully :) It was actually getting pretty hot!
What paint markers do you use? They look like the ones my friend used to have; the ones that don't erase easily with alcohol. I need some of those!
"POSCA" :)
Que maravilha eles estarem "vivos"! Muito bom!!!!
Thank you!
The sweetest victory moment is: yeah it worksssss!!!!
if the cache breaks ( the big longated square) disable cache with a another cpu in the bios and you got a pentium 1 but superfast one but still slow as a duron or athlon -imagine like k2-500 speeds or even slower
interesting thanks!
Just a question. Could the socket from the video by chance be found in old laptops to?
Old laptops often had socketed CPUs - though they were "mobile" CPUs. I am not sure about AMD though.
@@tony359 Just wondering because I have this old Intel Celeron Q038A171-08423 sitting in an old laptop. I know the CPU works because I've used it before.
Heh, sometimes it's the *little* things you miss!
Those CPUs could work with 133 MHz FSB with low multiplier (for example 11x model OCed to 133x9=1197 MHz).
interesting - can you change the multiplier on that board?
@@tony359 You can change it on the CPU (voltage too, so be careful).
Google for ''AMD Athlon pencil mod'' or ''AMD Socket-A 133MHz FSB/DDR Overclocking Guide''.
@@tony359 Oh the great days of pencil or conductive paint... some boards could set multiplier (dip switch or BIOS) but the multiplier bridges were cut on later CPUs - equally, the CPU also had bridges to set the multiplier itself and those could be reworked
Be carreful with those old AMD cpus, make sure to use some thermal paste since it is bare silicon and the overheat protection might not be fast enough to save it. Perhaps it was not shown in the video if you used thermal paste or not. 🙂
I did! :)
But thanks, I'm aware those CPUs are very delicate. In fact, it's why I didn't want to install/remove the heatsink every time, the spring is very stiff and I feel I am about to destroy everything every time! :)
But yes, if the heatsink tilts for a moment, bye-bye CPU.
😅❤
ONLY AMD are real CPU's !
ahah AMD forever :)