Hacker News new | past | comments | ask | show | jobs | submit login
Safe VSP – 30 year old Commodore 64 bug demystified (2013) (linusakesson.net)
265 points by qmr on June 6, 2016 | hide | past | web | favorite | 22 comments

Aswesome :)

"Finally, this phenomenon hinges on the exact timing of the RAS signal at the nanosecond level, and on many machines the critical situation simply doesn't occur. The timing (and thus the probability of a crash) depends on factors such as temperature, VIC revision, parasitic capacitance and resistance of the traces on the motherboard, power supply ripple and interference with other parts of the machine such as the phase of the colour carrier with respect to the dotclock. The latter is assigned randomly at power-on, by the way, which could be the reason why a power-cycle sometimes helps."

Well of course LFT was the one who discovered the reason! His work is impressive. Example: http://www.linusakesson.net/scene/craft/index.php

Reminded me a bit about the modern DRAM issue, i.e., row hammer -- although the details are of course different -- https://en.wikipedia.org/wiki/Row_hammer

Slightly off topic but it's pretty cool to live in a day and age where people can post logic analyzer dumps of hardware processes and debug them together on the internet.

As a neuroscientist, this makes me feel better about this:


Shit like this is where I wish we could have a demo scene for modern architecture. (Maybe there is one, and I'm just don't know about it)

I have always fantasized about going back to the bad old days, where you have to boot the computer into the program you want to run.

I dream about having the full capability of modern computing to just run one program. Without any overhead. Without an OS.

You might be interested in https://en.wikipedia.org/wiki/Unikernel

Nothing is preventing you from doing that.

It is unfortunate 4164s dates before CAS before RAS refresh. Not all 41256s supports it either I think. There were even SIMMs with labels indicating that they are for Mac Plus/SE only when Apple had to use Siemens chips that don't support it.

One should add '[2013]'.

Bias for newness is a bad thing.

Fully agreed. But HN's convention is to append the year to articles from previous years. That isn't a bias for newness, but a bias for oldness—that is, for things that stand the test of time.

Historical material is more than welcome on Hacker News. One key function of this site is to distribute historical knowledge into what otherwise tends to be an amnesiac industry/community/society. More historical material, everybody, please!

Edit: Speaking of which, if you haven't been following it, Alan Kay has been posting some top-notch comments to HN lately: https://news.ycombinator.com/posts?id=alankay1

Adding the year does not devalue the resource. It is quite common for old but cool articles to be reposted.

Adding the year isn't a criticism, it's simply a marker that this particular article posted on Hacker News isn't new, so you don't get confused reading an article about something that happened years ago.

Having read this several years ago, I think [2013] is appropriate.

If you scroll to the bottom of http://www.linusakesson.net/scene/safevsp/index.php you can read "Posted Wednesday 20-Mar-2013 23:23". :-)

Not the point. The point is to have it in the title so everyone knows BEFORE clicking.

I understand where your heart is, but as a grocery shopper, I disagree.

He should have waited ten more years, then it would have been a 40 year old bug. Much more impressive! ;-P

Depends on how much oak you like in your bug.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact