I had another brand-new, identical Samsung SSD, so I hooked both the old and new drive up to a Linux laptop (with USB cases) and tried to dd the old drive to the new drive. That mostly worked, but VERY VERY slowly: it would run fast for 5 seconds and then have no activity for 30 seconds. I had a fan blowing on the old drive to keep it cool because it was running very hot.

The dd copy would eventually fail and then I'd restart it with appropriate iseek and oseek values. I also did a cmp /dev/zero with the new disk to verify that it was all zeroes (it was brand new), and that allowed me to use conv=sparse on the dd. The reason for that was to avoid writing to ever sector of the new disk; I didn't want to copy sectors from the old drive that had never been accessed (she only used about 250GB of the 1TB).

It took a couple of days and about 5 restarts to finish the copy, but it did work, and as a precaution, I made another copy of the drive and ran a cmp of the original drive and the 2nd copy (also having to restart cmp several times). Since that compare worked, I knew that all 3 drives had identical content. The new drive worked fine in her laptop and she was mighty glad to see her Windows login screen.

The thing that made this work, IMO, is that Linux has a longer timeout for errors than Windows apparently does, especially during the boot sequence. Plus Linux allows adjusting the drive timeout, so if the device is doing error recover, which is sometimes slow, it gives it time to finish rather than reporting an error.

One of my theories was that the bad SSD was overheating, but if that was the case, a cold boot should have worked, with the failures only coming later.

The other theory is that one of the chips on the SSD failed, so the drive was having to use the ECC codes to correct for the missing information, and the correction process was taking longer than Windows boot would tolerate.

https://paddockpost.com/2025/02/14/executive-compensation-at...

If people are giving stuff to Goodwill with no compensation, I'd say they definitely would give eBay stuff for a 30-40% cut.

In the real world, you don't hire a plumber and expect him to also do your landscaping, fix your car, and tailor your clothes. It would seem like a much better use of resources if I could download an app that specialized in shell, Python, and C coding for example, or maybe even that would be 3 apps that communicated. Maybe I could even run them on a regular machine with 16GB of RAM. I don't need one huge model that can do that and code in Fortran, COBOL, and Lisp.

As humans, we've done pretty well by specializing. I hope this gets explored more with smaller, focused AI models vs the current path of one model to rule them all that can only be run in a data center the size of a country.

PID COMMAND %CPU 1200 Firefox GPU 319.2

I do use uBlock Origin and NoScript addons, but these are disabled in private windows and this GPU thing happens with either regular or private windows where the addons aren't running. I only noticed this because eventually the fan will ramp up.

Comments URL: https://news.ycombinator.com/item?id=47976527

Points: 2

# Comments: 1

See "The Forbin Project": https://vimeo.com/584593423

As author of HashBackup, I know people are using it with GCS, and I'd like to be able to test against it, but not enough to swallow a large surprise Google bill.

I ended up paying $12K to Mayo for a week of appointments. Private insurance, if I could have gotten it, would have been at least $1000/mo for premiums (in 2020) plus $10K deductible, so I actually saved money just paying Mayo instead of getting private insurance.

IMO the only reason insurance companies allowed the ACA to pass was the stipulation that everyone in the US was required to get insurance coverage or face a penalty. When the Supreme Court ruled that provision illegal, I'm sure the insurance companies were furious that they were duped.

It has been a very educational journey and remains interesting to me 18 years later - longer than any other project I've worked on. I've had thoughts about commercializing it over the years, but as has been mentioned, navigating online payment services, making a fancy portal, testing it with several browsers, etc. is not that interesting to me: the programming details are the interesting part.

HashBackup has always had a quarterly expiration date built in to the backup portion in case I did decide to commercialize it or a company wanted to acquire it, but recently I decided to ditch that idea and removed the expiration date. I've recently had some health issues that are going to cost society a not-insignificant amount, so letting HashBackup remain free is a way for me to feel like, in a small way, I'm paying it back a little.

My parents married right out of high school, which was pretty much the norm I think. I lived on a dead-end street where nearly every house had kids my age. Dads worked, moms didn't. Moms might babysit, iron, do laundry for others, etc., but moms took care of the house and the kids. The houses were 850 sq ft, most with 3 (small!) bedrooms, a kitchen a living room, and 1 bath. We lived in that house until I was 8 and my sisters were 6 and 2, so 5 of us in 850 sq ft.

My dad worked as a bag boy at Kroger during high school and could: - get married - buy a house after a year married - start a family at 20 - had 1 car for the family - had a boat - had a motorcycle right out of high school. There's no way an unskilled high-school kid could do that today. They'd be lucky to have a car and be able to fill it with gas and have car insurance.

I don't think most people today would consider that lifestyle feasible, but at the time, it was fine. I don't think it's doable today because both parents have to work since inflation over the decades has had a dramatic effect on prices.

I announced it on the Prime Usenet group and a guy wrote to say he was interested in using it for production, but had been burned before and wanted to make sure it worked. So I set him up a VM in my basement, he uploaded a bunch of his executables, and they all worked. I charged $1000/mo for the emulator. Since he was in the EU, I knew I wouldn't have a leg to stand on to get paid if they decided to stiff me, so I did some investigating into a Matrix dongle. It was pretty slick, and I coded up a copy protection scheme that allowed the emulator to run only when the dongle was inserted, and it contacted a license server in my basement to verify that it was allowed to run. To make sure my basement server wasn't a critical resource, I had the dongle setup like a battery that got periodically "charged": if the license server was down, the emulator kept running for up to N days on its local charge, giving me time to solve issues with the license server.

This went on for a few years, but they were never happy they had to actually keep paying for the technology. I heard every excuse in the book about why they needed a backup dongle. I knew the real reason they wanted one was to hire someone to hack it, but unfortunately, that would mean removing it, and that would halt the emulator within 5 minutes. So they came up with excuses like:

"Our state regulator requires us to have a backup computer system, so we need 2 dongles". No, you start computer 2, move the dongle, and you're fine.

One time I heard from a guy who said he heard about the emulator from a Prime parts guy in Chicago I knew, and wanted to try it out for a large EU bank. I asked where to send it, and surprise - it was the same city as my other customer! So I call R in Chicago and asked him how the conversation went with this "banking" customer, and R says "It was kinda weird. He said he had a Prime with a failing disk drive, but when I asked what model drive he had, he didn't know. He kept asking if there was any other solution for him besides hardware until R says "Well, there's a guy that has an emulator", which the banking guy jumped on immediately. So I go to Google and lookup this address the banker gives me to send the dongle to, and it's a massage parlor!

All in all, it was a good deal for me and a good deal for him, even if he wasn't happy about it. He was getting paid to support his own Prime software that he didn't have to port. But I learned my lesson as a youngin' about companies saying they will pay for something. I sold some printer/spooler minicomputer software in my early 20's and had a large investment bank (huge actually, as in everyone at the time would recognize their name) try it out for a month. They called and said they were processing the one-time license fee of $1500 but could I give them the access code early so they wouldn't have a disruption for a few days. I did it, and never got a dime from them. Hard lesson learned about being nice.

I meant to say fooxscale, but had a typo. Turned out to be interesting after all.

According to their AI:

"The optimum Python regular expression to match the words foo, fooxcale, and fooyscale as whole words is \bfoo(?:x|y)?cale\b|\bfoo\b"

That actually matches foocale (wrong), foo, fooxcale and fooycale (wrong), and doesn't match fooyscale as requested.

I have to give them credit though: their AI did give me good advice the other day about an SQLite GLOB query that was not using an index in my program but did use it when I typed the query in by hand. I worked on it for about 30 minutes, then asked. Turns out that if you use a ? parameter for GLOB, SQLite can't tell whether it starts with a wildcard or not, so never uses the index. To make it use the index, you have to string splice the parameter value (after SQL quoting of course) rather that use the ? for parameter substitution. I was pretty amazed it answered that exactly right and gave me the hint about doing the splice, and warning about injection attacks.

I also think it'd be a good idea if SQLite did this test inside the VM program it generates rather than at compile time, generating code to both use the index and not use it and choosing based on the runtime test of the first character. It confused me severely.

Comments URL: https://news.ycombinator.com/item?id=46561875

Points: 1

# Comments: 0