I am curious as to why they would offload any AI tasks to another chip? I just did a super quick search for upscaling models on GitHub (https://github.com/marcan/cl-waifu2x/tree/master/models) and they are tiny as far as AI models go.

Its the rendering bit that takes all the complex maths, and if that is reduced, that would leave plenty of room for running a baby AI. Granted, the method I linked to was only doing 29k pixels per second, but they said they weren’t GPU optimized. (FSR4 is going to be fully GPU optimized, I am sure of it.)

If the rendered image is only 85% of a 4k image, that’s ~1.2 million pixels that need to be computed and it still seems plausible to keep everything on the GPU.

With all of that blurted out, is FSR4 AI going to be offloaded to something else? It seems like there would be a significant technical challenges in creating another data bus that would also have to sync with memory and the GPU for offloading AI compute at speeds that didn’t risk create additional lag. (I am just hypothesizing, btw.)

Future chips not affected by THIS cpu bug yet.

When you powered an old CRT down, the image would collapse to a single horizontal line, and then to a point.

I was thinking you could just cut the power to the deflecting coils.

It would be close to infinite refresh rate as the electron beams wouldn’t need to move. Also, the phosphors in that spot would probably burn away in a few minutes resulting in no refresh rate at all.

FYI, you can download your photos in bulk with Google Takeout, but you need to have enough space in Google Drive to do it. (Takeout zips up all your photos and will drop 10GB chunks in Drive.)

I was doing something similar to you recently. I downloaded all my photos and de-duped by generating MD5 hashes for all the pictures that were downloaded. (I was moving all of my photos to a local NAS, so it wasn’t quite what you are doing.)

If your dups have consistent MD5 hashes, that might work for you but it’s hard to say.

That almost got me excited for a second.

I am going to need your 50 point summary of those obvious points in the longest form possible by this afternoon so I can be completely convinced that I have already made up my mind in the correct way. Thanks.

It was on old 3.5" drives a long time ago, before anything fancy was ever built into the drives. It was in a seriously rough working environment anyway, so we saw a lot of failed drives. If strange experiments didn’t work to get the things working, mainly for lulz, the next option was to see if a sledge hammer would fix the problem. Funny thing… that never worked either.

I used to take failed drives while they were powered on and kinda snap them really with a fast twisting motion in an attempt to get the arm to move or get the platters spinning.

It never worked.

Did you get bad sectors? Weird things can absolutely happen but having sectors marked as bad is on the exceptional side of weird.

Maybe? Bad cables are a thing, so it’s something to be aware of. USB latency, in rare cases, can cause problems but not so much in this application.

I haven’t looked into the exact ways that bad sectors are detected, but it probably hasn’t changed too much over the years. Needless to say, info here is just approximate.

However, marking a sector as bad generally happens at the firmware/controller level. I am guessing that a write is quickly followed by a verification, and if the controller sees an error, it will just remap that particular sector. If HDDs use any kind of parity checks per sector, a write test may not be needed.

Tools like CHKDSK likely step through each sector manually and perform read tests, or just tells the controller to perform whatever test it does on each sector.

OS level interference or bad cables are unlikely to cause the controller to mark a sector as bad, is my point. Now, if bad data gets written to disk because of a bad cable, the controller shouldn’t care. It just sees data and writes data. (That would be rare as well, but possible.)

What you will see is latency. USB can be magnitudes slower than SATA. Buffers and wait states are causing this because of the speed differences. This latency isn’t going to cause physical problems though.

My overall point is that there are several independent software and firmware layers that need to be completely broken for a SATA drive to erroneously mark a sector as bad due to a slow conversion cable. Sure, it could happen and that is why we have software that can attempt to repair bad sectors.

If I remember correctly from the good ol’ days, promoted sites/ads kinda stopped after the first few pages and it skipped past most of the SEO crap.

But yeah, I forgot how aggressive Google is with ads and AI now, those bastards.

It’s actually a good change, IMHO. You can just click directly on page 10 for anything that might be related to your original search.

It’s been around for a while. It’s the fluff and the parlor tricks that need to die. AI has never been magic and it’s still a long way off before it’s actually intelligent.

Yeah, I would think memory as well due to the screen artifacts in that low res mode. (That depends on how x86 memory is mapped these days, I suppose.)

Ultimately you need only a tiny fraction of that data to emulate the human brain.

I am curious how that conclusion was formed as we have only recently discovered many new types of functional brain cells.

While I am not saying this is the case, that statement sounds like it was based on the “we only use 10% of our brain” myth, so that is why I am trying to get clarification.

This bubble is quite bubbly. There is an AI company for anything and everything now. The market is almost fully saturated with “AI” everything.

Just like the web bubble, all of the intsta-AI shops need to fail so the real tech can grow. AI is never going to go away, but most of the scam companies will fail in due time.

We might have one big consolidation, or several. The hype will die and the quick money will disappear. It’s the same story, every time. One the magic AI box stops shitting out dollar bills, we should be good.

For anyone not wanting to read into the Reddit thread and is also from the US, it seems the “normal” tax rate is ~40% for people in the UK. So, 23% is a fairly large tax break.

I dunno how accurate that is, but if it’s wrong, Cunningham’s Law absolutely applies.