dagaci

this already happened -> https://news.ycombinator.com/item?id=49421158 -> My passing comment mentioning Apple Reference Image in the same thread was moderated down into oblivion for some reason!

I have a RTX PRO 6000 96GB when the pricing was way better than now i also have a RTX 5090 too.

What I noticed is that (1) the great local models are optimized run inference (diffusion & LLMs) well on 32GB VRAM <= GPU's because that that's what the target has ...

(2) The quality of local models (esp. in diffusion) is increasing faster than the need for more VRAM - additional reason for the value of these FAST GPUs to increase!

(3) RTX PRO 6000 96GB is really great for fine tunes (ai-toolkit) :) but doesn't outperform my RTX 5090 with inference by anything significant on the good local models.

I have never run an AI job on a Mac, i also have doubts about performance and compatibilities - since the reviews almost never compare directly.

This a variant of the provenance scheme: C2PA its implemented by all major Camera maker, and Google it seems, Apple support it with 3rd party apps on iPhone, but they have something called "Apple Reference Image" brewing.

Personally I think there is a good argument for being able to distinguish AI generated image and video...

Googling you can see the source code for watermarking here https://github.com/microsoft/InvisMark

Note: Those results are a little misleading the good LTX results were generated with a good workflow in ComfyUI, and a prompt expanding local model.

Whilst the H3 result was generated with the raw api using his raw prompt.

If you ask ChatGPT or other AI model to "improve" your prompt (with cinematic, good lighting) generally, you will get also very good result from H3 also.

My own H3 test show that H3 (API edition) is undoubtedly better than LTX in prompt adherence ! The real comparison will be with the edition of H3 we get to run locally.

Yes, google they paid $900 millions in 2015 for about 7% stake, 100X. % Number probably fell because of all the "merging".

So the general public get to bail-out & subsidize Anthropic / and downstream xAI/SpaceX via the State Wallet, but of course with no access.

IIS also sits at the back of a many "modern" cloud web type services.

Its understated, but this kind move is now systemic in the EU due to the sanctioning of ICC & EU officials and random people who hurt the presidents feelings requiring Microsoft to remotely kill access to resources tied to Microsoft Accounts.

Without rules of law its literally irresponsible for EU to have this kind of heavy dependency on US corporations.

A nice little AI review with comparison of the CPU/Power Draw & Networking would be interested in seeing a fine-tuning comparison too. I think pricing was missing also.

Imagine your a super smart 'aztec' citizen and you watch your leading priests sacrificing 1000s of people in the name of a god which is probably imaginary ... you have the choice to dumb down (re-occupy your mind with work, duties, mundane, gish gash) and stop over thinking or ... continue to contemplate the true horror's that your world is indulging in or something else even...

I liked Express VPN

Lol and I had to pay >$60 a share for my little bundle of AMD!

No problem — I'm sure their Supremely nice Leader up north kept a backup. He's thoughtful like that...

Amazing. iOS only, with region restrictions in 2025.

The criteria is someone with 1 million in liquid investable wealth - homes are not investable assets or liquid.

This is actually stated in the report xD

The media hysteria about a millionaire exodus happened in 2024, not 2025. 10900 articles were in 2024 not 2025 so the hysteria happened in 2024. The interesting part is: those spreading hysteria don’t need to provide evidence; they just simply repeat.

Then those locked into the X-rinse-and-repeat-loop echo the same hysteria, as if they’ve just discovered some kind of wisdom when they have simply been overtrained.

I hear American's have to get in debt for over 100k for the same education thats almost free in so many places, so it might be a kind of balancing things..

Jenson has managed to kneel into every market boom in a reasonable amount of time with his GPUs and tech (hardware and software). No doubt he will be there when the next boom kicks off too.

Microsoft fails consistently ... even when offered a lead on the plate... it fails, but these failures are eventually corrected for by the momentum of its massive business units.

Apple is just very very late... but this failure can be eventually corrected for by its unbeatable astroturfing units.

Perhaps AMD are too small keep up everywhere it should. But compared to the rest, AMD is a fast follower. Why Intel is where it is is a mystery to me but i'm quite happy about its demise and failures :D

Being angry about NVIDIA is not giving enough credit to NVIDIA for being on-time and even leading the charge in the first place.

Everyone should remember that NVIDIA also leads into the markets that it dominates.

SimpleTuner is dependant on Microsoft's DeepSpeed which doesnt work on Windows :)

So you probably better off using Ai-ToolKit https://github.com/ostris/ai-toolkit

Some error corrections: ebay bought skype in 2005 for 2.6B (1 million concurrent user), and Microsoft bought them in 2011 for 8.5B (22 million concurrent users) with 74 million concurrent users reported in 2013 falling to apparently 30 million now...

https://www.statista.com/chart/1417/skype-usage/

What is your criteria for stewardship ?

Surely they will have receipts for the tariffs paid…

Remember these are elons are script kiddie hackers, it only occurred to disable the outer firewall, azure ad will independently geoip block all by itself

Russian IPs were in the pool because it never occurred to them to check where these IPs were geo registered

MI300X (RunPod) 192gb ram Hourly Rate: $2.49/hr. Break-even Point: You can rent for 2,410 hours (~100 days of non-stop-continuous use) before reaching the cost of the $6000 Mac. Mac's top out at 192GB not 2TB ;) Consideration: If your AI training requires sporadic use (e.g., a few hours daily or weekly), renting is significantly cheaper. MI300X will also get you result many times faster too, so you could probably multiply that 100 days!

Because sensible people just use the cloud at this point, you can probably get several years of training for $6000

So based on the 4090 prices and the small footprint, great thermals of the 5090. It doesn't seem like a bad deal to get 32GB VRAM and the fastest inferencing performance for $2K. Its looking like a 40% increase in performance (not that the 4090 was slow by any means).

Devtoys is great and open-source and .NET and cross-platform * https://devtoys.app/

and in context to the thread files-community/files is great and open-source and .NET * https://files.community/ (its actually 96% .net and 3.1% c++)