Marvin

Mike Wuerthele said:

ssfe11 said:

Maybe I’m missing something but Tim meeting with Trump and in Feb and then both announcing the 500b US investment has total exemption written all over it. Again am I missing something? Would welcome comments. 

You're missing the fact that the President said yesterday that there were no exemptions.

There are some exemptions listed on the site:

https://www.whitehouse.gov/fact-sheets/2025/04/fact-sheet-president-donald-j-trump-declares-national-emergency-to-increase-our-competitive-edge-protect-our-sovereignty-and-strengthen-our-national-and-economic-security/

"Some goods will not be subject to the Reciprocal Tariff. These include: (1) articles subject to 50 USC 1702(b); (2) steel/aluminum articles and autos/auto parts already subject to Section 232 tariffs; (3) copper, pharmaceuticals, semiconductors, and lumber articles; (4) all articles that may become subject to future Section 232 tariffs; (5) bullion; and (6) energy and other certain minerals that are not available in the United States."

https://www.msn.com/en-us/money/other/semiconductors-are-exempt-from-trump-s-massive-32-tariff-on-taiwan-though-pc-gamers-will-still-feel-the-heat/ar-AA1Cev7t

It's not clear how much impact it will have on Apple's products.

This move is being widely criticized but other countries have been unfairly locking US companies out of trade for a long time. While this likely won't have a good outcome, continuing to let other countries engage in unfair business practices shouldn't just be accepted either. These tariffs are in response to the unfair trading practises of other countries. The expected outcome is that other countries start trading fairly and drop their original unfair business practises and tariffs and the US can do the same.

https://ustr.gov/issue-areas/reciprocal-tariff-calculations

If the US tariffs were removed tomorrow, all that happens is US businesses go back to being taken advantage of. It's clearly a ham-fisted approach with the intent of trying to get a quick fix similar to the war negotiations but the US has a huge debt running out of control:

https://www.msn.com/en-us/money/markets/us-could-hit-debt-ceiling-as-early-as-mid-july-study-shows/ar-AA1ByPeh

This is what the government spending cuts and tariffs are about. If there's a better way to fix the debt issue other than cutting spending and trying to make up trade deficits, they'd probably like to hear some ideas. These moves suggest they don't have any better ones.

michelb76 said:

I had high expectations, but my god this runs bad. 23fps on an M4 MacBook air at the low preset to 500p render target. Barely hitting 40fps on the medium preset on my M3 Max, with a lot of low fps moments. Crazy.

There were a few Macs tested here and they all run pretty badly:



M3 Max looks playable around 50FPS after turning raytracing down but it doesn't look good without raytracing effects.

It runs poorly on low-end PCs too, the Nvidia 3050 is just above the performance of the Air and gets around 30FPS:



At 3:43 they enable frame-gen and get 60FPS. MetalFX will have frame-gen at some point, Cyberpunk 2077 was reported to support it, maybe they will enable it at a later date.

It should perform better on the M3 Max chip, maybe they haven't used the Metal hardware RT APIs properly and it's falling back to software RT.

The minimum requirements for the game at 1080p/30 are actually pretty high, roughly 8TFLOPs GPU. Only the Max chips and latest Pro chips are at this level.
Recommended GPU is around 13TFLOPs, which is essentially M3 Max or above.

This is why Apple having a reference renderer library would help developers. On Mac and iOS, game devs could use Apple's rendering engine and know it will perform ok. They probably have one internally already that they use to help 3rd party developers out with their proprietary engines. If they partnered with Microsoft, Sony and Valve, maybe Pixar, they could make a reference renderer that supports DirectX, Metal and Vulkan. This would support path tracing, approximated path tracing, shadow maps, mesh shading and multi-pass rendering. Royalty-free, possibly open-source so game studios can modify it. Post-production rendering software has multiple rendering engines like V-Ray, Redshift, Arnold, Octane etc, game engines should have the same.

brianus said:

tiredskills said:

Why on earth would I want to run an AI model?  Locally or otherwise?

I’m sure this was meant to be snarky, but for me it’s a genuine question: what are the envisioned real world use cases? What might a business (even a home one) use a local LLM for?

The article mentions a hospital in the context of patient privacy, but what would that model actually be *doing*?

In hospitals, AI models are reviewing patient scans to detect cancer:

https://www.youtube.com/watch?v=Mur70YjInmI

This is image analysis rather than text but text models can be used for medicine. There's an online AI for free here:

https://duckduckgo.com/chat

It can be asked about medical issues like if there's a pain somewhere, what could it be and what treatments are available e.g 'What medicine is typically used to treat acid reflux?'.

In a clinical setting, a doctor would review the recommendations.

In business, they'd be better off using a custom AI model that is trained on high quality data. A legal company might train a model on past cases and they can quickly find similar cases to use as references.

Local models are usually more responsive (if the hardware is fast enough), don't get timeouts and you can save past prompts more easily. They would likely still be cloud-based so that all employees can access them from lightweight clients, just a company cloud server.

blastdoor said:

At about 5 minutes and 30 seconds he says that building this with consumer PC hardware would be "quite expensive." I was looking for a fair bit more precision than that. 

Specs are listed here:

https://geekbacon.com/2025/02/20/running-deepseek-r1-671b-locally-a-comprehensive-look/

It needs multiple 3090 or higher GPUs + 512GB RAM. There's a video here showing a $2000 setup but it only runs at 3 tokens/s:

https://www.youtube.com/watch?v=Tq_cmN4j2yY&t=2822s

Another uses an Nvidia 6000 that costs around $7k for the GPU:

https://www.youtube.com/watch?v=e-EG3B5Uj78&t=560s
https://www.newegg.com/pny-vcnrtx6000ada-pb/p/N82E16814133886

Performance is 4 tokens/s. The video in the article mentioned the M3 Ultra was around 17 tokens/s.

This is one area where Nvidia and AMD are worse value and they do it on purpose because they make a lot of their revenue from this where they lower the memory in the consumer GPUs and charge a lot for enterprise GPUs with more memory that is needed for AI.

This video tests Nvidia H100 GPUs x8 ($28k each - https://www.newegg.com/p/N82E16888892002 ), which gets 25 tokens/s:

https://www.youtube.com/watch?v=bOp9ggH4ztE&t=433s

If Nvidia sold a model of the H100 with 512GB of memory, it could probably compete with M3 Ultra but would cost more than $30k just for the GPU.

Applications that need lots of unified memory is where Apple's hardware design is very competitive and they knew this when designing it.

Alex_V said:

Errors, misinformation, hallucinations, dubious results,, ill-advised stuff, you name it. So Apple pressed ‘pause.’ Because 50% reliable is not good enough if you’re Apple. 

50% reliable is fine for companies like ChatGPT, or any one of the dozens of other AI entrants. Never mind that their entire business model rests on plagiarising and circumventing copyright of the original creatives around the world. 

That's what pundits like Gruber are missing. Other products that Apple has made in the past had a deterministic outcome whether it was software or hardware. It doesn't matter how someone uses an iPhone or an OS, it will behave mostly as intended. AI, especially generative AI, is non-deterministic and unpredictable because the amount of inputs and outputs are so many that it can't be fully tested.

People like to rewrite history about Apple delivering timely products. There were promises about the G5 chip for years that never panned out, reaching certain clock speeds and making it into laptops. This was promised for multiple years by Steve Jobs and this was a deterministic product.

Attempting to sugar-coat Apple's history is just an excuse to complain about Apple. Apple's products reach over 1.5 billion people overnight, they have to work much harder to make sure they perform as expected.

wozwoz said:

Seriously, how long can it take to release a monitor update? The existing model sits too high ... and I have been waiting a long time for one that is not ergonomically compromised.  The current height variable Studio Display only goes UP:  hopefully, Apple will figure out how to make it go down in height as well.

That's what the VESA mount adapter option is for. Then get a 3rd party stand or swivel arm:

https://www.amazon.com/StarTech-com-Single-Monitor-Stand-Computer/dp/B00U8KSWB6
https://www.amazon.com/VIVO-Adjustable-Counterbalance-Free-Standing-Tabletop/dp/B074P7Y9ZG

https://www.youtube.com/shorts/ZOcFZlIHuYY