Besides, Mick Herron believes Nigel will soon die in an accident and he has been right about Farage matters before.

Binface on the other hand would for obvious reasons survive being hit on the head by 15L of paint. So he might get my vote for being the more resilient candidate.

Or how about slack, that lives in an alternate reality where outlook calendar can be ignored?

Does ananyone outside deepseek have a working code for the v4 compressed attention mechanism?

Has any other provider managed to bypass CUDA and program the compute engines in their native assembly language to get 10% more performance out of them?

There is your answer.

1. Show the local government that rich people are not paying enough tax.

2. Start alobbying group against the upcoming tax raise

3. Accept donations from rich people.

Anyway, does anyone know what makes a submarine "high tech" ?

They have published tons of articles dedicated to performance and efficiency engineering. Feel free to have a look...

Someone is working *really hard* to make this political. And a lot of people on both sides are buying it. Resulting in long essays like this one...

In the mean time (a) dual AC system for heat/cold is pretty standard in many European countries and (b) nobody is stopping you from doing anything.

If you can get hold of one, go ahead and install an AC. Install five for what I care. The inverter system is very efficient and you will probably not have the cold on for more than a few weeks per year anyway. And that during a time where Europe usually produces more power from solar panels than the infrastructure can handle..

Ignore whoever told you it's bad for the environment. Also ignore anyone who told you the green conspiracy is killing the elderly by the millions.

Now go and have a great summer.

Learned a few thing about Rust performance tuning too :)

They have always been able to do this, but this time they did have the option to pass.

/s

(Seriously, look at these comments. What is happening to HN? Is this what previously productive people now do with their free time while their agents are churning?)

The first Spark sw update improved performance significantly. Maybe AMD software team can get their act together and do the same? :)

Anthropic claims deepseek has made 150K requests to their servers. Even if this number is correct, it takes far more requests to distill from a 3.2T model into a 1.6T model. 150K is closer to running a few benchmarks.

If anything, deepseek together with googles deepmind are the ones innovating while Anthropic and openAI are spending money and time on politics to try to hinder or ban competition.

That is a tiny tiny system. OpenAI uses _milions_ of GPUs for training

On the other hand, this probably reuses the existing deepseek v4 architecture and weights. Maybe didn't need that much compute.

But at the same time, you should not be expected to give up your culture (as long as it is not doing something against the law).

I am reading up on GPU / ML micro architecture and am looking for some good sources.

AOG models are fine, but who knows if my company will be approved by WH to use their next model.

Besides, from a technical point of view the new open models seem to be where most leaps are being made so even if you don't use them it is good to keep an eye on them.

https://news.ycombinator.com/item?id=48704714

Source

https://x.com/AllardKlipp/status/2060109271635771457

Sure ads and AI are horrible but they are root of like 5% of the Windows problems.

A good example of a real windows problem is the garbage filesystem performance