Viewing profile — fancyfredbot

Using AI is vanity? I've never heard that before. Why is that? Refusing to use AI honestly seems more like vanity to me. Like "no machine could ever do what I do". That said I woul…

The link is to a really poor summary of a better article by the register. https://www.theregister.com/ai-and-ml/2026/08/03/as-larry-el... The register article is about this post: h…

I found this touching. As someone who has tried to read a new kind of science, but struggled with the arrogant style, the love and respect this shows he had for his wife came as a …

Since we're (as a society) spending tens of billions of dollars to serve models like this, it's a pretty interesting advertisement.

If I'm squeezing enough margin out of my customers that I can afford to sponsor a football team with some of the leftover cash then you probably don't want to deal with me.

It's the GNU/OSS AI bros who are upset. The other AI bros are using LLVM/Clang I expect. GCC is much less relevant than LLVM in the AI world. NVIDIA's entire CUDA compiler stack is…

It's a pun on "efficient markets hypothesis". The efficient markets hypothesis says the market prices incorporate all available information and there's no such thing as a cheap sto…

I don't entirely understand why they "didn't want to touch it"? Surely they were capable of soldering a new port. I'm sure they'd have had to charge a lot. There's unlikely to be m…

Correct

ancient intel did 80 bit fp until your intermediate result was written from registers to memory. Since you typically have no control over when that happens, even in very low level …

I think the term in this case is marketing. NVIDIA used to use the term supercomputer more often five or six years ago. It's now fully committed to the term AI factory. I think thi…

It's mind blowing seeing these multi exaflop single rack systems. The world's first exaflop supercomputer was Frontier. It was launched only 4 years ago in 2022. It's not a fair co…

Pensions will have exposure, probably through private credit, and probably in the range of 5% to 10% of assets. It's a lot, and it'll hurt people. But I'm not at all sure it is eno…

Yes they are going to ramp up capacity. It would only make business sense to do nothing if all your competitors were also doing nothing, which would probably require some level of …

I bet there's a better way than unpacking, running sequentially and repacking. Even if the algorithm is very branchy you save a pack and unpack.

The relatively slow depreciation of GPU value is an artifact of supply constraints. If you run fp4 inference and could choose freely between Hopper and a Rubin, the performance per…

s/banks/private credit/

Things have changed since 2009. It would be private credit which fails this time, not the banks. Private credit is not supposed to be systemically important and it's not supposed t…

A lot of OpenAI/Anthropic compute actually belongs to the hyperscalers. They have long term contracts and will be able to charge some kind of margin for access to the compute they'…

The term vibe coding actually originated from a tweet from Andrej Karpathy and so there is actually a sort of definition of the term. He said vibe coding is where you "forget the c…

I increasingly feel that the reason vibe coding specifically prohibits reading the generated code is that it's impossible to forget the horror that lurks in these python files. The…

Buffoon (not bafoon).

They are taught the difference through reinforcement learning with verifiable rewards. Pretending you've solved the task or making up a story about how you solved it won't do well …

Normally people refer to the compute-bound phase as "prefill". Nothing wrong with saying it's building the kv cache though, it's accurate just unusual.

[dead]