adolph
Joined 17 years ago
Karma
8,432
Submissions
5,734
Member since
Jan 27
Howdy!
View on Hacker NewsRecent submissions
Comment · 4 days ago
The 5090 does have a bandwidth advantage, but also must pass more data from main to GPU memory for any particular task. Comparisons of a 5090 to a DGX Spark at 273 GB/s are illustrative since the Spark uses unified memory similar to Apple's M-series. [0] The 5090 certainly wins most benchmarks but that doesn't mean the DGX Spark is not useful.
0. https://float16.cloud/en-en/ai-benchmark/dgx-spark-vs-rtx-50...
Comment · 4 days ago
Given that a group of people is not typically of a single mind about the vast array of things for which a position might be expressed [0], the continued presence of Wyden reflects an ambivalence within the group and/or a controlled opposition to cite when the music stops.
Comment · 4 days ago
One of the interesting slights of hand Apple marketing must do is simultaneously say the new processor is awesome while selling the higher spec versions of the previous processor as upgrades:
The M6 chip joins M5 Pro and M5 Max to create our most advanced line-up of chips ever built for a pro laptop. With a 12-core CPU, 12-core GPU and our first ever Dual 16‑core Neural Engine, M6 brings next-generation speed and advanced on device AI. All three chips deliver phenomenal single- and multithreaded CPU performance and faster unified memory. . . .Comment · 5 days ago
Interesting. Since this was implemented on the S3, which uses the ISA XTensa LX instead of the RISC-V of most newer ESP32 series chips, I looked into how Espressif is handling SIMD on the latter ISA. Apparently it is only present in limited fashion as a processor extension on the P4 and S31 [0]. The S31 datasheet claims SIMD [1] and the P4 datasheet [2] files them as "Processor Instruction Extensions."
0. https://bitbanksoftware.blogspot.com/2026/04/esp32-p4-simd-e...
1. https://documentation.espressif.com/esp32-s31_datasheet_en.p...
2. https://documentation.espressif.com/esp32-p4_technical_refer...
Comment · 5 days ago
People waiting for >v1 Duo? Same processor, 1" larger screen, lower cost but you still carry a phone.
Comment · 8 days ago
Question: do you advocate for police to maintain dossiers on people's activities in order to arrest them once some combination of those activities cross the threshold? Like KYC for shoplifters? Do you think these records should be jurisdictionally/geographically limited?
Comment · 9 days ago
The interesting thing about thresholds is that antisocial behavior models itself to the threshold, whether by conscious awareness or by an evolutionary pattern.
The pair is accused of stealing around $3,200 worth of merchandise, primarily meat, seafood, and produce, from one or more Safeway locations across 14 separate incidents.
Comment · 9 days ago
Do people just deploy from "Agents" to prod (i.e. trivial in seconds)?
Comment · 9 days ago
It also has a link to Austin, TX. The American company involved in the Grasberg mine is Freeport-McMoRan. In the 1980s-90s it also attempted to develop housing along the Barton Creek watershed. This lead to the ordinance and political group called "Save Our Springs." At the time many people tied the environmental degradation of Papua New Guinea with what the company planned for Austin.
https://www.austinchronicle.com/news/the-battle-for-barton-s...
Comment · 11 days ago
> That's Kami Kaze for US frontier labs. There's no way from here for them towards trillion dollar valuations.
I don't understand the reference to Kamikaze in this context. Is Deepseek destroying itself in order to deep-six US frontier labs?
The term is used generically in modern warfare for an attacking vehicle which is itself destroyed when attacking a target; for example, a kamikaze drone. [0]
Comment · 11 days ago
The technical addendum [0] indicates they refer to "model" as the entire system, not just open-source weights, which is the focus of the g-parent comment [1].
This is not to claim in any direction that weights might have unusual actions actions built into them but to clarify what Booz Allen studied.
Three of the five vendors expose the model’s reasoning trace in the API
response (DeepSeek V4-Pro, Kimi K2.5, MiniMax M2.5). Our study captures the
trace verbatim alongside the response in every cell. Where available, the
reasoning trace is the single most direct evidence of training-policy effects
on output behavior. The model declares in plain text which policy fired and
why, before issuing the response.
Comment · 11 days ago
Yes, wish there was more journalism like this. So many interesting queries, like How are the physical and informational domains synchronized? Does scrap signage get a higher or lower price in recycling? Is there software to make a cut list from all possible signs in the backlog in order to minimize waste from each aluminum blank?
The signs they make seem to cross jurisdictional boundaries, like city streets and state administered interstate highways. How is billing or cost-capture peformed?
Comment · 12 days ago
> Our own nerves produce a huge amount of data.
I wonder if collection of training data is going to be a purpose for Neurallink.
Comment · 19 days ago
This is the truth. Aside from solar there are plenty of folks with backup generators that automatically come on whenever grid power fails. When maintenance crews came through my neighborhood to upgrade the old pole fixtures, they grounded out each block to keep any backfeed off what they were working on.
Comment · 20 days ago
> They should for sure list the annual price and the monthly price but it's not "burried".
Buried prices are the next frontier of consumer protection. I've gotten burned by looking up a store product online and then when attempting to purchase finding that unless you ordered it through the site for in-store pickup, you get a higher price. Car dealerships were the same the last time I bought one. And while I appreciate Tesla's challenge in expressing TCO, subtracting average fuel costs from the price of a car smells bad in the same way.