HNHacker News
TopNewBestAskShowJobs

elorant

35,994 karma · joined January 18, 2011

.Net developer

cbarak@gmail.com

submissionscomments
elorant··on Reinvent the Wheel
In my almost 15 years in this community this is the first comment where I don’t have a fucking clue of what is described.
elorant··on What do wealthy people buy, that ordinary people know nothing about? (2015)
Wealthy people usually get concierge services. You go to a museum as a regular folk, it’s overcrowded and there’s like 20 other people looking at the same art piece you’re looking, trying to get a better shot from it. You visit as a wealthy individual, you get a private tour after hours with a dozen others when the museum is empty with a trained guide who can answer any questions you may have. You can view any art piece at your own pace from every angle you like. This goes on for pretty much every public service out there.
elorant··on Vacheron Constantin breaks the world record for most complicated wristwatch
Good luck finding a compatible battery for it after 200 years.
elorant··on The Llama 4 herd
It’s an x86 PC with unified RAM based on AMD’s new AI cpus. Pretty unique offering. Similar to Mac studio but you can run Linux or Windows on it, and it’s cheaper too.
elorant··on Dow plunges 2,200 points, Nasdaq enters bear market
Who said anything about sweatshops? You pay the minimum wage expected by law. Apple has $180bn in profits. They could surely afford to pay US workers and lose some of that profit. But everything has been dominated by greed these days.
elorant··on Dow plunges 2,200 points, Nasdaq enters bear market
In theory companies like Apple could bring manufacturing back to US.
elorant··on I genuinely don't understand why some people are still bullish about LLMs
I like LLMs for what they are. Classifiers. I don’t trust them as search engines because of hallucinations. I use them to get a bearing on a subject but then I’ll turn to Google to do the real research.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
So eventually you'll nuke them. But you don't think that will start WW3. Okey dokey.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
And how would you safeguard the next regime? With F35s and bombers?
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
If you don't get boots on the ground you can't change the regime. Bombings alone would only weaken the infrastructure and probably lead local people to hate you even more. Plus in a conflict the opposite side retaliates. Once body bags start pilling back home the general public will demand an invasion.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
Because Syria is fragile, and so is Iraq, Afghanistan, Libya, Yemen. Those are all countries where Iran has very strong influence. The world isn't the same as it was 20 years ago, and Iran is way more powerful than what Iraq was back then.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
A ground invasion is an inevitability if you start bombing strategic installations.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
Sure, here's one. Iran hits US military bases in UAE and Bahrain. Houthi rebels launch long-range missile attacks at Saudi Arabia's and UAE's oil installations. Hezbollah attacks Israel. At the strait of Hormuz Iran directly targets US and European oil tankers. US is left with no other choice than a large scale military mobilization in the area. You can take it from there.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
Any type of action has a reaction. Iran can target US military bases all around Middle East. They can damage tankers passing the Persian Gulf causing havoc to worldwide markets.
elorant··on B-2 Spirit Stealth Bombers Begin Unannounced Deployment to Diego Garcia
Great. Let's end the war in Ukraine and start WW3 with Iran.
elorant··on An AI bubble threatens Silicon Valley, and all of us
One of the reasons I'd like this bubble to burst is to see Nvidia's stock collapse. I'm sick and tired of their exuberant prices.
elorant··on Qwen2.5-VL-32B: Smarter and Lighter
True. The only problem with lower quantization though is that the model fails to understand long prompts.
elorant··on Qwen2.5-VL-32B: Smarter and Lighter
The average user won't self-host a model.
elorant··on Qwen2.5-VL-32B: Smarter and Lighter
You don't need 16-bit quantization. The difference in accuracy from 8-bit in most models is less than 5%.
elorant··on Why Tech Bros Overestimate AI's Creative Abilities
Overestimated or oversold?
elorant··on The race is on to build the most complex machine
EUV lithography machines are borderline science fiction. I mean if you look at all the steps included it's mind boggling. Shooting tiny droplets of tin with two lasers at a speed of 50k times/second, harvesting uv light through atomic precision mirrors, moving wafers with nanometer precision and all that in perfect vacuum. You watch the video demos and seems like you're watching a sf movie.
elorant··on When the Dotcom Bubble Burst
That one hit me hard. I had just quit my job from a prestigious company the previous year to pursue my dream with two friends of building an internet related company and just a few months after we launched the market tanked. And it tanked so hard that we practically went out of business a year later. The problem was that the web back then was so fragile as a business case that once the bubble burst a lot of companies lost interest in investing in it.
elorant··on AMD's Strix Halo under the hood
But why did they choose to build this as a mobile cpu though? I don’t need 128GB of unified RAM on my laptop. It’s the desktop where things happen.
elorant··on What's next after the AI bubble bursts?
The new Mac Studio is great for this job too. With the Max chip you get 128GBs of unified RAM for $3.5k. Memory bandwidth is at 546GB/s which is good enough for most tasks.
elorant··on What's next after the AI bubble bursts?
I do too, but I still believe there is a bubble. Reason is that I can do most of my work with self-hosted models so there’s little need to pay for a subscription. Sure, I’m a minority for now because I’m tech savvy, but with that in mind it’s hard to justify the valuations of most AI companies. We don’t know if they’re keep accumulating users and most LLMs seem to have hit a plateau of performance.
elorant··on Nvidia's RTX 5090 power connectors are melting
All these incidents are from aftermarket cables.
elorant··on Building a personal, private AI computer on a budget
Runs is an overstatement though. With 4 tokens/second you can't use it on production.
elorant··on Mistral Small 3
I’m using Llama-3 8B to classify html files. It’s surprisingly good, and I run it on an RTX 4060 Ti at 8-bit quantization. No complains so far.
elorant··on DeepSeek could represent Nvidia CEO Jensen Huang's worst nightmare
What happens with excess hardware that’s been already bought? It could take years for all of that to be absorbed by smaller companies. And secondly, if this opens the era of self-training models then why go for a 671B one and not a smaller that’s fine-tuned to a company’s specific needs and can then run on consumer GPUs. At 70B with 8-bit quantization you’re good with just three 3090s.
elorant··on Open-R1: an open reproduction of DeepSeek-R1
You know what was exciting back then? Tech magazines. That’s where you discovered all the great stuff and read about the research that was happening.
← PreviousPage 6 of 34Next →