2,943 karma · joined October 15, 2007
Imo they don't charge all that much relative to their value, depending on who you're asking.
Having a non-user replaceable battery is a really easy way to ensure a product stops working after 3-4 years though.
And the criticism is typically directed at companies like Apple, who does make things that last physically, but then force you to upgrade by way of battery.
And this was a mistake, one that they slowly tried to reverse once they realized Ukraine wasn't going to be a pushover especially with modern weaponry.
It _is_ possible with the combo of weapons and economic sanctions. Ukraine (had) more economic support, so it is both a war of blood and endurance, endurance which Ukraine could have by being a porcupine with next-gen weapons and non-stop bleeding for Russia until things start hitting the oligarch's pocketbooks too far and the people of Russia begin to demand regime change.
Russia is a glorified gas station. You can hurt it dramatically with economic sanctions because its whole GDP is concentrated into one sector.
Imo if Trump really cared about America's long term security, he would be trying to fund research into cheaper / easier to produce next-gen weapons for Ukraine to test for us, instead of complaining how much our HIMARS and tanks currently cost, firing top military brass, cozying up to a dictator, and trying to exploit a ravaged democratic ally.
Would if I could figure out how to uninvite my otterpilot ^^'
Also, if you weren't going to use Google's offerings you probably wouldn't be down to buy their HW either :)
FWIW your PTSD with Tensorflow is shared by everyone at Google, it was just unavoidable because Google needed ML to be performant and they started by sacrificing devex for performance whereas Pytorch went ergo-first, performance later. In retrospect a prescient move by Meta, but now making the high performance stuff ergonomic is proving out for Google via XLA and JAX.
Really? Nobody uses TF but plenty of people use XLA, which is what you allude to saying that TPU in Pytorch is possible. That's arguably the most important piece; the sledgehammer that makes good performance and good devex compatible.
> I wish Google would compete with Nvidia directly and let me buy a TPU
I don't see the utility of buying _any_ GPUs. It's mainly a cost + availability optimization to own them yourself if you're doing a lot of continual training. Outside of the foundation model companies, most of us just use cloud services -- and I want those to be cheap and always use the latest thing.
Google has a highly optimized infrastructure to make sure unused resources (CPU, TPU, RAM, disk) are properly allocated. This means preemptible instances, highly co-located compute + data, etc. In theory everyone should win when you use ML hw through hyperscalers, because they collect on their structural advantages and your TCO is lower.
> Google could have Nvidia’s business and OpenAI’s
They arguably have a better _business_ on their hands but a worse _speculative outlook_ in the eyes of Mr. Market. Those are different things. There's an argument to be made that Google's valuation could be as high if they ran their public relations strategy as well as OpenAI / Nvidia / Tesla, which are all riding monumental hype.
You can easily train and serve models on TPU through Google's cloud services. Colab for example makes it easy to train models on TPU.
> TPUs are not faster than H100s
There are several generations of TPU with different characteristics, some oriented toward serving, some toward training, and all toward minimizing TCO. You might give up FLOPs but have much faster memory bandwidth, or be overall similar but have a much more efficient data pipeline thanks to Google's scaled infrastructure, netting out ahead. The OP isn't a wanker for pointing out that long-term infra investment is paying off for Google.
Why do you want the hardware vs just using it in the cloud? If you're training huge models you probably don't also keep all your data on prem, but on GCS or S3 right? It'd be more efficient to use training resources close to your data. I guess inference on huge models? Still isn't just using a hosted API simpler / what everyone is doing now?
L2 driving record vs L4 driving record are insanely different. You can't fall asleep in an L2 car without risking death. You can fall asleep in an L4 car.
> I have my opinion, it’s just not the same as yours.
This is literally the TL;DR for what I wrote.
Exactly what I illustrated earlier: your developer productivity metrics. If you're turning code around faster, setting up your network better, turning around insights faster, the AI is working.
> It depends. For a sysadmin maybe not, but for data scientists, the bar would be pretty high just to understand the math jargon.
Why does an AI coding agent need to understand math jargon -- it just helps you write better code. Are you even familiar with what data scientists do? Seems not because if you were, you'd see clearly where the tool would be applied and do a good/bad job.
Reminder: we're talking about evaluating whether Codebuff / alternatives are "pretty good" at X. Just go play with the tools. tgtweak expressed their opinion on how good the tool rates at some tasks {sysadmin, data engineering, cloud architecture} and your response was to question how someone could have an opinion about it. The obvious answer is that they used the tools and found it useful for those tasks. It may only be _subjectively_ good at what they're using for but it's also a rando's opinion on the internet. As another rando I very much agree with what the person you responded to is saying. You're not going to get more rigor from this discourse - go form a real opinion of your own.
Admittedly the last time I used manicode was a while back but I even preferred Cursor to it, and Cursor hallucinates like a mf'er. What I liked about cursor is that I can just tell composer what files I want it to look at in the UI. But I just use Cline now because I find its performance to be the best.
Other datapoints: backend / ML engineer. Maybe other kinds of engineers have different experiences.
Sure but it's not high at all.
Your typical sysadmin is doing a lot of Googling. If perplexity can tell you exactly what to do 90% of the time without error, that's a pretty good sysadmin.
Your typical programmer is doing a lot of googling and write-eval loops. If you are doing many flawless write-eval loops with the help of cline, cline is a pretty good programmer.
A lot of things AI is helping with also have good, easy to observe / generate, real-time metrics you can use to judge excellence.
> It’s a weird catch-22 giving praise like that to LLMs.
It's a bit asymmetrical though isn't it -- judging quality is in fact much easier than producing it.
> you might be able to intuit and fill in the gaps left my the LLM and not even know it
Just because you are able to fill gaps with it doesn't mean it's not good. With all of these tools you basically have to fill gaps. There are still differences between Cline vs Cursor vs Aider vs Codebuff.
Personally I've found Cline to be the best to date, followed by Cursor.
Also, if you do embedded, there's a lack of something akin to nanopb for capnproto.
Pedantic to say that Columbus subjugating entire people groups with mass murder doesn't qualify here. The manpower skew was huge, so how exactly did they subjugate the Tainos without genocide? 7M tainos died when all was said and done.
They also weren't all the same. Not everyone goes around enslaving people and using their life to seek riches and fame. I don't think anyone complains about historical figures who, flaws and all, fought for higher ideals like personal liberty.
Typical HN comment writing off significant thoughtfulness as "not an innovation" lol