Hacker Newsnew | past | comments | ask | show | jobs | submit | more steinvakt2's commentslogin

Remarkable


> I basically want to build a custom e-reader with a RasPi Zero for learning/home use, 8-10inches would be great.


Isn't 5090 FE (roughly 2500 USD in my country) pretty good FLOP value? 32 GB VRAM (and flash attention pushes it even faster compared to apple/mps relatively cheap "vram")


Not really:

5090: 210 TF / $2k == 105 TF/$k

B200: 2250 TF / $40k == 56 TF/$k

Getting only 2x the FLOPs per dollar probably isn't worth the hassle of having to rack 10x as many GPUs, while having no NVLink.


One of the reasons they removed NVLink from consumer cards (they supported it before). There’s also an issue with power consumption (1xB200 vs 10x5090)


Sure, but when spending 20x more, getting almost twice the compute per buck seems expected


I had a 5090 some months ago but couldnt get flash attention to work. Does it now work natively? What about 5080?


Pytorch now has native support for the Blackwell architecture:

https://pytorch.org/blog/pytorch-2-7/


It does, but the performance is pretty bad, worse than Hopper.


Curious what issues you were having. The kernel should compile natively if you pass nvcc the correct arch flags, although it probably won't take advantage of any new hardware features.


High-performance GPU code typically uses nonportable features that are not supported across generations.


Does it fit on a 5080 (16gb)?


Haven't tried myself but it looks like it probably does. The weight files total 13.8 GB which gives you a little left over to hold your context.


It fits on a 5070TI, so should fit on a 5080 as well.


And flash attention doesn't work on 5090 yet, right? So currently 4090 is probably faster, or?


I don't think the 4090 has native 4bit support, which will probably have a significant impact.


> And flash attention doesn't work on 5090 yet, right?

Flash attention works with GPT-OSS + llama.cpp (tested on 1d72c8418) and other Blackwell card (RTX Pro 6000) so I think it should work on 5090 as well, it's the same architecture after all.


I also have this feeling that I'm 2-10x more productive. But isn't it curious how a lot of devs feel this way, but no devs that I know have the experience that any of their colleagues have become 2-10x more productive?


<raises hand> Our automated test folks were chronically behind, struggling to keep up with feature development. I got the two assigned to the team that was the most behind set up with Claude Code. Six weeks later they are fully caught up, expanding coverage, and integrating AI code review into our build pipeline.

It's not 10x, but those guys do seem like they've hit somewhere around 2x improvement overall.


10x means to me that i can finish a month of work in max 2 days and go cloud watching. What does it mean for you?


Sometimes 10x can mean that I start things that I would have never started before, knowing it would take a long time. Or that I can have any of the agentic stuff "explore" libs, stacks and frameworks I wanted to look at, but had no time. Or distill some vague docs and blog posts to find common use cases for tech x. And so on.

It's not always a literal 10x time for taskA w/ AI vs taskA w/o AI...


A 60 minute script becomes 6 minutes


Did you run it the best way possible? im no expert, but I understand it can affect inference time greatly (which format/engine is used)


I ran it via Ollama, which I assume uses the best way. Screenshot in my post here: https://bsky.app/profile/pamelafox.bsky.social/post/3lvobol3...

I'm still wondering why my MPU usage was so low.. maybe Ollama isn't optimized for running it yet?


Might need to wait on MLX


Wondering about the same for my M4 max 128 gb


It should fly on your machine


Yeah, was super quick and easy to set up using Ollama. I had to kill some processes first to avoid memory swap though (even with 128gb memory). So a slightly more quantized version is maybe ideal, for me at least.

Edit: I'm talking about the 120B model of course


https://munin.watch

It offers searching in transcriptions of all Norwegian podcasts in (roughly) real-time. Also offers subscribing to alerts, for example if anyone mentions your business name. Launched some days ago


great work


Thanks!


As a Norwegian, it is absolutely baffling how someone can seriously utter the words "I am in full support of Israel's actions in Gaza". Is the media coverage that different in USA than in Europe? How is it possible?


I try very hard to find credible primary and secondary sources of what's true on the ground. For Gaza, but also for Ukraine, Myanmar, Sudan, I am basically a nerd about conflict zones. From all of my research, I believe the IDF gets a bad rap but is no worse in Gaza than any Western army would be in the same situation. And indeed we can see that in dug in cities like the coalition forces vs ISIS in Mosul, civilian casulaties happened at quite similar rates to Gaza.


The United States occupation of Iraq was like a trip to an amusement park compared to Gaza. Please don't compare the IDF to the US military. At least the US military and its soldiers had RoE that they respected. Any breaking of those rules did not get ignored like they do in the IDF.


People who actually study this stuff say otherwise https://macdonaldlaurier.ca/im-a-war-scholar-there-is-no-gen...


That's a right-wing propaganda outlet, not an institution of studying


Here is the same guy on NPR. https://www.npr.org/2025/07/29/nx-s1-5478643/war-scholar-dis...

He is also the Chair of Urban Warfare Studies at West Point, he is a real subject matter expert. I could also point you to plenty of similar sources.


Is it possible that "any Western army" is also pure evil?


Ulysses Grant, hero of the Civil war for the Union, said, "War is cruelty and cannot be refined." But evil is up to you.


As someone from the USA what I see from media coverage is genocide in Gaza. They would have to be truly ignorant and intentionally uninformed to say something like that or be in favor of genocide.


did you ever consider that media coverage is very selective and biased ?

i remember half an year ago, during ceasefire when gaza was swamped in aid and food was rotting on the streets, laura coates at prime time said that "hundreds of gazans die from starvation daily". never happened. not even dozens.

or when at abby phillips show yesterday, somebody tried to say that presenting images of children with genetic diseases as image of children who are dying from starvation is manipulative, abby phillips stopped this person and said that it's not important.

nyt for example quietly updated it's story on this topic https://www.jpost.com/israel-news/article-862667

you must have seen numerous mentions that 500 trucks of food to gaza is minimum (and actually needed even more), because it's the number of trucks that were entering gaza before war ? but did you see even once mention that 500 is total number of trucks that included construction materials, animal feed, etc.. etc.. and maximum of trucks of food that entered gaza in 1 day before war was 72(82?)

you don't get journalism in mainstream media coverage anymore. you get activism https://www.thefp.com/p/friedman-when-we-started-to-lie


[flagged]


Struggle harder to explain how 10's of thousands of civilian deaths out of a population of 2 million and which has been increasing during the "genocide" can possibly be a "genocide".


Well, that's also something they said. We may or may not find the final numbers after the war. Remember the official death count is frozen since March 2024, since they killed the people who were counting the deaths.


[flagged]


I really have no idea what you are talking about. There's a genocide. The people who are doing the genocide told us they were doing a genocide, so there's no need to read the tea leaves. Genocides are bad.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: