Hacker News: rft

New comment by rft in "216M Spy TVs – The LG Smart TV Problem [video]"

rft — Mon, 07 Sep 2026 14:17:19 +0000

I have a "Homatics Box R 4K Plus" with CoreELEC/Kodi installed. The UX is then determined by your Kodi skin or apps. I wouldn't exactly recommend that box for ease of installation, but it works. It only needed to boot once into Android, annoyingly requiring an internet connection for setup. Since then it only runs CoreELEC. Works via HDMI CEC via the TV remote, which is still a bit annoying, because they had space for a dozen branded streaming provider buttons, but pause/forward etc. are on a mode switch button that overloads the navigation buttons.

For far too many options, see this massive sheet for Kodi support, including newer formats like Dolby Vision and HDR variants: https://docs.google.com/spreadsheets/d/1Uyuw_1dADsRw7u1LGW6q...

Overall I am pretty happy with Kodi on my Android box. 90% of the content is streamed from a local Jellyfin server, so I can't comment on how well it works with DRM media. It occasionally [1] seems to hang, requiring a power reset. The next time it does, I will hook it up to a smart plug to simplify that as well. If you are into tinkering, CoreELEC is based on Linux and you have some basic tools already installed with more available from the addon repos. I have SSHed into LibreELEC to debug network issues with iperf and on a previous iteration to kick over Kodi with systemctl restart.

[1] gonna say less than once a month, I don't remember when it last happened

New comment by rft in "Ask HN: What is your favorite lightweight tool or CLI utility in 2026?"

rft — Sun, 16 Aug 2026 03:10:28 +0000

I intentionally avoid this, especially if I am not the only one connecting to a server. I see potentially breaking others' expectations for my convenience as a clear downside. Plus there is a level of paranoia there, I might have unintentionally included a credential in some config that gets copied over. That potentially increases the blast radius of a server compromise beyond what is strictly required for that specific server. I might copy over some scripts into a directory not in $PATH, easy enough to add it to my shell session if needed, CTRL+R helps as well.

New comment by rft in "Ask HN: What is your favorite lightweight tool or CLI utility in 2026?"

rft — Sun, 16 Aug 2026 02:56:33 +0000

ssh, scp, tmux, (grml-)zsh, grep, less, file, tar, git

The time saving of a well filled ~/.ssh/config is impressive, especially once you start juggling ProxyJump hops.

New comment by rft in "Exploiting System Management Mode with a very long interrupt"

rft — Tue, 11 Aug 2026 00:34:21 +0000

Chris just has some fun projects. Sandsifter was pretty well known back then due to finding that VIA x86 opcode, the movfuscator is just an amazing piece of mostly useless engineering and the fun reverse engineering psychological warfare was my first contact with his projects. I guess someone stumbled on one of his projects and others clicked through to his other projects. Many of them are a perfect fit for HN, no wonder they got posted.

New comment by rft in "Mark Zuckerberg attacks 'closed' AI rivals as Meta returns to open models"

rft — Mon, 10 Aug 2026 15:19:58 +0000

I get that argument, but I don't think it is that bad. Open weights that are out of reach for local use are still within reach for groups of people or small to medium companies. Even if you only rent enough cloud GPUs to run the weights for a few hours, you still get to do anything you want with it.

Worst case you can still archive the weights and hope for capable hardware to get affordable. At the same time this archive is the base line for a new frontier lab to start over if required. I see open weights as a pure upside even though I can't run the bigger models in my home lab.

New comment by rft in "DMARC has been public since 2012 but most company domains still don't enforce it"

rft — Tue, 28 Jul 2026 13:35:51 +0000

Article is missing a note on the existence of MX records for the domains. Sure, you can easily have a send-only domain without an MX record, but the common case is likely to setup both send and receive capability. It would be interesting to have that number included as domains without MX and DMARC might just not be configured for email at all. Worst case the 45% of domains without DMARC are simply not relevant for email and thus not configured at all. I would find "x% of domains with configured email don't enforce DMARC" more interesting.

New comment by rft in "Gemma 4 12B: A unified, encoder-free multimodal model"

rft — Thu, 04 Jun 2026 21:37:14 +0000

Those numbers are wrong for most use cases. Likely the LLM did not take SWA (Sliding Window Attention) of G4 into account. Without SWA those numbers could be correct, I can't load a q8 without SWA on a 24GB card.

I tested this with the at the time newest llama.cpp master on a Linux system with 2 3090 24GB, only one was used for testing. q8 without any KV quant, 256k context, mmproj loaded takes less than 20GB VRAM. This runs at about 1.5k to 2k tok/s pp and 40-50 tok/s gen (slightly lowered power limits & undervolted). q8 with 64k non-quant context and mmproj takes just under 16GB VRAM. Drop down to the q6k model, no mmproj, 64k non-quant context and it fits in 12GB VRAM. All the way down to q4km and some batch size tweaking and it barely fits into 8GB VRAM.

64k context is the minimum for Hermes agent, so a vision capable "agentic" model fits into a 16GB card. This is very impressive. I am currently testing how smart the model is and it does decently so far, had one looping issue it recovered after a lot of tokens, did some basic tool calling.

New comment by rft in "I put a datacenter GPU in my gaming PC"

rft — Sun, 31 May 2026 20:34:36 +0000

Ah, ok. I took a look at the 3090 numbers and they list 400 tok/s prefill, so if I normalize my expectations to that base line the numbers you posted do make sense. I haven't dug deep into that site's methodology, but their estimates seems way off. Especially since they don't take into account cache quant when deciding whether or not you can run a model. Overall I found that website a bit confusing, but maybe the UX just didn't click with me.

New comment by rft in "I put a datacenter GPU in my gaming PC"

rft — Sun, 31 May 2026 19:59:36 +0000

> The MI100 is roughly double the performance on Qwen 3.5 35B A3B Q5_K_M to the R9700 (462 token/s prefill vs 239 tokens/s, 217 tokens/s vs 118 token/s for inference)

Those prefill numbers look really low to me. I can run nearly that same model (qwen 3.6) at q4km with q6 cache on a single 3090 and get 2.3k-4.4k prefill and 100-170 generation. Just based on raw numbers I would expect the R9700 to land around 70-90 generation (about 2/3 of memory bandwidth of a 3090) and at least the same or higher prefill (nearly 3x FP16 TOPS on the R9700). That means the numbers really don't add up. Is the benchmark done with some special settings, e.g. parallel requests or with very low prompt length?

New comment by rft in "Anthropic acquires Stainless"

rft — Tue, 19 May 2026 00:00:32 +0000

For those who, like me, had to do a double take on that number: https://www.businessinsider.com/openclaw-peter-steinberger-a...

Yes, $1.3M in token cost in less than 30 days and some days were even off-peak, if you can call it that with that insane scale that likely hides quite a lot of tokens in the lower bars.

HN thread: https://news.ycombinator.com/item?id=48159227

New comment by rft in "Poland is now among the 20 largest economies"

rft — Fri, 08 May 2026 13:25:32 +0000

Without having firm data, I can see a few factors that are different. After the collapse of the GDR, it was easier for eastern Germans to move to west Germany than for Polish to move to a different country in the west. Mostly younger and educated people would have made that move, hampering future generations. With the Reunification also came the whole Treuhand issue which essentially sold off a good chunk of eastern Germany for pennies to western investors, because eastern investors had no capital. That meant the east lost out on the profits from its economy as they would accumulate in the west instead. Even today a large part of east German rentals are owned by western landlords or corporations. Then the industrial base of west Germany was setup far more for competing on the open world market with automotive companies in the NW (VW), SW (Daimler) and SE (BMW) plus the big industrial area Ruhrgebiet. So you naturally got an economic focus even after Reunification on the old BRD with the previous GDR requiring decades to hopefully catch up to the rest of the new country.

New comment by rft in "Edit store price tags using Flipper Zero"

rft — Tue, 21 Apr 2026 11:17:57 +0000

And a minority of those kids will get curious about the How and Why. Those are the security nerds of the future securing the networks against both the kids they were themselves and actual malicious actors.

Source: Early interest in wifi security, including in other people's networks, lead me down an education and career in security

New comment by rft in "Bet on German Train Delays"

rft — Wed, 04 Mar 2026 15:04:08 +0000

In 2019 there was a talk about data mining the DB arrival data [1] (yes, this problem is nothing new). One of the takeaways was that on some connections you can actually buy a "Sparticket" (cheaper, but only valid for a specific train), but get it upgraded to a "Flexticket" (more expensive, can take any train on the route) for free. This works because a delay of more than X minutes removes the specific train requirement and some routes are nearly always delayed by at least that threshold.

[1] https://media.ccc.de/v/36c3-10652-bahnmining_-_punktlichkeit... (German)

New comment by rft in "Go Home, Windows EXE, You're Drunk"

rft — Fri, 16 Jan 2026 13:06:02 +0000

Related, there is a stable way to detect whether your .exe is running under Wine and even which version by looking up the address of the export wine_get_version in ntdll [1]. Years ago I actually had to do this to work around some weird path bug when we were testing our Windows build under Wine (easier to setup Wine than a full Windows CI).

[1] https://www.winehq.org/pipermail/wine-devel/2008-September/0...

New comment by rft in "39th Chaos Communication Congress Videos"

rft — Fri, 02 Jan 2026 20:11:35 +0000

I still have to go through my watch list, the age old issue of not having my slides done before congress...

The 10 year of Dieselgate is interesting just from a "how bad is it really?" PoV, I saw the part about curves and other defeat devices already [1].

The Rowhammer talk is likely going to be great as well, I like Daniel's work [2].

The practical Cross-VM Spectre was interesting to show this is still a problem [3].

The opensource secure element was good for trying such a thing, but I wasn't that impressed with the content [4].

[1] https://cfp.cccv.de/39c3/talk/7MSRA7/ https://media.ccc.de/v/39c3-10-years-of-dieselgate

[2] https://cfp.cccv.de/39c3/talk/3JXAJJ/ https://media.ccc.de/v/39c3-rowhammer-in-the-wild-large-scal...

[3] https://cfp.cccv.de/39c3/talk/ATYLN9/ https://media.ccc.de/v/39c3-spectre-in-the-real-world-leakin...

[4] https://cfp.cccv.de/39c3/talk/9DYZXG/ https://media.ccc.de/v/39c3-lessons-from-building-an-open-ar...

New comment by rft in "The Angry Path to Zen: AMD Zen Microcode Tools and Insights [video]"

rft — Fri, 02 Jan 2026 09:18:37 +0000

Author here, happy to answer any questions you might have.

New comment by rft in "WinApps: Run Windows apps as if they were a part of the native Linux OS"

rft — Sat, 29 Nov 2025 11:26:35 +0000

SteamVR works ok, but last I checked it still performs worse than on Windows. If you are feeling adventurous, you can try a FOSS VR stack [1]. It works for Steam games running Proton and when it works it provides better performance. I had some troubles with it, sometimes you need to switch versions or you get some artifacts in games, or some games just don't work at all. Good thing is, switching between FOSS and SteamVR is as simple as launching either first before starting the VR game in Steam.

I guess the Linux VR stack might get a bit of love from Valve for the Steam Frame, so things might improve in the near future.

[1] https://lvra.gitlab.io/docs/fossvr/envision/

New comment by rft in "A monopoly ISP refuses to fix upstream infrastructure"

rft — Sun, 23 Nov 2025 09:36:35 +0000

At this point escalating, or threatening to, might be the better option. But I can't help trying to figure out how to solve a people/organizational problem with a technological solution.

Github is still famously IPv4 only. I don't know if there is a split between the SSH (if you use SSH to access the repos) and HTTPS (the tarballs) setup on their end, so maybe you get full speed on IPv6 and limited on IPv4 (or the other way around). Try disabling IPv6 on your end, if the speeds match then this might be it. If IPv6 is fast using an IPv4 gateway that tunnels via an IPv6 VPN might be a workaround.

I also had a similar problem a while back. Some speedtests showed more bandwidth than I could get in regular HTTPS downloads. I could get multiple downloads running at the same time that in total added up to the expected speed. In my case the line was just lossy enough (TCP retransmits in Wireshark) for TCP to never scale up its window size properly beyond a certain limit per connection. I verified this by running iperf in TCP and UDP against a gigabit server, UDP reached near full speed because it didn't care about a few lost packages. Working around that issue might be a bit harder, maybe [1] via [2] can provide some ideas to look into.

[1] https://github.com/apernet/tcp-brutal

[2] https://news.ycombinator.com/item?id=38164574

New comment by rft in "Germany: States Pass Porn Filters for Operating Systems"

rft — Fri, 21 Nov 2025 12:18:08 +0000

As much as I hate where this might lead, I can't help but amuse myself over a scene like this: You walk up to a bearded fellow in a dark alley and quietly whisper "Stallman..." only for him to quietly respond "... was right.". He then opens his coat for you to choose the install USB of your favorite distro (Arch of course!). He dutifully hands you the stick and a printed copy of the GPLv5 for you to hide in your coat as you walk past the telescreens back to your home.

New comment by rft in "Germany: States Pass Porn Filters for Operating Systems"

rft — Fri, 21 Nov 2025 11:58:17 +0000

That part also caused my tin foil hat to heat up. At least they get the credit of including it directly instead of adding it in a later revision that gets even less news coverage. It is hard not to grow cynical when you see this.

I am also worried about another detail:

> The states also want to prevent the circumvention of blocking orders by erotic portals ... using so-called mirror domains – i.e., the distribution of identical content under a minimally changed web address. For a page to be treated as a mirror page and quickly blocked without a new procedure, it must essentially have the same content as the already blocked original.

Note the part "quickly blocked without a new procedure" so there is a way to block sites with even less process and oversight. That just invites overblocking without accountability.