Marcus Müller

Public Key

npub1p3y6c327k0x5dhy9f5qcpvhsusga2vcd3540q3gycux8kkva687q8jh309

NIP-05 Address

funkylab@mastodon-social.mostr.pub

Profile Code

nprofile1qqsqcjdvg40t8n2xmjz56qvqktcwgyw4xvxc62hsg5zvwrrmtxwarlqpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0ds3zqk7a

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1p3y6c327k0x5dhy9f5qcpvhsusga2vcd3540q3gycux8kkva687q8jh309

Show more details

Last Notes

2026-06-09 18:25:47 UTC

Legit cables that look like shitposts: SMA to 3.5 mm TRS
https://files.mastodon.social/media_attachments/files/116/721/552/264/061/934/original/7a89f8c461feb1bb.png

2026-06-09 14:02:33 UTC

- reply

@nprofile…3tdq is this half- or full-duplex?

2026-06-01 15:57:02 UTC

- reply

@nprofile…mv7s yep, need to shift the time values to a "local environment" if you will (extreme example: the interpolator doesn't care about absolute time, you can always shift all times by an integer value so they're zero-centered, locally)

2026-06-01 15:41:45 UTC

- reply

@nprofile…mv7s yeah, never keep time in floating point; it's typically not a thing where that makes a lot of sense (this was learned by SDR software *several* times). Your timestamps should just be int64 multiples of some pre-defined time quantum, and you'll be fine. You don't need sub-attosecond accuracy, and neither do you need to be able to measure in millions of years; the fixed 64 bit range suffices.

2026-05-29 14:36:09 UTC

- reply

@nprofile…m7ah or you might be someone with enough statistics to say which parameters correlate with trouble, both only possible to work with data that actually is given to you.
But saying "Memory and PCIe are not RF links where you waste capacity" is really um, not a great sign of insight here, imho.

2026-05-29 14:35:09 UTC

- reply

@nprofile…m7ah why do you think DDR controllers do link training? That's a noisy RF link with reflections! Of course you need to treat it like that to squeeze out the last bit of performance to get a 3600 MHz clock rate going over stupid FR4 in a bunch of narrowly routed microstrip lines (not even "proper" transmission lines!).
I honestly think you're way deep in denial here how ubiquitous "parameters that nobody but a very narrow group of experts can understand" are.
You might be one of the experts,

2026-05-29 14:30:41 UTC

- reply

@nprofile…m7ah
> Memory, PCIe, etc. are not RF links where you have to fully utilize your channel bandwidth/link budget and waste efficiency by not doing so.
You're, as you noted above, are very wrong about that! How do you *think* PCIe works *at all* at 2.5 GT/s over a transmission line? It's a noisy RF link. PCIe 5.0 with 32 GT/s: That literally uses FEC extensively as you'd find in a wireless RF link. To cite the PCI-SIG: "We're better off retrying a packet with 10⁻⁶ (or 10⁻⁵) prob.…"!

2026-05-29 14:26:21 UTC

- reply

@nprofile…m7ah
> Memory shouldn't have errors, ever. It's an emergency feature, nothing more.
Assuming this is about modern server RAM in well-cooled servers: hm, that's an expensive stance to take. But OK, your time might be worth more than RAM.
Assuming this about any other kind of memory, be it flash mass media, hard drives, optical media or tape: you're nowhere close to guessing how deep ECC needs to go to make any of these technologies usable at all.

2026-05-29 14:24:14 UTC

- reply

@nprofile…m7ah yes, absolutely, and that's fine. Now, your "trained eye", however, knows the parameters to look for. And the vendor's "trained dev" knows which parameters to display at all (which often is also a tradeoff with complexity and power). So, it's "you're not close to the truth here" all the way down, you just, and I get that desire fully, want to get the proven-to-be-useful information that you *might* use for anomaly detection.

2026-05-29 14:19:44 UTC

- reply

@nprofile…m7ah respectfully: God, no!
If your error-corrected link *never* produces an error, you're underutilizing it, or it was badly designed and wastes capacity on pointless robustness. (The point of ECC RAM is that you *can* correct memory issues. Otherwise it'd be "checksum RAM"; the point of PCIe transaction failure modes is that you *can* retransmit.)
Your shiny server SSD? Thousands to millions of bit errors that get corrected in operation. That's the way things work. Links and media are…

2026-05-29 14:16:48 UTC

- reply

@nprofile…m7ah hm, fair. In a professional server environment, where people might be able to pull someone who understands any of the numbers seriously, other than "Oh it was half that with our server last year, what's an Eb/N0 anyways? why is it so high here?", sure. I just try to be aware of the fact that as an EE coming into computers, I lack skills on the computing side of things, and that vice versa, I wouldn't trust someone who knows to run racks of servers to have the slightest about signals.

2026-05-29 14:13:31 UTC

- reply

@nprofile…m7ah and I mean, for PCIe, you at least get the error counts; for ECC RAM controllers the count of reconstructions and retries, for network cards with checksum offload the number of failures, for Wifi cards you often get an actual CSI estimate and so on.

2026-05-29 14:11:22 UTC

- reply

@nprofile…m7ah @nprofile…w7qn but isn't that the honorable job? The interface *has* to just work. Nobody wants to debug why their new USB mouse sometimes behaves badly. The interface not working needs to be that much of an exception to the rule that by all means, it's not something users ever should have to care about, especially since it's something they can't do much about. (But might believe they do. "insufficient timing advance, try cleaning connectors" will bring out the audiophile in way too many)

2026-05-28 15:25:16 UTC

Hey, people into SDR: If you happen to be in the US, specifically the New England region, close to Worcester, Mass., on June 4/5:
NEWSDR , the New England Workshop on Software Defined Radio, is free (registration required, by *tomorrow*!), and organized by great people. Oh, and if you have something you can put into a poster by the time the workshop starts, they're still open for that!
https://newsdr.org/workshops/newsdr-2026/
poster abstract sub: https://docs.google.com/forms/d/e/1FAIpQLScYvl_NpCt2RLHv7UoqGMy1APVh_3MfT5DZgoiknACl0ENliw/viewform
#sdr #workshop
https://files.mastodon.social/media_attachments/files/116/652/894/268/219/214/original/c664ced017d61a82.webp

2026-05-28 14:00:45 UTC

- reply

@nprofile…q6qh wat? Ok, that's nonsense.

2026-05-27 10:06:44 UTC

- reply

@nprofile…s5ng "Mehr Zeit für Zwischenmenschlichkeit – DB & #evaTraindestroyer"

2026-05-13 08:33:51 UTC

- reply

@nprofile…y789 @nprofile…0sfk this makes me a bit excited for what's to come!

2026-05-13 08:30:35 UTC

- reply

@nprofile…y789 @nprofile…0sfk which of the national governments are the driving force behind radical change? It feels like we have some influential govs that don't like to spend on their state rail (Deutsche Bahn), and some that really never wantes to make it easy for people to buy tickets (SNCF), and some that really try to avoid playing nicely with their neighbors. I do get why you're nervous, just surprised improvements in cross-border ticketing are actually happening!

2026-03-27 12:24:29 UTC

- reply

@nprofile…mv7s sounds much like you really ought to instead have a factory that spits out these objects:
I'd frankly have smth like the "actual type" you want to work with, and a wrapper class. Hint: you can use `std::unique_ptr<class object_t, class Deleter>`'s `Deleter` to implement the "remove pointer from registry, then actually call `delete` on the stored pointer" functionality. The registry could be `std::unordered_set<object_t*>`.
Factory is function returning these `unique_ptr`s.

2026-03-27 11:57:52 UTC

- reply

@nprofile…mv7s that's a registry, and be careful of the *static initialization order fiasco* if this is C++

2026-03-24 13:29:53 UTC

- reply

@nprofile…m7ah @nprofile…qhvj nice, that's super helpful!

2026-03-24 13:21:05 UTC

- reply

@nprofile…m7ah @nprofile…qhvj hey, because energy for a year can quickly become more expensive than the hardware: do you happen to have a wall-power measurement for the media converter? (just an order of magnitude, not something detailed. I want to know whether we're talking 4€/a in electricity, or 50€/a.

2026-03-24 10:55:13 UTC

- reply

@nprofile…m7ah how do you do the mechanical part of pulling the cable through walls? I have a flat with power outlets that I hope are connected via contiguous pipes, but there's obviously already a power cable in these. Can that still work?

2026-02-07 07:59:50 UTC

- reply

@nprofile…k3ke @nprofile…29up yes, as explained; sorry getting tired of re-explaining, see other thread

2026-01-30 08:40:22 UTC

- reply

@nprofile…m7ah @nprofile…h46y @nprofile…pzxv well uefi specifies boot via HTTP https://uefi.org/specs/UEFI/2.11/24_Network_Protocols_SNP_PXE_BIS.html#http-boot ; there's an entirely (nah, let's go with homeopathically) reasonable world where the BMC acts as HTTP server, using as much of its own RAM as read cache as possible,to minimize that latency penalty (physical CD-ROM drives also benefit from locality, because seeking across half a disk is probably slower than awaiting replies from residential V.90 modems)

2026-01-30 08:21:34 UTC

- reply

@nprofile…m7ah @nprofile…h46y @nprofile…pzxv yeah I was wondering where images were stored, but assumed there must just be enough flash for that; I was wrong!

2026-01-30 08:10:24 UTC

- reply

@nprofile…m7ah @nprofile…h46y @nprofile…pzxv not sure I fully understand, doesn't that mean the browser page with the ISO "upload" needs to stay open and active for the whole duration of what you're doing with the bootable image?

2026-01-13 06:38:26 UTC

- reply

@nprofile…7k5d @nprofile…un2n codeberg.org is hosted forgejo

2026-01-12 19:43:49 UTC

- reply

@nprofile…ds0l @nprofile…7q7j I might need to know what that tag is, specifically, for looking-at-cats-being-measured reasons.

2026-01-07 11:37:42 UTC

- reply

@nprofile…mv7s -ly, then you can of course use that instead. In practice, as ling as the actual impulse response is symmetrical enough around its maximum (i.e., where it contributes most), sinc interpol is fine, and we iften find ourselves doing it with but the observed maximum sample and one to each side of it, and approximating the sinc x by x², because that makes finding the place of extrema easy

2025-12-27 10:46:01 UTC

- reply

@nprofile…mrzh wait, Mini Berliner? Are they filled with jam?

2025-12-16 12:20:44 UTC

- reply

@nprofile…nmzg @nprofile…d4jx how to end up on various lists in 1 easy step (-up converter)

2025-12-10 12:14:35 UTC

- reply

@nprofile…2cds @nprofile…aax0 I do have a few questions on what the actual magnetic field looks like for these, but I might just want to sit dowmln with your inductors first and think. But: first row of your 16 examples, the density's highest in the center, field lines there normal to pcb. Last row I find really hard to visualize.

2025-12-10 12:09:43 UTC

- reply

@nprofile…2cds @nprofile…aax0 I can see that (but think it's an unlikely use case that you'd want a ferrite sheet applied to the surface, but not use a discrete inductor).

2025-12-10 12:04:24 UTC

- reply

@nprofile…2cds @nprofile…aax0 huh, but thats already feasible sizes for SMPSes, neat

2025-11-24 16:27:59 UTC

- reply

@nprofile…syqt I'd argue that hinges on the synchronization methods you can and want to use, and which kind of representation you want to give that in C++?
(It makes a difference whether access has to happen through something that causes a memory barrier, or whether you know that all writes are atomic and `volatile` and a linker script are good enough)

2025-11-06 08:35:38 UTC

- reply

@nprofile…fqnk persönliches Dankeschön dafür

2025-10-25 13:03:31 UTC

- reply

@nprofile…mrzh I don't speak Hindi *at all* nor can I read sanskrit, but the difference in font sizes really bothers me

2025-10-16 18:30:37 UTC

- reply

@nprofile…3qhu muss ich mal schauen – ich hab den Haken am Anfang von Corona auch häufig benutzt, bin davon aber abgewichen, da die Gummis trotzdem am Ohr scheuerten; aber ich war auch zwischendurch auf selbstbau-verlängertem-Doppelhaken, und das war echt OK.

2025-10-15 09:42:15 UTC

- reply

@nprofile…mrzh I love how a quasi-monopolist tells potential customers of Nintendo, a company not especially known for their shy legal department, that no, that product does not exist.

2025-10-12 20:50:24 UTC

- reply

@nprofile…8x2j @nprofile…4g7k sounds like gimp's cage transform (tools->transform tools->cage)

2023-11-28 22:06:37 UTC

- reply

@npub10ae…nlyc you're mean – the URL on the silkscreen isn't live yet!

2023-10-04 09:57:22 UTC

- reply

@npub19la…hm4g hm. But that basically necessitates solid RPC between an unprivileged client and a privileged system management daemon, which in itself is exactly where we tend to find parsing bugs for decades (like the glibc one triggering this discussion is).
I'd honestly rather see a kind-of-single-syscall-suid mechanism that only works with a sensible verifiable pledge()-equivalent. Like, think of an xattr that contains ebpf code describing that a privileged open can open exactly one file, and...

2023-08-29 12:42:16 UTC

- reply

@npub1e7m…7x48 The safe way is *really* having the grid voltage handled in a closed, isolated enclosure, and having a 5V cable come out of that. Anything else won't be importable into the EU, pretty much.
So, I'd argue "USB wall wart" is a) an extremely cheap, multi-sourcable, astonishingly resilient and reliable power source and b) "put your electronics into a case with USB connector somehow attached to it" is pretty much preferable to integrating it into the wall wart, for cost,
2/N

2023-08-29 12:38:30 UTC

- reply

@npub1e7m…7x48 Again, sounds like USB supply + protoboard w/ USB cable soldered on + double-sided sticky tape to me :)
But yeah, what you describes exist in various forms; there's simply "wall-wart enclosures" [1], but I'd strongly discourage anyone to just throw a power supply in one and ship that to consumers ("Makers" is just pretty wording for "unskilled person willing to experiment", in terms of danger), as you'd be exposing them to the chance of electrocution.
1/N
[1] https://www.tme.eu/de/en/katalog/power-supplies-enclosures_100628/

2023-08-29 12:22:58 UTC

- reply

@npub1e7m…7x48 You mean like a protoboard with a USB cable to plug into a power bank? I don't think "soldering a USB cable to a protoboard" really is worth making that a product…

2023-08-25 13:31:22 UTC

I don't want to alarm anyone, but archive.org @npub1xrj…9q43 is down, and especially not @npub1s2r…7gng / @npub1qz2…nlrg …
https://files.mastodon.social/media_attachments/files/110/950/498/841/474/923/original/85f3b48858a2bfe9.png

2023-08-09 19:22:07 UTC

- reply

@npub1sly…t8j7 der FTL macht eigentlich als Kernaufgabe ganz andere Dinge – nämlich aus einem Haufen mittelmäßig zuverlässiger Speicherzellen mittels Fehlerkorrekture, wildem Rumgemappe und der dafür nötigen Tabellenpflege schönen, zuverlässigen, und linear denkbaren Speicher darstellen, dass dein Dateisystem nicht selbst ECC, wear leveling, remapping… implementieren muss.