enumerator4829
@enumerator4829@sh.itjust.works
- Comment on Windows 11 to add an AI agent that runs in background with access to personal folders, warns of security risk 2 weeks ago:
If you want to encode information into only the depth of your recursive identically named folders, you have 128 different depths, one character for the name, one for the slash, per level. Yields about 128 possible levels. Leave one off for the last filename, 127.
If we want to name our folders something longer than a single character, we can store less files. If we want to store our files on linux, by default we get 4096 characters to play with, so about 2k levels (unless we compile our own linux kernel with PATH_MAX set for this very specific purpose). If we run CIFS we may be able to reach up to 16k levels.
That was my interpretation of OPs (admittedly bad) idea. Personally, I try to avoid implementing inodes as Church numerals.
- Comment on Google CEO: If an AI bubble pops, no one is getting out clean 2 weeks ago:
Who do you think pays for the bailouts in the end?
- Comment on Windows 11 to add an AI agent that runs in background with access to personal folders, warns of security risk 2 weeks ago:
You can now have 128 files in your filesystem.
- Comment on You can do anything at Zombocom 3 weeks ago:
All connectors are compatible given enough violence.
- Comment on You can do anything at Zombocom 3 weeks ago:
The VGA connector is actually reversible! Once. given enough force. It also doesn’t actually work properly when installed upside down.
- Comment on Nvidia reveals Vera Rubin Superchip for the first time — incredibly compact board features 88-core Vera CPU, two Rubin GPUs, and 8 SOCAMM modules 5 weeks ago:
Sorry, no OpenGL support.
- Comment on Microsoft seemingly just revealed that OpenAI lost $11.5B last quarter 5 weeks ago:
It’s literally the same chip designers, production facilities and software. Every product using <5nm silicon fabs compete for the same manufacturing capabilities (fab time at TSMC in Taiwan) and all Nvidia GPUs share lots of commonalities in their software stack.
The silicon fab producing the latest Blackwell AI chips is the same fab producing the latest consumer silicon for both AMD, Apple, Intel and Nvidia. (Let’s ignore the fabs making memory for now.) Internally at Nvidia, I assume they have shuffled lots and lots of internal resources over from the consumer oriented parts of the company to the B2B oriented parts, severely reducing consumer focus.
And then we have any intentional price inflation and market segmentation. Cheap consumer GPUs that are a bit too efficient at LLM inference will compete with Nvidias DC offerings. The amount of consumer grade silicon used for AI inference is already staggering, and Nvidia is actively holding back that market segment.
- Comment on Microsoft seemingly just revealed that OpenAI lost $11.5B last quarter 5 weeks ago:
I want cheap GPUs at home please!
- Comment on Google Confirms Non-ADB APK Installs Will Require Developer Registration 1 month ago:
Normal people aren’t flashing custom ROMs. The audience for some FOSS software will shrink by several orders of magnitude.
But the pain really kicks in when your government/bank/streaming apps require attestation of a signed boot chain and Google Play services running.
- Comment on You should know how to coil cables 2 months ago:
For the record - analog multis can burn in hell. Nowadays, not running all of the show over Cat6 should be criminal.
- Comment on You should know how to coil cables 2 months ago:
For anyone working on or around stages:
Most sane production companies standardise on over-under. Even if you find some other method superior (nothing is), you’ll get thrown out headfirst if you don’t follow the standard. Having a tech fuck around with a non-compliant cable during a changeover is far too risky.
Should be noted that there are special cases. For example, thicccc cables (i.e. 24ch analog multi) that have their own dedicated cases often go down in an 8 instead - easier to pull out and you can use a smaller case. Thank god for digital audio.
(Also, when using over-under correctly, you can throw the cable and it will land straight without any internal stresses winding it up like a spring)
- Comment on Google's shocking developer decree struggles to justify the urgent threat to F-Droid 2 months ago:
I can agree on Apple not really having a properly supported hardware repair ecosystem, and actively working against third party repair.
But the software? When Samsung and friends had 2-4 years of security updates, Apple had almost twice that. The iPhone XS still has support, 6 years after end-of-sale, 7 years from release. Normal people can’t be expected to flash their phones with LineageOS. The situation is slightly better nowadays, but Samsung still seems to be depreciating 3 year old devices: endoflife.date/samsung-mobile
- Comment on Tailscale difficulties 2 months ago:
Here I am, running separate tailscale instances and a separate reverse proxy for like 15 different services, and that’s just one VM… All in all, probably 20-25 tailscale instances in a single physical machine.
Don’t think about Tailscale like a normal VPN. Just put it everywhere. Put it directly on your endpoints, don’t route. Then lock down all your services to the tailnet and shut down any open ports to the internet.
- Comment on Those who are hosting on bare metal: What is stopping you from using Containers or VM's? What are you self hosting? 2 months ago:
My NAS will stay on bare metal forever. Any complications there is something I really don’t want. Passthrough of drives/PCIe-devices works fine for most things, but I won’t use it for ZFS.
As for services, I really hate using Docker images with a burning passion. I’m not trusting anyone else to make sure the container images are secure - I want the security updates directly from my distribution’s repositories, and I want them fully automated, and I want that inside any containers. Having Nixos build and launch containers with systemd-nspawn solves some of it. The actual docker daemon isn’t getting anywhere near my systems, but I do have one or two OCI images running. Will probably migrate to small VMs per-service once I get new hardware up and running.
Additionally, I never found a source of container images I feel like I can trust long term. When I grab a package from Debian or RHEL, I know that package will keep working without any major changes to functionality or config until I upgrade to the next major. A container? How long will it get updates? How frequently? Will the config format or environment variables or mount points change? Will a threat actor assume control of the image? (Oh look, all the distros actually enforce GPG signatures in their repos!)
So, what keeps me on bare metal? Keeping my ZFS pools safe. And then just keeping away from the OCI ecosystem in general, the grass is far greener inside the normal package repositories.
- Comment on Report: Microsoft's latest Windows 11 24H2 update breaks SSDs/HDDs, may corrupt your data 3 months ago:
$ su - # rm -rf —no-preserve-root /
Should do the trick. (Obviously don’t try it unless you know what you are doing and know what may happen when it hits your EFI variables.)
- Comment on HELP HIM. 3 months ago:
Computational biochemistry is slowly getting there. Alphafold was a big breakthrough, and there is plenty of ongoing research simulating more and more.
We can probably never get rid of animal testing entirely for clinical research, we’ll always need to validate simulations in animals before moving on to humans.
I do however agree that animal testing outside of clinical research approved by a competent independent ethics committee can fuck right off. (Looking at you, cosmetics industry)
- Comment on Popup Ads in Your Pickup Truck? RAM Trucks Now Feature Scammy Ads on the Center Display 3 months ago:
I don’t think there is much overlap between the sets of people
- buying these cars
- having the competence to hack them
- having the willingness and finances to potentially brick the car
- Comment on Spotify fans threaten to return to piracy as music streamer introduces new face-scanning age checks in the UK 3 months ago:
I wonder if ancient crunchy low bitrate mp3s will be an aesthetic, the way that dusty vinyl or worn out tapes are?
- Comment on Duckstation(one of the most popular PS1 Emulators) dev plans on eventually dropping Linux support due to Linux users, especially Arch Linux users. 4 months ago:
Most arch users are casuals that finally figured out how to read a manual. Then you have the 1% of arch users who are writing the manual…
It’s the Gentoo and BSD users we should fear and respect, walking quietly with a big stick of competence.
- Comment on China advances toward tech independence with new homegrown 6nm gaming and AI GPUs — Lisuan 7G106 runs Chinese AAA titles at 4K over 70 FPS and matches RTX 4060 in synthetic benchmarks 4 months ago:
Yeah, that’s the thing.
The gaming market only barely exists at this point.
- Comment on China advances toward tech independence with new homegrown 6nm gaming and AI GPUs — Lisuan 7G106 runs Chinese AAA titles at 4K over 70 FPS and matches RTX 4060 in synthetic benchmarks 4 months ago:
Pheasantsgamers buycheap inference cardsgaming cards.The absolute majority of Nvidias sales globally are top-of-the-line AI SKUs. Gaming cards are just a way of letting data scientists and developers have cheap CUDA hardware at home (while allowing some Cyberpunk), so they keep buying NVL clusters at work.
Nvidia’s networking division is probably a greater revenue stream than gaming GPUs.
- Comment on Thoughts?? 4 months ago:
I have fucked around enough with R’s package management. Makes Python look like a god damn dream. Containers around it is just polishing a turd. Still have nightmares from building containers with R in automated pipelines, ending up at like 8 GB per container.
Also, good luck getting reproducible container builds.
Regarding locales - yes, I mentioned that. Thats’s a shitty design decision if I ever saw one. But within a locale, most Excel documents from last century and onwards should work reasonably well. (Well, normal Excel files. Macros and VB really shouldn’t work…). And it works on normal office machines, and you can email the files, and you can give it to your boss. And your boss can actually do something with it.
I also think Excel should be replaced by something. But not R.
- Comment on Thoughts?? 4 months ago:
R, the language where dependency resolution is built upon thoughts and prayers.
Say what you want about Excel, but compatibility is kinda decent (ignoring locales and DNA sequences). Meanwhile, good luck replicating your R installation on another machine.
- Comment on Very large amounts of gaming gpus vs AI gpus 4 months ago:
the H200 has a very impressive bandwith of 4.89 TB/s, but for the same price you can get 37 TB/s spread across 58 RX 9070s, but if this actually works in practice i don’t know.
Your math checks out, but only for some workloads. Other workloads scale out like shit, and then you want all your bandwidth concentrated. At some point you’ll also want to consider power draw:
- One H200 is like 1500W when including support infrastructure like networking, motherboard, CPUs, storage, etc.
- 58 consumer cards will be like 8 servers loaded with GPUs, at like 5kW each, so say 40kW in total.
Now include power and cooling over a few years and do the same calculations.
As for apples and oranges, this is why you can’t look at the marketing numbers, you need to benchmark your workload yourself.
- Comment on Very large amounts of gaming gpus vs AI gpus 4 months ago:
Well, a few issues:
- For hosting or training large models you want high bandwidth between GPUs. PCIe is too slow, NVLink has literally a magnitude more bandwidth. See what Nvidia is doing with NVLink and AMD is doing with InfinityFabric. Only available if you pay the premium, and if you need the bandwidth, you are most likely happy to pay.
- Same thing as above, but with memory bandwidth. The HBM-chips in a H200 will run in circles around the GDDR-garbage they hand out to the poor people with filthy consumer cards. By the way, your inference and training is most likely bottlenecked by memory bandwidth, not available compute.
- Commercially supported cooling of gaming GPUs in rack servers? Lol. Good luck getting any reputable hardware vendor to sell you that, and definitely not at the power densities you want in a data center.
- TFLOP16 isn’t enough. Look at 4 and 8 bit tensor numbers, that’s where the expensive silicon is used.
- Nvidias licensing agreements basically prohibit gaming cards in servers. No one will sell it to you at any scale.
For fun, home use, research or small time hacking? Sure, buy all the gaming cards you can. If you actually need support and have a commercial use case? Pony up. Either way, benchmark your workload, don’t look at marketing numbers.
Is it a scam? Of course, but you can’t avoid it.
- Comment on Huawei shows off data center supercomputer that is better “on all metrics” 7 months ago:
Please note that the nominal FLOP/s from both Nvidia and Huawei are kinda bullshit. What precision we run at greatly affect that number. Nvidias marketing nowadays refer to fp4 tensor operations. Traditionally, FLOP/s are measured with fp64 matrix-matrix multiplication. That’s a lot more bits per FLOP.
Also, that GPU-GPU bandwidth is kinda shit compared to Nvidias marketing numbers if I’m parsing correctly (NVLink is 18x 10GB/s links per GPU, big ’B’ in GB). I might read the numbers incorrectly, but anyway. How and if they manage multi-GPU cache coherency will be interesting to see. Nvidia and AMD both do (to varying degrees) have cache coherency in those settings. Developer experience matters…
Now, the real interesting thing is power draw, density and price. Power draw and price obviously influence TCO. On 7nm, I guess the power bill won’t be very fun to read, but that’s just a guess. The density influences network options - are DAC-cables viable at all, or is it (more expensive) optical all the way?
- Comment on What could possibly go wrong? DOGE to rapidly rebuild Social Security codebase. 8 months ago:
Document databases are the future /s
- Comment on FBI warnings are true—fake file converters do push malware 8 months ago:
What? Just base64 encrypt it before you store it in the git hub
- Comment on 'Writing is on the wall for spinning rust': IBM joins Pure Storage in claiming disk drives will go the way of the dodo in enterprises 8 months ago:
The flaw with hard drives comes with large pools. The recovery speed is simply too slow when a drive fails, unless you build huge pools. So you need additional drives for more parity.
I don’t know who cares about shelf life. Drives spin all their lives, which is 5-10 years. Use M-Disk or something if you want shelf life.
- Comment on Alibaba doubles down on RISC-V architecture with a new secretive 'server-grade' chip that will put AMD and Intel on alert 8 months ago:
I agree with you, mostly. Margins in the datacenter are thin for some players. Not Nvidia, they are at like 60% pure profit per chip, including software and RnD. That will have an effect on how we design stuff in the next few years.
I think we’ll need both ”GPU” and traditional CPUs for the foreseeable future. GPU-style for bandwidth or compute constrained workloads and CPU-style for latency sensitive workloads or pointer chasing. Now, I do think we’ll slap them both on top of the same memory, APU-style á la MI300A.
That is, as long as x86 has the single-threaded advantage, RISC-V won’t take over that marked, and as long as GPUs have higher bandwidth, RISC-V won’t take over that market.
Finally, I doubt we’ll see a performant RISC-V chip from China the next decade - they simply lack the EUV fabs. From outside of China, maybe, but the demand isn’t nearly as large.