Numerlor

6 ms·

31.
by Numerlor 9mo ago
Have you tried e.g. Mojo that can vectorize/do SIMD without having to do intrinsics everywhere?
32.
by Numerlor 9mo ago
Computer monitors have been getting a lot better while being cheaper, with no ads or services. You can get a high resresh rate 4K ips for about $200 nowadays. Display tech is just advancing faster than other tech at the moment
33.
by Numerlor 9mo ago
Why is their browser using so much memory? I'm quite bad at closing tabs since I've switched to vertical and just open new windows instead. And even with that I don't think I ever broke 10gb on edge except for when I opened
34.
by Numerlor 9mo ago
> LNL throttles heavily even on the default profile, not just power saver modes. This does also show it not changing in other benchmarks, but I don't have a LNL laptop myself to test things myself, just going off of what people I kn
35.
by Numerlor 9mo ago
Nobody is duplicating the macbook experience because Apple is integrating both hardware and os, while others are fighting Windows, and OEMs being horrible at firmware. LNL should only power throttle when you go to power saver modes, battery
36.
by Numerlor 9mo ago
Lost faith from what? On x86 mobile Lunar lake chips are the clear best for battery life at the moment, and mobile arrowlake is competitive with amd's offerings. Only thing they're missing is a Strix halo equivalent but AMD messed
37.
by Numerlor 9mo ago
It was with the fast profile using both software and hardware deciding, important detail I forgot was that the video was av1. Don't have the link to it now but it was from jellyfin's test files
38.
by Numerlor 9mo ago
I did recently see someone compare mpv and vlc on a 8k HDR @ 60fps file with mpv really lagging while vlc doing it fine. I could confirm the mpv lags but don't have vlc, so not sure if it's just better in that specific case or di
39.
by Numerlor 9mo ago
Most articles I've seen from them on here have been negative enough that I'd assume bias the other way
40.
by Numerlor 9mo ago
From my tests defender has minimal impact on performance even when doing a full scan, except for making some io slower when you're e.g. unpacking new files but NTFS is plenty slow by itself there Enterprise likes to layer multiple inva
41.
by Numerlor 9mo ago
There aren't because nobody is betting on ai demand to last. Then they'd have a couple billion dollar fab sitting around doing nothing and employees that'd have to be fired. There already was scaling back for dram and and pro
42.
by Numerlor 9mo ago
Afaik production of nand was reduced as some of the lines can be repurposed for dram that's more in demand. Significantly increasing supply is also a huge multi year investment into a new fab that'd likely not pay out when the art
43.
by Numerlor 10mo ago
Value wise when trying to spec out my personal Lenovo laptop on framework, it'd never get anywhere close to being worth it even if I completely made use of the hardware after a future upgrade. Framework makes sense if you're going
44.
by Numerlor 10mo ago
The polish is what they seem to have trouble with lately. I much prefer their ides to say vscode, but their development has been a mess for a while with half-assed implementations and long standing bugs
45.
by Numerlor 10mo ago
The socket io locks in the amount of memory channels. Some pins could be repurposed but that's pretty much a new socket anyway. They could in theory do on package dram as faster first level memory, but I doubt we'll see that anyti
46.
by Numerlor 10mo ago
Fwiw if you want a simple pastebin, I've been running pinnwand for a couple years without any issues off of a single short docker compose file, I think running it on host also shouldn't be complicated
47.
by Numerlor 10mo ago
DDR4 manufacturing is mostly shut down, so if any real demand picks starts there the prices will shoot up
48.
by Numerlor 10mo ago
This article about them got published just yesterday... https://news.ycombinator.com/item?id=46124883 There's a lot of misleading information in what they publish, plagiarism, and I believe some information that wouldn
49.
by Numerlor 10mo ago
But also note that a memory test passing doesn't mean the ram (or CPU) is fine. I overclocked my system for fun and for ram you'd run 2-3 different tests to be able to call it stable with some confidence, as they all have differen
50.
Influence as a Service: SemiAnalysis Under the Microscope(jon4hotaisle.substack.com)
6 points by Numerlor 10mo ago|0 comments
51.
by Numerlor 11mo ago
I'll admit, bit of a poor choice of word,l. But when you need to do e.g. physical security, costs add up quickly over what you'd spend on cloud in say a year, and the compliance companies are usually a huge headache to deal with s
52.
by Numerlor 11mo ago
For company hosting cloud solutions gets you the various compliance stuff for free which can be worth it if you're not too large, and of course faster turnaround if you need to get a product out. For personal a cheap vps will end up co
53.
by Numerlor 11mo ago
Yes it should work with the codec pack, and they're keeping it on some of their laptops like with dedicated gpus
54.
by Numerlor 11mo ago
I recently found out I've been banned from all of their repositories on GitHub, while as far as I'm aware our only interaction was on a duplicated bug issue I created, as I didn't manage to find the original with GitHub'
55.
by Numerlor 11mo ago
Then the outrage can come when that happens? This comment section is 50% people that haven't used windows in 10 years complaining
56.
by Numerlor 11mo ago
Or, they simply prefer smaller towns without high-rises
57.
by Numerlor 11mo ago
Also anyone living at higher latitudes won't be thrilled by having the sun up even earlier. I don't live that high up compared to quite a lot of people in Europe, and during summer the sun already starts shining from below the hor
58.
by Numerlor 1y ago
Main "issue" is that evaporative cooling is just ridiculously effective compared to a normal heat exchanger with only air. You can use water or air internally but then to get rid of the heat from the facility there aren't man
59.
by Numerlor 1y ago
I believe just inference is bandwidth limited, prompt processing and other tasks on the other hand needs the compute. As I understand it, the workstation is also as a whole focused on the local development process before readying things for
60.
by Numerlor 1y ago
The cooling should be better, especially with higher wattage chips (either out of the box, or overclocked). The AIOs have gotten a lot better with faster pumps. Noise should also be lower again assuming you'd be actually cooling enough

More ›