reroute22

9 ms·

1.
by reroute22 5mo ago
I agree in both parts. In particular, repeatability is only one part of it, another is in person physical presence, or a weaker form of it - on-screen presence, or at the absolute least - voice. All of the latter plays a huge role, a lot of
2.
by reroute22 5mo ago
Execs was indeed more of a metaphorical description. A small subset of individuals working there. You don't need 80k people to make policy decisions, Meta is that big not because of policy makers, but because of infrastructure people a
3.
by reroute22 5mo ago
While reality can be anything but. As far as I understand, there is plenty of research there in disciplines raging from social studies through psychology to game theory and economics, as well as informal simulations, that strongly suggest t
4.
by reroute22 5mo ago
Don't confuse employees with execs. It's a gigantic company with almost 80k employees. Most cultures around the world are acutely aware that the actions and opinions of their leaders are not a reflection of behaviors and opinions
5.
by reroute22 3y ago
Such a fantastic paper overall, it was a pleasure to read, it's very accessible, and greatly informative. If anyone is new to the idea and is seeking a definition of AGI, reading this paper is easy and is immeasurably superior to merel
6.
by reroute22 3y ago
https://arxiv.org/abs/2310.02989 ("xVal: A Continuous Number Encoding for Large Language Models", Submitted on 4 Oct 2023)
7.
by reroute22 3y ago
It conceptually never admits ignorance and never asks for clarifications. It always produces something, to the best of its ability. It _seems_ to be a minor technical limitation (there is plenty of traditional ML systems producing confidenc
8.
by reroute22 3y ago
Thank you for providing this. Bing Chat runs on GPT-4, however [1]. And Bing gets this wrong in all 3 of its modes (Creative, Balanced, and Precise) as of time of writing. Given this experiment and similar others presented around here, it s
9.
by reroute22 3y ago
Just tried and it completely fell over: https://chat.openai.com/share/ea9cf93f-0632-48b9-aa47-1bbf31... Tried on Bing in "Precision" mode as well, and it fell over just the same, but starting with C instead o
10.
by reroute22 3y ago
I'm afraid I'm with your colleagues. I value being able to grasp what the code does at a glance while scrolling through. Your first snippet is entering my brain like as if without even reading at all, all at once and instantly. Yo
11.
by reroute22 3y ago
An easier to read research article that's narrower in subject and seemingly more relevant to the OP: https://research.nvidia.com/sites/default/files/pubs/2012-12... ("Unifying Primary Cache, Sc
12.
by reroute22 3y ago
There is a fairly informative survey on the subject: https://www.osti.gov/servlets/purl/1332070 (A Survey of Techniques for Architecting and Managing GPU Register File)
13.
by reroute22 3y ago
I'll leave your first question to the other comment here from frogblast, as I really battled with how to answer it well, given my limited knowledge and being an elbow deep into an analogy, after all. I got a writer's block, and fr
14.
by reroute22 3y ago
I doubt anyone will be able to answer questions this fine grained, not now (if the implementation is architecturally exposed - leaks into the ISA - and Asahi Linux group figures some of it out), or possibly not ever (if it's architectu
15.
by reroute22 3y ago
Not really, there is no such definition that you're referring to. Perf of a GPU can be limited by any one of the thousand little things within the micro-architectural organization of the GPU in question, any on-chip path can become the
16.
by reroute22 3y ago
Sorta, yeah! Also, your "128 cuda cores" of Skylake variety run at higher frequencies and work off of much bigger caches, so they are faster (in serial manner)... ...until they are slower, because GPU's latency hiding mechani
17.
by reroute22 3y ago
"128 * the-number-of-cores" of threads can make progress truly in parallel (at the same time). 24,576 threads (or however many, I didn't validate the number and it depends on the occupancy, which depends on thread resource us
18.
by reroute22 3y ago
Okay, I went through the other video they reference ("Discover new Metal profiling tools for M3 and A17 Pro" [1]), and there is actually a whole bunch of extra very relevant (IMO) information on the subject, starting about 13:30 o
19.
by reroute22 3y ago
Thanks!
20.
by reroute22 3y ago
Judging by the output/GUI of their GPU profiler, "complex" there is more like "complex instructions", think f32 (floating point) ops that aren't additions and multiplications (and FMAs), but trigonometry, squar
21.
by reroute22 3y ago
To be completely fair, base 14" MBP is only Pro in the name. That laptop has only 2 ports, supports only 1 external monitor, has a low core count CPU, and a relatively weak GPU, disqualifying it for any "Pro" work well before
22.
by reroute22 3y ago
Wow, that sounds like a very impressive level of separation of concerns. May I kindly ask for some links for further reading? If you have any handy.
23.
by reroute22 3y ago
Unfortunately, hardly. Ampere's (Nvidia 3000 series), Ada's (Nvidia 4000 series), and RNDA 3's (AMD 7000 series) GPUs have doubled up their FP32 units in ways that differ in implementation (between AMD and Nvidia) but are rel
24.
by reroute22 3y ago
I've no idea exactly how MS licenses uses of DX, but just for context Imagination Technologies just released a custom GPU design that implements DirectX Feature Level 11_0 (which corresponds to earlier versions of DX 12 [1]): https:&#
25.
by reroute22 3y ago
I'd say the beginning sounds like an introduction to GPU architectures in general, not marketing. Somewhere in the ballpark of 5:30-6:00 or so it describes prior hardware design of the Apple's shader core, and starting 7:00 it goe
26.
by reroute22 3y ago
A pretty detailed hypothesis showed up recently: https://forums.macrumors.com/threads/3d-rendering-on-apple-s...
27.
by reroute22 3y ago
We don't know anything about Apple's implementation, but for perspective, in Nvidia's first implementation of RTX RT cores occupied about 6-7% of the area of a TPC, where TPCs constitute the majority, but not the entire die,
28.
by reroute22 3y ago
See paragraphs 31 & 32. If I'm reading it right, they answer that question, and positively.
29.
by reroute22 3y ago
This is all blatantly false and the exact opposite is true. > The M1 doesn't really benchmark in pure CPUs tests much higher that the Intel chips it replaced It does. See Cinebench 2024 ST at https://www.cpu-monkey.com&#x
30.
by reroute22 3y ago
The answer to these sort of questions (OP) is always the same: it was done because it made all the sense in the world to do. Most likely, 150 GB/s -> 200 GB/s results in a fairly small improvement (when paired with a processor

More ›