refibrillator

7 ms·

31.
by refibrillator 2y ago
Pay attention to IO bandwidth if you’re building a machine with multiple GPUs like this! In this setup the model is sharded between cards so data must be shuffled through a PCIe 3.0 x16 link which is limited to ~16 GB/s max. For refere
32.
by refibrillator 2y ago
In theory no, a network partition is indistinguishable from infinite latency. Though in common use I think “latency” tends to imply a response and measurable timing.
33.
by refibrillator 2y ago
CAP theorem is great, though it does omit latency. Which leads you to the logical extension, PACELC [1]: If there is a network Partition you must choose Availability or Consistency, Else the tradeoff is Latency vs Consistency. This offers p
34.
by refibrillator 2y ago
I recognize the author Jascha as an incredibly brilliant ML researcher, formerly at Google Brain and now at Anthropic. Among his notable accomplishments, he and coauthors mathematically characterized the propagation of signals through deep
35.
by refibrillator 2y ago
Direct link to the PDF: https://drive.google.com/file/d/1YaG9xpu-WQKBPUi8yQ4HaDYQLUS...
36.
by refibrillator 2y ago
> Pgvector is very slow, seconds to 10’s of seconds, on filter and order by queries This is an absurd claim to make with no qualification on index type, number of rows, etc. Really undermines the authority of the post, even though the ot
37.
by refibrillator 2y ago
Hmm there may be a bug in the authors’ python script that searches google scholar for the phrases "as of my last knowledge update" or "I don't have access to real-time data". You can see the code in appendix B. The
38.
by refibrillator 2y ago
This paper is light on background so I’ll offer some additional context: As early as the 90s it was observed that CPU speed (FLOPs) was improving faster than memory bandwidth. In 1995 William Wulf and Sally Mckee predicted this divergence w
39.
Windows 9x TLB Invalidation Bug(blog.stuffedcow.net)
3 points by refibrillator 2y ago|0 comments
40.
by refibrillator 2y ago
Call me a skeptic but this whole charade about privacy aware infrastructure seems more likely to be about plausible deniability rather than genuine compliance. If it’s truly such a complex and monumental effort for a global company like Met
41.
by refibrillator 2y ago
You are incorrect, this is an interactive simulation that is playable by humans. > Figure 1: a human player is playing DOOM on GameNGen at 20 FPS. The abstract is ambiguously worded which has caused a lot of confusion here, but the pap
42.
by refibrillator 2y ago
Just want to clarify a couple possible misconceptions: The diffusion model doesn’t maintain any state itself, though its weights may encode some notion of cause/effect. It just renders one frame at a time (after all it’s a text to imag
43.
by refibrillator 2y ago
There is no text conditioning provided to the SD model because they removed it, but one can imagine a near future where text prompts are enough to create a fun new game! Yes they had to use RL to learn what DOOM looks like and how it works,
44.
by refibrillator 2y ago
Nice, what are the odds though? I searched around but couldn’t find any hard stats. Is this like nuclear bunker in my basement type of precaution?
45.
by refibrillator 2y ago
Novel view synthesis via 3DGS requires knowledge of the camera pose for every input image, ie the cam position and orientation in 3D space. Historically camera poses have been estimated via 2D image matching techniques like SIFT [1], throug
46.
by refibrillator 2y ago
I love her sense of humor! One story she tells is about the world’s first computer bug [1], I had never heard it nor the history of the word. She also mentions they were using computers to enhance satellite photos, it took 3 days to process
47.
The Bug in the Computer Bug Story (2022)(daily.jstor.org)
36 points by refibrillator 2y ago|10 comments
48.
by refibrillator 2y ago
Upvoted to spread the immense wisdom in this post. But golly I must say the line can get blurry quickly. > Would anything of value be lost if this or that chunk of it was removed? In early stage projects I’ve seen this mentality backfire
49.
by refibrillator 2y ago
I found some AirPods on a remote trail awhile back, case and all. Batteries were completely dead. Once charged they paired to my iphone with no indication of a previous owner, besides the device name of John’s AirPods or whatever. I tried b
50.
by refibrillator 2y ago
Yeah jitter is generally used to mitigate “thundering herd” type problems because it reduces the peak load by spreading it out over time.
51.
by refibrillator 2y ago
Running a server just to set the user agent header to the googlebot one for some requests feels a bit heavyweight. But perhaps it’s necessary, as it seems Firefox no longer has an about:config option to override the user agent…am I missing
52.
by refibrillator 2y ago
Paper (2020): https://arxiv.org/pdf/2003.07267 As a layperson I found the first page to be more succinct and intuitive than the article. > Let Alice have such a processor that implements fast information scrambling
53.
by refibrillator 2y ago
In 3DGS, spherical harmonics are used to model view-dependent changes in color. https://en.m.wikipedia.org/wiki/Spherical_harmonics Basically for each Gaussian there is a set of coefficients and those are used to calcu
54.
by refibrillator 2y ago
If you’re interested in learning more about the rich human history and ingenuity underpinning the Hubble “constant”, please do yourself a favor and scroll through The Cosmic Distance Ladder by Terence Tao of UCLA: https://terryta
55.
by refibrillator 2y ago
I definitely watched it but missed a couple of those sentences so thanks for the timestamps above. Yes he does name soil migration as the biggest problem with subsurface drains, but to be fair the video is contextualized around dams. The sp
56.
by refibrillator 2y ago
I’ve built quite a few french drains in residential settings, some lasting longer than others - I’ll share a few hard earned lessons here: Soil migration is the number one failure mode. This is mentioned but perhaps a bit understated in the
57.
by refibrillator 2y ago
Actually the provisional patent # 60/376,055 was filed on April 25 2002. So if the paper was indeed published in Oct 2003 then all is well.
58.
by refibrillator 2y ago
Here is a link to that HN thread from 2013: https://news.ycombinator.com/item?id=5723863
59.
by refibrillator 2y ago
The code has a comment which seems to hint that Tri Dao was working on FA3 as early as April 2022, the month after Hopper/H100 was announced. I find it mildly curious that over 2 years has elapsed before the code was released today. Pe
60.
by refibrillator 2y ago
Previous discussion this week: https://news.ycombinator.com/item?id=40817852

More ›