Great article! Shame it left me curious to know how a further implementation using more pure Assembly would’ve performed vs using intrinsics. Anyone know or is it easy to just assume “faster”?
I've generally found no/minimal change between assembly and intrinsics. Once I start using them I tend to look at the generated assembly to see what's actually being generated and to make sure the compiler isn't doing something surprising.
https://bayesianneuron.com - I barely have time to write anymore but its a blog in 2 parts, one is a modern more professional-esque where I write about ML, discrete optimization, and security. The other is a good old fashioned 90s style geocities type “space” where I rant and do random writings
How is it worse than having to use up all my RAM in a few browser tabs with unresponsive PWAs such as Outlook for work? I much rather have Thunderbird or Evolution silently in the background popping up notifications when I get emails - not to mention it’s much easier to compartmentalise within a DE, i.e I know my email is workspace #4 instead of having to flash scroll through all my open tabs
> today the scanning is only for images uploaded to iCloud, and only for CSAM
This, to me, would be exponentially less privacy invasive as I’ve come to assume all major cloud hosting providers implement something like this (look at Google Drive), but Apple has said that the scanning is done on-device, meaning whether or not you upload your photo library to iCloud, your local photos will be scanned with an on-device database of hashes.
Essentially iOS photos now implement a direct API call to the feds with some vague “human verification” layer if you go above an unknown threshold
Right, this is what all the articles on this matter are getting wrong, unknowingly or otherwise. It’ll be one flag switch to change from scanning uploaded images to scanning local images.
This is an erosion of fundamental human rights under the guise of “think of the children“ so that anyone who stands up against this tyranny can be labelled a “pedophile”. 1984 wasn’t like 1984 but 2021 is surely looking that way.
It’s a delayed concern for us in the West. First the bait and switch will happen in all parts of the third world, autocratic, and dictatorial governments. After all, our list of hashes will be different from what their agencies will provide. Across the globe Apple will assist in this.
Then one day we get another Trumpian President who adds more categories of hashes in a place like America.
What Apple is not willing to do is have that philosophical and ethical discussion with its customers. It has simply made the decision.
> I’ve come to assume all major cloud hosting providers implement something like this
They have, for the past decade.
>The system that scans cloud drives for illegal images was created by Microsoft and Dartmouth College and donated to NCMEC. The organization creates signatures of the worst known images of child pornography, approximately 16,000 files at present. These file signatures are given to service providers who then try to match them to user files in order to prevent further distribution of the images themselves, a Microsoft spokesperson told NBC News. (Microsoft implemented image-matching technology in its own services, such as Bing and SkyDrive.)
I feel the same, just with the times flipped around. For me the morning is great for small random/admin type tasks. Even reading a research paper or spending some time learning about something I need to do/use.
In the afternoon is when I can concentrate for 2-4 hours and do some intense work. Take a break after that, and iron out anything small before calling it a day.
I love the demoscene. I’ve been trying to get into it for some time now but it’s actually hard in terms of that are very little if any resources aimed at beginners and demo examples and tutorials.
The resources I’ve found are either incomplete or assume one is already familiar with the domain.
I understand graphics to the extent I’ve written a raytacer and a “semi-caster”, but can’t find my foot in demos, especially translating my existing graphics knowledge to ASM.
Anyone have any resources or information I may have missed?
Praise the Lord,
I found https://in4k.github.io/wiki/win32 to have useful links. Particularly Compofiller studio is quite easy to get started with and rather well done. To make a final compressed executable takes a couple of steps, where you have to switch the mode and minify the shader code which is on the Compofiller UI before creating the executable. If you can do some coding on shadertoy.com it is rather easy to transfer to a production here.
For the music, a suggestion would be to download the free version of renoise.com which is full featured and works fine for making demos. It is a good piece of software and worth the amount they ask for the full version if you can do it too i think. And then download the http://4klang.untergrund.net/ 4klang vst and use it with Renoise. You set up multiple instruments that link to the same instance of the VST instead of multiple instances of the VST and use different channels for each instrument (basically use the version that shows up top after you create one instance) and you can start by loading one of the preset patches included with 4klang. You record the songs in the 4klang vst and then put the generated files in the compofiller studio project directory and once it recompiles it has your new song.
I played around with this for the release i helped make for revision 2020 ( https://www.pouet.net/prod.php?which=85364 glsl source included ) Glory to Jesus, After this you can go lower level with the other 4k tutorials learning how to integrate with other assemblers or perhaps C++ compilers including Visual Studio community edition. It is possible to make even 4ks in C. If you want to get started in assembler i would recommend starting with the 256byte intros in a freedos installation. While I included source code in a 256byte that was presented at revision, i would start out with other tutorials.
Start out by researching some basic instructions to be able to set memory in a loop to a pointer in es:di. The basic instructions to set memory and registers, increment, compare, jump. Look a little bit into interrupts to be able to detect a keypress, or set graphics mode with 256 colors.
So for a freedos program you can do this with "mov ax, 13h; int 10h;" at the beginning of the program, set a label for the frame_loop_aleluya, then a label for the pixel_loop_aleluya (praising God in my code makes it more enjoyable) write to 0xA0000 for 320*200 bytes in the pixel_loop then detecting if a keypress has been made with "mov ah, 1;int 16h;jz frame_loop_aleluya" and then back to text mode "mov ax, 3h; int 10h" at the end of your program with ret. You can play around with that to make your first own little asm intro. Compile it with nasm to a .com. This is included in freedos.There is a little more boilerplate that goes into a .com as well, put "BITS 16; org 100h; section .text;" and a start: label . This works if you have no memory variables of your own, which is fine for a first step.
Sorry, this is my first hacker news comment so i will see how this looks and make corrections if needed, and post a small full sample program as a reply. God guide us and bless you in Jesus name
Praise the Lord! https://gist.github.com/loveJesus/29b85c3d8d6f4fa5cc85d4f33c... is a verbosely commented simple intro that compiles to 44 bytes. The source is just over the length where I think it would be better to visit the link than post it directly. If someone else could tell me their thoughts on proper protocol here i appreciate it. God guide us and bless you in Jesus Holy name
May I recommend to add some breaks into your text for easier reading. You have to insert an empty line for it to work. (Press return not just once but twice.)
I feel like "herd immunity" is very speculative right now and not fully thought out. There are even reports and studies suggesting immunity might not be something everyone gets...
I write about ML, optimization & CS, and... well, whatever I want or find interesting. I have a public backlog of projects and blog posts too.
I started it because I kinda like writing, but also because when I do write (technical topics) I enjoy giving very clear, fully understandable explanations.
Not a fan of the “here’s concept A, it’s very straightforward... we’re now at concept Z which as you can see makes use of A” style of writing. I understand it’s need and use, you can't always explain everything, especially if the topic is already very niche and highly complex, or maybe you don’t have the space.
I find that it lets me learn more or jogs my memory on other subjects when I have to fully explain and try to teach things.
I’m an ML engineer (was) and am looking for contract work during this time. In previous I also worked as a sysadmin and a python engineer. I’ve experience managing large servers and AWS. I also have experience in optimization and love hard problems. MSc in CS
Location: GMT
Remote: Yes (ONLY)
Willing to relocate: No.
Technologies: Python, PyTorch, Sklearn, Postgress, Mongo, AWS
Resumé/CV: I wish to remain anonymous on here so please use my email and I’ll get back to you.
Email: kl3u4z+a3od9aa48ibiw@sharklasers.com
