40x Faster Binary Search
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
"I suck" -ThePrimeagen
The PrimeTime
55.8k views
It's all fake
The PrimeTime
88.9k views
1 year, no job
ThePrimeTime
184.3k views
We did it?
ThePrimeTime
102.6k views
Final stretch
ThePrimeTime
34.3k views
THIS BLEW MY MIND
ThePrimeTime
198.9k views
Claude 4 System Prompt
ThePrimeTime
93.3k views
AI Skeptic Friends
ThePrimeTime
344.3k views
Why Buying GPUs Is a Disaster
ThePrimeTime
211.0k views
Terminal Family Feud
ThePrimeTime
127.7k views
Top Comments (10)
A 62-minute article fits in a 84 minute video. This is the 1.35x developer.
Disappointed that he used Rust instead of Assembly directly.
39:55 SSE SIMD was added to CPUs in 1999 and LLVM didnt get auto vectorization until 2012. Vast majority of SIMD is written by hand.
This guy: "Then I figured the clock pulse takes 0.1 phantom seconds to go to from L3 to then process unit, so I patched some nano graphene wires to the cpu and improved 1.764x the running time"
1:13:49 oh dang, jump scare when this banger blog post references a paper from someone I've worked with (Paul Medvedev)
I have no clue what the fuck this is, i dont know programming but i love how he goes trough it. Whats even funnier is that even him cant keep up... CLASSIC ASSEMBLY
I think it's the first time I've heard this layout being referred to as an Eytzinger layout. It's typically called a heap layout or a breadth-first layout. I've been using this for many many years. It's the starting design for any kind of N-ary tree for which performance matters. If performance really matters, at a large scale, you should consider a von Emde Boas layout, which is essentially a recursively subdividing breadth-first layout of subtrees, which has some nice cache-oblivious performance properties (i.e., cache-aware optimization means coming up with an implementation that performs optimally for a specific memory architecture, and thus, the implementation depends on that parameters of the architecture (e.g., cache sizes, latencies, etc.), while cache-oblivious optimization yields a single near-optimal implementation for any memory architecture, which obviously scales much better). A saying I've adopted over the years is: "If you're not maxing out your system's memory bus, you could be doing better."
We had 3 slots, 4 children, a root node of 3, 17, 69 (nice), and a whole galaxy of numbers stored in it. Not that we need all of that to search a value, but once you get locked in into a serious data structure, the tendency is to push it as far as you can.
My man still got the Internet Comment Etiquette references down to a subconscious level
The video was definitely encouraging. My college education, decades ago, was very adv data structure oriented and seeing how a person can optimize such an old search algorithm is inspiring.
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
A 62-minute article fits in a 84 minute video. This is the 1.35x developer.
Disappointed that he used Rust instead of Assembly directly.
39:55 SSE SIMD was added to CPUs in 1999 and LLVM didnt get auto vectorization until 2012. Vast majority of SIMD is written by hand.
This guy: "Then I figured the clock pulse takes 0.1 phantom seconds to go to from L3 to then process unit, so I patched some nano graphene wires to the cpu and improved 1.764x the running time"
1:13:49 oh dang, jump scare when this banger blog post references a paper from someone I've worked with (Paul Medvedev)
I have no clue what the fuck this is, i dont know programming but i love how he goes trough it. Whats even funnier is that even him cant keep up... CLASSIC ASSEMBLY
I think it's the first time I've heard this layout being referred to as an Eytzinger layout. It's typically called a heap layout or a breadth-first layout. I've been using this for many many years. It's the starting design for any kind of N-ary tree for which performance matters. If performance really matters, at a large scale, you should consider a von Emde Boas layout, which is essentially a recursively subdividing breadth-first layout of subtrees, which has some nice cache-oblivious performance properties (i.e., cache-aware optimization means coming up with an implementation that performs optimally for a specific memory architecture, and thus, the implementation depends on that parameters of the architecture (e.g., cache sizes, latencies, etc.), while cache-oblivious optimization yields a single near-optimal implementation for any memory architecture, which obviously scales much better). A saying I've adopted over the years is: "If you're not maxing out your system's memory bus, you could be doing better."
We had 3 slots, 4 children, a root node of 3, 17, 69 (nice), and a whole galaxy of numbers stored in it. Not that we need all of that to search a value, but once you get locked in into a serious data structure, the tendency is to push it as far as you can.
My man still got the Internet Comment Etiquette references down to a subconscious level
The video was definitely encouraging. My college education, decades ago, was very adv data structure oriented and seeing how a person can optimize such an old search algorithm is inspiring.