Communications of the ACM
Refine your search:
How the “Frontier” Became the Slogan of Uncontrolled AI
Artificial intelligence (AI) has been billed as the next frontier of humanity: the newly available expanse whose exploration …
From Daniel Lemire's Blog
Iterating in batches over data structures can be much faster…
We often need to iterate over the content of data structures. It is surprisingly often a performance bottleneck in big-data applications. Most iteration code works...
From Daniel Lemire's Blog
Introducing GapminderVR: Data Visualization in Virtual Reality
I am a big fan of sites such as Gapminder and Our World in Data. Such data visualization sites are like intellectual pornography. You want to know which countries...
From Daniel Lemire's Blog
Science and Technology links (April 13th, 2018)
Somewhat depressingly, there is very little evidence that you can improve people’s overall cognitive abilities: Although cognitive ability correlates with domain...
From Daniel Lemire's Blog
For greater speed, try batching your out-of-cache data accesses
In software, we use hash tables to implement sets and maps. A hash table works by first mapping a key to a random-looking address in an array. In a recent series...
From Daniel Lemire's Blog
Science and Technology links (April 7th, 2018)
Mammals have a neocortex, some kind of upper layer on top of our ancestral brain. It is believed to be the key evolutionary trick that makes mammals smarts. Yet...
From Daniel Lemire's Blog
Caching hash values for speed (Swift-language edition)
In my posts Should you cache hash values even for trivial classes? and When accessing hash tables, how much time is spent computing the hash functions?, I showed...
From Daniel Lemire's Blog
Science and Technology links (March 30th, 2018)
People who score higher on intelligence tests tend to have larger brains. Twin studies suggest the same genetic factors influence both brain size and intelligence...
From Daniel Lemire's Blog
Should you cache hash values even for trivial classes?
Hash tables are a fundamental data structure in computing, used to implement maps and sets. In software, we use hash values to determine where objects are located...
From Daniel Lemire's Blog
When accessing hash tables, how much time is spent computing the hash functions?
Suppose that you create a large set of objects that you store in a hash table. Let us take 10 millions objects. It is large enough that it probably will not fit...
From Daniel Lemire's Blog
When shuffling large arrays, how much time can be attributed to random number generation?
It is well known that contemporary computers don’t like to randomly access data in an unpredictible manner in memory. However, not all forms of random accessesContinue...
From Daniel Lemire's Blog
Science and Technology links (March 23rd, 2018)
Sending your kids to highly selective schools is maybe less useful than you think: “However, once we controlled for factors involved in pupil selection, (…) the...
From Daniel Lemire's Blog
Science and Technology links (March 16th, 2018)
From the beginning of the 20th century to 2010, the life expectancy at birth for females in the United States increased by more than 32 years. The 3 major causes...
From Daniel Lemire's Blog
Iterating over hash sets quickly in Java
There are many ways in software to represent a set. The most common approach is to use a hash table. We define a “hash function” that takes as an input our elements...
From Daniel Lemire's Blog
Science and Technology links (March 9th, 2018)
The Audi A8, which goes on sale this year, will be the first car to offer Level 3 autonomy, which means that as a driver, you are expected to be able to relinquish...
From Daniel Lemire's Blog
Iterating over set bits quickly (SIMD edition)
Suppose that you have a long sequence of bits 10101011100000… you want to visit all the bits set to 1. That is, given 10101011100000, you would like to get theContinue...
From Daniel Lemire's Blog
Science and Technology links (March 2nd, 2018)
Flashing lights might cure Alzheimer’s, according to Nature. There is no paradox: being obese is definitively bad for you. Class attendance predicts success inContinue...
From Daniel Lemire's Blog
Vectorized shifts: are immediates faster?
Our processors are all equipped with vector instructions also called SIMD (single instruction multiple data). One common instruction is the “shift”. Roughly speaking...
From Daniel Lemire's Blog
Science and Technology links (February 24th, 2018)
Samsung is manufacturing and will sell 30TB hard drives. That’s huge. It is enough to record everything you see and hear for three years. Of the cases of early-onset...
From Daniel Lemire's Blog
Iterating over set bits quickly
A common problem in my line of work is to iterate over the set bits (bits having value 1) in a large array. My standard approach involves a “counting trailing zeroes...
From Daniel Lemire's Blog