[ 00 ] a new runtime for the k family
l is a new runtime for k4, q, and qSQL code. It is a fast database and array language interpreter most commonly used on Wall Street. l has compressed vectors, SIMD, and automatic parallelism by default and runs existing k4/q/qSQL without modification.
the q/k/qSQL you already have, just faster
l>sum 1 2 3 4 5 6 21modern HPC runtime
0.02 ms
SIMD reduction · scalar syntax
[ 01 ] what is l
The K and Q languages are array programming languages that offer an elegant syntax for high performance operations. In K/Q arrays are the unit of thought. l keeps those languages unchanged but adds transparent compressed vectors as the unit of execution: primitives run directly compressed structures so the full array is never rebuilt. Modern hardware starves for bandwidth long before it runs short of compute — so moving fewer bytes helps make L fast.
[ 01A ] compatible
run the syntax, idioms, and K4/q code you already have. l provides full compatibilty for tables, dicts , partitions and splays out of the box.
[ 01B ] compressed
data stays compressed at rest, in memory, and in transit. primitives run directly on the compressed vectors, so less data moves.
[ 01C ] transparent
the runtime picks the execution path—scalar, SIMD, threaded, or
offloaded—with no annotations. you write sum x; l does
the rest.
[ 02A ] in practice
Fully qSQL compatible database including all joins and partition logic. Same legendary speed.
// 86.5M-row analytics table - single CPUl> count AN_pageView86585107jl> select sum isBot, count i, avg isBot by date.month from AN_pageViewmonth isBot cnt botRate
2024.05 29404 1634214 0.01799275
2024.06 49596 1495228 0.03316952
2024.07 194442 2812446 0.06913626single CPU · no decode pass
86.5M rows
[ 02B ] the same primitives, faster
L implements automatic SIMD and threading for operations on vectors. L intelligently threads when vector lengths demand it - small vectors are SIMD , large are SIMD+threaded.
l> avg 1 2 3 4 5 63.5// -> fsum (SIMD reduction) -> / countl> avg 1 2 3 4 ... 999999// worker0: sum x[0..hi0]// worker1: sum x[hi0..hi1]// worker2: sum x[hi1..hi2]// ...// sum = s0 + s1 + s2 + ...// avg = sum / n499999.5l> remoteLinux:hopen `:10.0.0.1:8080connectedSIMD reduction · thread fan-out · remote execution
one expression
[ 03 ] anonymous mail group
in the spirit of classic technical mailing lists: notes, benchmarks, design questions, release logs. no profiles, no directories.
join with email only