The Oliver blog

Notes from the engine room.

What the team building Oliver measured, and what we learned from it.

Subscribe by RSS

Two indexes over the same log data drawn one tile per megabyte: 128 tiles for the existing index and 37 for the new one, which finds a common word 714 times faster.714×faster on a common word7 µsto find it, in under a third of the spaceexisting index · 128 MBnew index · 36.9 MBone tile = 1 MB of index · same log data
Search

Full-text search is now up to 714× faster.

We built a new full-text index for Oliver. It finds a common word in 7 microseconds, takes up to 3.5 times less space, builds nearly five times faster, and returns exactly the same answers as our existing index.

7 October 20262 min read
Achilles in full armour sprinting across a Greek hillside toward a tortoise, drawn as a row of successive strides that close the gap one step at a time.
Frontier

Oliver: a Frontier Database™.

Frontier models run on GPUs, and now the database they ask does too. On one GPU, Oliver answers over 6,000 questions a second across 100 million rows, up to 2,529 times faster than asking them one at a time.

29 September 20264 min read
244 query shapes, one tile each, shaded by how much faster Oliver answered: a 3.06 times median, the brightest 290 times faster.290×one user, 50 million rows3.06×median speed-up1.1×60×+under 1.1×Oliver vs ClickHouse 26.3 · 16 Sep 2026
Benchmarks

Oliver beats ClickHouse.

Across 244 query shapes on identical servers, Oliver was 3.06 times faster than ClickHouse at the median, and it ran all of them in less time than ClickHouse took to load the data.

27 September 20263 min read