GPU-Native Infrastructure for Real-Time AI

We build the primitives that autonomous systems depend on: sorting, caching, rate limiting, and retrieval. Engineered for GPUs from the ground up.

NVIDIA Inception Partner Tested on DGX Spark (Blackwell)
Validated Performance

MASH Sort benchmarked on NVIDIA Blackwell GB10. Speedup vs. standard GPU radix sort. Geometric mean across 100M–3B keys.

Presorted 8.6x
Reverse 4.3x
Uniform Random 1.4x
Zipfian (Heavy-Tail) 1.3x

On presorted 1B-row workloads (the kind you see in HFT, logging, and time-series) MASH is ~9x faster. At 7B elements, standard radix sort crashes. MASH keeps running.

Read "Sorting on Blackwell" →
NVIDIA
Inception Partner
CUDA
Native
DGX Spark
Tested
Engineering

The technical foundations behind Voxell's products.

View all engineering articles →
Get In Touch

Building systems where latency and consistency matter? We'd like to hear about your challenges.

24h reply • NDA ok • No IP needed