Chip Foundry Services · Live Lab← Back to site

This page is not a simulation. When you opened it, a real matrix multiplication started on a real machine in the CFS lab. Everything below streams live from that machine.

Connecting…

Live data path

What is running

Machine

GPU
…
Host
…
CPU
…
Cores / RAM
…
OS
…
NVIDIA driver
…

Application

Name
CFS MatMul Demo
Stack
Python 3 + numpy
Kernel
…
Backend
…
Build
…

Workload

Operation
C = A × B
Matrix size
1024 × 1024
Precision
float32
Work
2.147 GFLOP per multiply
Input
random, seed 2600
Repeats
5 timed runs

The matrices (8×8 block averages of the 1024×1024 data)

A
×
B
=
C (computed on the machine)

Result

–
Best multiply time
–
GFLOPS
–
Verified vs CPU float64
–
Speedup vs CPU

Throughput

Each timed run (ms)

Live JSON flow

Demonstration only · a fixed built-in benchmark runs on real Chip Foundry Services lab hardware · no user code is executed.