DSUMO for Supercomputing Users

What is DSUMO?

DSUMO, or Distributed Supercomputing Using Microcompute Operations, is a method of massively parallel computing using small functional operations on anywhere from one to millions of compute nodes.

Microcompute operations are performed in JavaScript within the DSUMO Framework on the client web browsers of of DSUMO-enabled sites. The compute operations are considered "micro" because they don't heavily tax the client's CPU, and are individually non-performant. The speed and power of DSUMO comes from combining millions of minimally-used client CPUs towards completing a single task.

All code written for execution on the DSUMO platform is in standard, web-compliant JavaScript. JavaScript was chosen for its ubiquity, both within the web browsers of compute nodes (the client web browsers) and in the skillset of developers.

DSUMO is easy to use: Code can be added to projects within your account via either the DSUMO website, or through a flexible API that supports multiple request/response content types. Your code is then executed by making a request to the API with results returned asyncronously, either by pushing it to your servers or by polling for it from ours.


DSUMO is meant to satisfy the needs of companies and individuals requiring access to vast computational resources without the imposition of the significant expense and time creating their own computing cluster or renting time on a "typical" supercomputing cluster.

The inspiration for DSUMO came from observing the CPU utilization of a web browser as it loaded and displayed advertisements that contained dynamic content. With over a billion personal computers in use worldwide, even a small amount of load on the CPU of each one equates to massive computational power. Rather than allocating those CPU cycles to the display of ads, DSUMO aims to put them to better use.

The Cost

Setting up DSUMO Computing is free, and there is no recurring fixed cost: You pay only for what you use. You can test out the DSUMO Framework and API for free within our development sandbox. Pricing per microcompute operation is demand-based, with an associated fixed fee for bandwidth utilizaiton.

Completed MO1 GB Transferred2
$5 per 10,000 MOs
(minimum bid)
$0.02 per GB

1 A completed microcompute operation is one that has run to completion and for which a result has been stored.

2 GB (gigabyte = 1,000,000,000 bytes) transferred represents the sum total of data submitted to and returned from each microcompute operation. See the DSUMO Computing documentation for more details.

Price Comparison

It is understandably difficult to compare the cost of DSUMO computing with traditional methods of grid and distributed computing. JavaScript is less efficient than native code, processors available via DSUMO are neither homogenous nor fully utilized, and inter-node communication is relatively slow. DSUMO works because it leverages the absolutely vast number of processors of the web browsing public, and no other computing platform does this.

Your success with the DSUMO platform will depend both upon your specific application and your implementation. We would be happy to assist you with optimizing a solution to run on the DSUMO platform.

Giving Back

KisoLabs is committed to providing non-profit, academic research and development projects with needed computational power through DSUMO. A portion (currently 15% or more) of our total computational capacity will be donated to such projects.

If you represent such a project and are interested in partnering with KisoLabs to utilize the power of DSUMO, please contact us and let us know a little bit about yourself, your project, and its goals.

Adding DSUMO to Your KisoLabs Account

At this time, KisoLabs is in the process of beta testing the DSUMO platform with acedemic research partners at no cost to compute users. If you represent a university, research laboratory, or non-profit organization and would be interested in participating in our beta program, please email dsumo@kisolabs.com with details on your organization and your anticipated computing needs.

© KisoLabs 2017