The Solution for stream encryption


We offer a unique solution - MALT-C, which is C/C++ programmed chip with the dissipation power of an embedded microprocessor. The MALT-C’s performance-per-watt is as high as FPGA’s used on all popular stream ciphers. More than that, the value of MALT-C is competitive with the value of a universal processor! There hadn’t been a microprocessor of such characteristics designed before!


Designed on MALT-C’s processor, the cryptographic accelerators are intended to solve the tasks of symmetric encryption including stream ciphers. The most preferable processing is multi-channel. Our processors are designed on the trusted local architecture. At the end of 2017 we got our first MALT prototypes in silicon from the factory which were successfully tested.


Look at the characteristics: the basic chip - 160-core MALT-C with an index of 7Mb160PLV, with frequency of 1 GHz and peak power consumption of 5W provides the maximum throughput of 195 kbps in the AES128 encryption task (ECB mode, not including the generation of the key).Try MALT! Apply MALT-C instead of foreign FPGA and cryptographic coprocessors in your projects! Get the detailed comparison in the table below.


The table The comparison of the throughput of MALT-C 7Mb160PLV with CPU, GPU and specialized integrated systems on various encryption algorithms. The data doesn’t include the generation of the key, the block ciphers are used on stream encryption mode.

Algorithm 7Mb160PLV
CPU, Gb/s GPU, Gb/s ASIC, Gb/s
AES 128 195 Семейство Intel Xeon E5v2 (3)
NVIDIA GTX 1080(4)
RC4 160     Result of the modelling (5)
Salsa20 444   NVIDIA GeForce GTX 590(6)
Chacha20 364 Intel Core i7-6700(7)



1) AES-GCM Encryption Performance on Intel® Xeon® E5 v3 Processors.

2) High Performance CUDA AES Implementation: A Quantitative Performance Analysis Approach.

3) High-Performance Hardware Implementation for RC4 Stream Cipher.

4) New Speed Records for Salsa20 Stream Cipher Using an Autotuning Framework on GPUs.

5) AES-NI SSL Performance.

6) ChaCha-IP-13 / EIP-13 ChaCha20 accelerators.

Get more information about these and other sources in the “Articles” section.




The Solution based on MALT. The stream encryption accelerator


One of the examples of MALT hardware implementation is a stream encryption accelerator for PCI bus. It is intended for the on-the-fly encryption of the data, which transmitted by one of three directions: PCIe -> 10G Ethernet, 10G Ethernet -> PCIe or PCIe -> PCIe, therefore, it can be represented as a PCIe bus network encoder/decoder or as a PCIe bus stream encryption accelerator. The module represents a half-height x4 PCIe card. The data flow processing is implemented by MALT-C cryptoprocessor, equipped with its own RAM, ROM, a few 10G interfaces and essential auxiliary ports (UART, JTAG, I2C, 1G Ethernet) for setup and management. An interface FPGA is used for the operation on PCI Express Gen2/3 bus.


Parameter Characteristic
Input/output interfaces PCIe Gen1/2/3 x4, 1000BASE-T, 10GBASE-T
Throughput capacity of the data over the network Up to 10 Gbps on any algorithm without linkage
Data encryption algorithm AES 128, RC4, Salsa20, Chacha20, etc.
Power consumption, maximum Up to 20W in operation, 0.25W in standby mode
Scalability Possible connection of multiple boards to the host
Programming Fully programmable device in C language




An application based on MALT. Standalone network encoder


Another example is a standalone encoder. The equipment represents a standalone device for stream encryption on the Ethernet. There are two independent 10G Ethernet ports (the encryption carries out between them) used for connection to trusted and open network. The data flow processing is implemented by a MALT-C cryptoprocessor, equipped with its own RAM, ROM, two 10G interfaces and essential auxiliary ports (COM, JTAG, i2c, 1G Ethernet) for setup and management.


Parameter Characteristic
Purpose «10G Ethernet - 10G Ethernet» encryption
Throughput capacity of the data over the network Up to 10 Gbps on any algorithm without linkage
Data encryption algorithm AES 128, RC4, Salsa20, Chacha20, etc.
Power consumption, maximum Up to 15W in operation, 0.2W in standby mode
Programming Fully programmable device in C language




The trusted blockchain solution


Blockchain is an approach, ensuring the data reliability kept on decentralized peer-to-peer network. No matter what operation is applied - transferring funds in Bitcoin, Ethereum contract execution or blockchain transaction - it always needs rapidity and power efficiency.


MALT-C processor family specifically developed for implementation of computationally complex cryptographical transformations, in particular, blockchain transactions with maximum of energy efficiency. By means of our processor you can provide trusted digital connection where the information and the data are successfully kept and transmitted with the use of blockchain technology and cryptographic tools. MALT-C processor family is low-power, so you can create a compact blockchain terminals, including embedded and portable ones. The equipment based on MALT architecture is guaranteed to be protected from software and hardware backdoors - you control both hardware and software! Who else in the world can make such a great offer to a domestic customer? By applying your own cryptographic algorithm on MALT-C, you can be sure that no one could hack the hash function by direct enumeration using the application-specific integrated circuit (ASIC). This way your blockchain system is securely protected. You can implement a full node on our ASIC.




Graphics card alternative


Nowadays graphics processors or GPU often find their use for the tasks which are not related to the computer-game visualization. It is connected with exceptional computing power of modern GPU containing thousands of computing elements, with advanced tools of GPU programming, and, what more important, with low cost. Nevertheless GPUs come as close as possible to universal processors, its essence is in graphics processing of computer games, because its architecture takes the whole potential and real productivity approaches to the highest level. As for other tasks, such as blockchain, big data and mathematical physics, their functionality is, on the one hand, excessive, because there is no floating-point operation in blockchain applications, on the other hand, insufficient, since the limited support for branching and random addressing in shared memory may reduce the performance by one or two orders of magnitude relative to theoretical limit in the ‘big data’ tasks.


There is a specialized solution: what if we create a chip, 80% of which will be composed of one-type computational elements designed ‘from scratch’ for a specified target class of tasks? What if we unite the computing elements into groups conducted by compact universal RISC processors united into computational array that to a programmer will look as a typical multi-threaded processor programmed on C/C++? We’ll get an extremely productive easy-to-use specialized processor! We designed three families of such processors and named the project ‘MALT’. Get more information about the architecture in the corresponding section.