As organizations prepare for the transition to quantum-resistant cryptography, a Chinese standards organization has published a suite of technical guidelines designed as China’s benchmarking framework that will standardize how next-generation cryptographic algorithms are evaluated across software and hardware platforms.
Released recently by the Institute of Commercial Cryptography Standards (NICCS), the four documents establish common methodologies for benchmarking implementations on x86 processors, ARM processors, field-programmable gate arrays (FPGAs) and application-specific integrated circuits (ASICs). The guidelines cover key encapsulation mechanisms (KEMs), digital signature schemes, key exchange protocols and cryptographic hash algorithms, providing developers with a consistent framework for measuring correctness, performance and resource efficiency.
While the documents are highly technical, they reflect a broader trend in post-quantum cryptography. As governments and industry move beyond selecting algorithms, attention is increasingly shifting toward how those algorithms perform in real-world hardware and software environments.
The guidance also aligns with the Institute’s broader mission of advancing commercial cryptography standards in anticipation of the security challenges posed by quantum computing.
China’s Benchmarking Framework: Beyond Choosing Algorithms
Much of the discussion surrounding post-quantum cryptography has centered on which algorithms should replace RSA and elliptic curve cryptography. The U.S. National Institute of Standards and Technology has already standardized several post-quantum algorithms, and governments worldwide have begun planning large-scale migrations.
China’s benchmarking framework addresses a different stage of that process.
Rather than evaluating the mathematical security of algorithms, the documents focus on implementation quality. They establish standardized procedures for measuring whether an implementation is correct, reproducible and efficient regardless of the computing platform.
For software implementations on x86 and ARM architectures, the framework requires developers to submit multiple implementation versions, including a reference implementation emphasizing portability and correctness, a performance-optimized version, and a resource-optimized version designed for memory-constrained environments. Each submission must include source code, benchmarking tools, dependency information and a detailed self-assessment report.
The documents also recommend standardized software environments, including Linux, GCC and CMake, while specifying compiler options to improve reproducibility across evaluations. Reference implementations emphasize ISO C99 portability, while optimized versions may use architecture-specific instruction sets such as AVX2 on x86 or NEON and SVE on ARM to maximize throughput.
Measuring More Than Speed
There is an emphasis on standardized performance reporting rather than raw benchmark results.
Developers are instructed to evaluate implementations using Known Answer Tests (KATs) to verify correctness before reporting metrics including CPU clock cycles, throughput, static and peak memory usage, key sizes, ciphertext sizes and communication overhead. The goal is to ensure that reported results can be independently reproduced and compared across implementations.
For public-key algorithms, the framework measures operations such as key generation, encapsulation, decapsulation, digital signatures and shared-secret derivation. Hash algorithms are evaluated separately using throughput metrics.
The guidelines distinguish between three implementation philosophies.
Reference implementations prioritize readability and portability. Performance versions allow aggressive compiler optimization and architecture-specific instructions. Resource-optimized versions instead minimize memory usage and storage requirements for constrained systems.
Hardware Becomes Part of the Migration
Perhaps the most significant aspect of China’s benchmarking framework is that it extends well beyond conventional software benchmarking.
Separate documents establish equivalent evaluation frameworks for FPGA and ASIC implementations, recognizing that many commercial security products rely on dedicated cryptographic hardware rather than general-purpose processors.
The FPGA guidance recommends a standardized AMD Zynq-7000 development platform together with common synthesis, timing and power-analysis workflows using Xilinx Vivado. Beyond measuring throughput and latency, evaluations include hardware resource utilization, including lookup tables (LUTs), flip-flops, block RAM, digital signal processing units, power consumption and throughput-to-area ratios.
The ASIC guidance goes even further by defining normalized metrics intended to allow comparisons across semiconductor manufacturing processes. Developers are instructed to report synthesized area, operating frequency, power consumption and normalized throughput relative to equivalent NAND2 gate area, along with detailed information about process libraries, memory macros and synthesis environments.
These measurements are particularly important for hardware security modules, embedded controllers, networking equipment and specialized cryptographic accelerators, where silicon area, power consumption and throughput directly affect manufacturing cost and deployment feasibility.
Preparing for Deployment
The documents on China’s benchmarking framework suggest that Chinese standards efforts are increasingly focused on implementation readiness rather than solely on cryptographic theory.
Successful deployment of quantum-resistant cryptography requires algorithms that can operate efficiently across cloud servers, enterprise infrastructure, mobile devices, industrial controllers and dedicated security hardware. Those environments have very different performance constraints, making standardized benchmarking increasingly valuable.
The guidance also reflects growing recognition that reproducibility matters. Rather than allowing developers to report isolated performance figures, the framework requires documentation of build environments, optimization methods, compiler settings, dependencies and testing procedures so that results can be independently verified.
Although the documents do not mandate specific post-quantum algorithms, they establish the infrastructure needed to evaluate next-generation public-key cryptography consistently across multiple computing architectures. As governments and industry continue planning long-term migration away from today’s public-key systems, implementation benchmarking is becoming an increasingly important part of determining which algorithms are practical outside the laboratory.
For the quantum security community, the release illustrates that the global race toward quantum-resistant security now extends beyond cryptographic research and into the engineering disciplines needed to deploy those algorithms at scale.
