Fundamentals 14 min read

From x86 to ARM: The Next Computing Pivot and Cross‑Platform Migration Practices with Huawei Kunpeng

The article reviews the historical shift from mainframes to x86, highlights the emerging limitations of x86 in the era of mobile and cloud workloads, and details how Huawei's ARM‑based Kunpeng platform and its migration toolchain enable efficient cross‑architecture software migration and performance optimization.

Architects' Tech Alliance
Architects' Tech Alliance
Architects' Tech Alliance
From x86 to ARM: The Next Computing Pivot and Cross‑Platform Migration Practices with Huawei Kunpeng

Historically, computing moved from expensive mainframes (e.g., IBM System/360) to affordable minicomputers (DEC PDP‑11) and then to Intel x86‑based PCs, making computers accessible to enterprises and individuals alike.

Today, almost all software runs on x86, but the rise of smart terminals, cloud gaming, HTTPS traffic, and distributed big‑data workloads exposes the architecture’s shortcomings, especially for mobile and high‑concurrency scenarios.

ARM’s smaller core size and higher core density make it attractive for these workloads; Huawei’s Kunpeng 920, an ARM‑based processor with up to 64 cores, 2.6 GHz frequency, 8 DDR4 channels, PCIe 4.0, and 100 Gb RoCE, demonstrates significant performance and efficiency gains over comparable x86 CPUs.

The Kunpeng ecosystem already supports mainstream web servers (Nginx, Apache, Tomcat), middleware (Memcached, Redis, Kafka), databases (MySQL, MariaDB, PostgreSQL), big‑data frameworks (Hadoop, Hive, HBase), and programming languages (Go, Java, C/C++, Python, Ruby, Perl, JavaScript), running on OSes such as openEuler, Ubuntu, CentOS, and UOS.

Cross‑platform software migration is divided into three categories:

Interpreted languages (Java, Python, etc.) – run unchanged on ARM64 if no native C/C++ dependencies exist.

Compiled languages (C, C++, Go) – require recompilation with an ARM64 toolchain; Huawei provides migration assistance when source code is unavailable.

Assembly code – usually needs to be rewritten or translated using future instruction‑set translation tools.

For C/C++ applications, migration steps include ensuring Linux‑based builds, removing x86‑specific assembly, and recompiling with ARM‑compatible GCC options (e.g., -march=armv8.1-a -mtune=tsv110 ).

Common migration challenges are incompatible compiler flags, missing libraries, unsupported third‑party packages, and inline assembly that must be rewritten for ARM.

Huawei’s Kunpeng developer suite offers four key tools:

Dependency Advisor : scans binaries and source to assess portability and estimate effort.

Porting Advisor : provides modification suggestions for build files, source code, and assembly, supporting both CLI and web interfaces.

Tuning Kit : collects performance metrics (CPU, memory, I/O, network) and offers optimization recommendations, including Java profiling.

Acceleration Libraries : optimized glibc and HMPP interfaces that improve compute‑intensive workloads, comparable to Intel’s IPP.

A typical migration workflow consists of four steps: (1) inventory and compatibility analysis of the software stack, (2) dependency analysis using Huawei’s migration tools, (3) detailed C/C++ porting recommendations, and (4) performance analysis and tuning.

Beyond technical support, Huawei promotes the Kunpeng ecosystem through university collaborations, publishing over 20 textbooks and 200 online courses, offering 24 professional certifications, and providing financial incentives (e.g., cloud coupons for startups) via the "沃土" (Wotu) program.

In summary, while moving from x86 to ARM (Kunpeng) requires careful planning and tool‑assisted porting, the platform delivers notable performance, energy‑efficiency, and a growing software ecosystem that empowers developers to embrace the next computing paradigm.

Cross-Platformx86ARMCPU architectureHuawei Kunpengsoftware migration
Architects' Tech Alliance
Written by

Architects' Tech Alliance

Sharing project experiences, insights into cutting-edge architectures, focusing on cloud computing, microservices, big data, hyper-convergence, storage, data protection, artificial intelligence, industry practices and solutions.

0 followers
Reader feedback

How this landed with the community

login Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.