IBM POWER8, POWER9 & POWER10 Software Development & Deployment

Elyan Labs ports, builds, optimizes, and deploys software on IBM Power servers (ppc64le / Linux on Power) — one of the few independent shops that runs Power hardware in-house and contributes to the PowerPC toolchain upstream. Remote, worldwide.

The wedge: almost nobody runs big-endian / ppc64 / POWER hardware in-house anymore — so flagship projects carry real bugs on those targets that no active contributor can reproduce. We own the hardware and the skill, so we walk in with a fix they can’t get anywhere else. That’s how a first-time contributor lands patches in OpenSSL, HACL*, Falco, and LLVM. For perspective: fewer than 90 people in history have ever contributed to HACL* (the verified crypto inside Firefox, Python, and the Linux kernel) — we’re one of them.

Start a Power project See the proof

What we do on IBM Power

Whether you're modernizing on Power, adding ppc64le to your support matrix, or squeezing more out of the hardware you already own — we work at the level the job needs, from a single build fix to a full port and deployment.

Port x86 → ppc64le

Bring x86/amd64 software to IBM Power: build-system changes, dependency porting, and validation on real POWER hardware.

Big-endian & portability fixes

Track down endianness bugs, alignment issues, and architecture assumptions that break on Power — the hard-to-find class of bug.

POWER performance tuning

Hand-optimization with VSX/AltiVec and POWER10 MMA, cache and prefetch tuning, NUMA-aware threading. We made llama.cpp ~9× faster on POWER8.

Build, CI & packaging

Reproducible Linux-on-Power builds, ppc64le CI pipelines, and packaging so Power stays a first-class target, not an afterthought.

Deployment to Power servers

Get code running and supported on your IBM Power systems — build, deploy, and harden, on your hardware or ours.

AI / LLM inference on Power

Run modern AI workloads on POWER — we built a sovereign LLM inference stack on POWER8 with custom VSX kernels.

Why us — proof, not promises

Most consultancies have never touched a Power server. We run one, ship for it, and improve the toolchain everyone else depends on:

Who this is for

Enterprises on IBM Power

Running AIX, IBM i, or Linux on Power and need modern software built, ported, or deployed on it.

ISVs & open-source projects

Need ppc64le added to your build/support matrix, or a stubborn Power-only bug fixed and upstreamed.

AI / HPC teams

Want to run inference or compute workloads on POWER and need the kernels and tuning to make it fast.

Migration projects

Moving to (or off) Power and need the porting, validation, and deployment handled by someone who's done it.

Common questions

Do you handle POWER9 and POWER10, or just POWER8?

All three. We run POWER8 in-house and our upstream LLVM work targets the POWER10 ISA (Power ISA v3.1); the same ppc64le toolchain and techniques apply across POWER8/9/10.

Can you port our x86 app to ppc64le?

Yes — that's a core service: build-system changes, big-endian/SIMD fixes, CI, and validation on real Power hardware.

Do you work on our Power servers or yours?

Either. We can build and validate on our own POWER8, or connect to your Power systems for build and deployment.

Where are you located?

Lake Charles, Louisiana — but IBM Power work is remote and we take clients worldwide.

Have a Power workload? Let's scope it.

From a single ppc64le build fix to a full port-and-deploy — the first conversation is free.

Email Scott — IBM Power project