Pek HPC Ansys

Ansys HPC Packs - Parallel and GPU Computing Licences

Every Ansys solver licence includes four cores, and your hardware has far more. Ansys HPC Packs unlock the rest, scaling a single simulation from 4 cores to more than 2,000 across CPUs and GPUs, on lease or paid-up terms.

Because the scaling is exponential, where packs are deployed matters as much as how many you own, so we size them to how your team actually solves.

Ansys HPC Packs and GPUs in 2026 R1

• Penyelesai GPU yang lancar : isipadu bendalir (VOF) dengan tenaga dan jaringan gelongsor kini dijalankan pada GPU, untuk simulasi berbilang fasa yang lebih pantas seperti pencampuran dan pemutaran jentera.
• HFSS GPU frequency sweeps (beta): GPU-solved frequency sweeps with cuDSS, distributed across multiple nodes for larger models.
• Polyflow on multiple GPUs (beta): Polyflow can now run with several MPI processes and multiple GPUs on a cluster.

Read more: Ansys 2026 R1 release highlights

How HPC Pack Scaling Works

Each pack multiplies parallel capacity rather than adding to it: one pack unlocks 8 cores, two unlock 32, three 128, four 512 and five 2,048, on top of the four included cores. Capacity beyond the entitlement sits idle, so two packs on a 48-core machine solve on 36 cores.

The same four packs deliver very different results depending on deployment: one pack on each of four machines gives 48 cores in total, while all four on one machine give 516 cores on a single solve. Consolidate onto a queued node when a few large models dominate; split across machines when several engineers solve at once. A pack cannot be split between users or analyses, and packs can run concurrent parametric variations through Workbench design points, Electronics Desktop and CFX operating maps.

HPC Packs for GPUs

Packs license GPU compute on the same entitlement, counting streaming multiprocessors (SMs) on NVIDIA hardware, or compute units on AMD, in place of CPU cores; an Ansys CFD Enterprise licence includes 40 SMs before any pack is needed. SMs are fixed in silicon and a Fluent case occupies the whole card, so a GPU that crosses a bracket boundary forces the next pack tier, which makes card selection a licensing decision as much as a hardware one.

Compatible Products

HPC Packs draw from one shared pool across the Ansys solver portfolio, including Fluent, CFX, Polyflow, Mechanical, Autodyn, Motion, Aqwa, HFSS, Maxwell, Q3D Extractor, Icepak, SIwave, Speos and Lumerical FDTD. Under HPC licensing, GPU acceleration is available in Fluent, Mechanical, HFSS transient, Polyflow and Lumerical FDTD. LS-DYNA, Rocky and EnSight use their own HPC licences.

Common HPC Licensing Consultation Topics

• Pack sizing: matching packs to model sizes and the number of concurrent solves.
• GPU selection: choosing cards with the licence brackets in mind.
• Cluster deployment: consolidating packs onto shared nodes.
• Benchmarking: measuring speed-up on representative models before purchase.

Ada projek di luar senarai ini? Hantarkan spesifikasi kepada kami. Sekiranya sesuatu kajian memerlukan pengetahuan pakar, kami menggunakan rangkaian rakan kongsi Ansys kami yang lebih luas.

Produk Berkaitan

• Ansys Fluent: the most common GPU use for HPC Packs.
• Ansys CFD: Enterprise includes 40 GPU SMs before packs are needed.
• Ansys Gateway powered by AWS: cloud hardware for burst capacity.
• Ansys TECS Reinstatement and Trade-Ins: consolidating piecemeal HPC licences when cover is renewed.

Kemas kini Okt '26 oleh Pasukan CADIT Ansys.