Skip to content
Tech News
← Back to articles

AMD contrasts MI430X native FP64 with Nvidia's FP64 emulation at SC26

read original get AMD Instinct MI300X Accelerator → more articles
GoKawiil Brief

At Supercomputing 2026, AMD promoted its Instinct MI430X accelerator's native double-precision (FP64) support, contrasting it with Nvidia's approach of emulating FP64 via techniques like Ozaki schemes on GPUs optimized for lower-precision AI math. AMD cited its own profiling data claiming that fewer than 10% of HPC workloads are dense matrix multiplications suited to such emulation, and said the method triples memory usage while poorly serving workloads like sparse linear algebra, stencils, and molecular dynamics.

Why It Matters

GoKawiil's interpretation of the reporting above, not reported fact.

The comparison highlights a strategic divergence between the two chipmakers: Nvidia has shifted die area toward AI-favored low-precision math as that market has grown, while AMD is positioning native FP64 hardware as a differentiator for scientific and national-lab customers who need it. AMD's framing suggests it sees an opening to court HPC buyers who may view Nvidia's emulation-based approach as a compromise rather than a true substitute for dedicated double-precision hardware.

Key Takeaways
Worth a Look

AMD Instinct MI300X Accelerator — If you're following AMD's push for native FP64 performance in HPC, the Instinct series accelerators represent that same architecture philosophy in AMD's current lineup. These GPUs are built for scientific computing and simulation workloads that demand genuine double-precision math rather than emulated approximations. Great for researchers and enthusiasts wanting to explore AMD's data center compute approach firsthand.

See AMD Instinct MI300X Accelerator on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.

Source: tomshardware.com — Jeffrey Kampman, 2026-11-17

Published there as: “AMD attacks Nvidia's FP64 emulation at SC26, touts native double-precision on Instinct MI430X — says Ozaki schemes fit less than 10% of HPC workloads and triple memory use and emulated FP64 is a poor fit for many HPC workloads”

Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.
Keep reading
nvidia 1118 briefs → amd 212 briefs → sc26 3 briefs →