[Hardware] Startup claims its Zeus GPU is 10X faster than Nvidia's RTX 5090: Bolt's first GPU coming in 2026

[Drennn]

a GPU startup from Sunnyvale, California, this week introduced its Zeus GPU platform that is designed for gaming, rendering, and supercomputer simulations. The company says that its Zeus GPU not only supports features like upgradeable memory and built-in Ethernet interfaces, but it can also beat Nvidia's GeForce RTX 5090 by around 10 times in path tracing workloads, according to slide published by ServeTheHome. There is one major catch: Zeus can only beat the RTX 5090 GPU in path tracing and FP64 compute workloads because it does not support traditional rendering techniques. This means it has little of no chance to become one of the

RISC-V multi-chiplet GPU

Unlike GPUs from AMD, Intel, and Nvidia that rely on proprietary instruction set architectures, Bolt's Zeus relies on the open-source RISC-V ISA, according to slides published by ServeTheHome. The Zeus core relies on an open-source out-of-order general-purpose RVA23 scalar core mated with FP64 ALUs and the RVV 1.0 (RISC-V Vector Extension Version 1.0) that can handle 8-bit, 16-bit, 32-bit, and 64-bit data types as well as Bolt's additional proprietary extensions designed for acceleration of scientific workloads.

As Zeus is aimed at path tracing rendering technique as well as compute workloads, it does not seem to have traditional fixed-function GPU hardware like texture units (TMUs) and raster operation units (ROPs), so it has to rely on compute shaders (or similar methods) for texture sampling and graphics outputs. This saves precious silicon real estate for compute elements. Nonetheless, each Zeus GPU has one DisplayPort 2.1a and one HDMI 2.1b output.

Like many processors these days, Zeus relies on a multi-chiplet design. The entry-level Zeus 1c26-032 features a single processing unit that is equipped with 32 GB of LPDDR5X memory at 273 GB/s and can be mated with up to 128 GB of DDR5 memory using two SO-DIMMs (at 80 GB/s). The Zeus GPU is also paired with an I/O chiplet that packs a QSFP-DD port for 400GbE/800GbE, two PCIe Gen5 x16 slots with CXL 3.0 on top (enabling efficient memory sharing across multiple cards), and a GbE port for BMC. The GPU chiplet connects to its I/O chiplet at 256 GB/s.

The more advanced Zeus 2c26-064/128 uses two Zeus processing units, an I/O chiplet, and supports either 64 GB or 128 GB of LPDDR5X memory. The most powerful — Zeus 4c26-256 — implementation integrates four processing units, four I/O chiplets, 256 GB LPDDR5X and up to 2 TB of DDR5 memory. The quad-chiplet Zeus implementation is not a card, but rather is a server.

Unlike high-end GPUs that prioritize bandwidth, Bolt is evidently focusing on greater memory size to handle larger datasets for rendering or simulations. Also, built-in 400GbE and 800GbE ports to enabling faster data transfer across networked GPUs indicates datacenter focus of Zeus.

Link: https://www.tomshardware.com/pc-components/gpus/startup-claims-its-zeus-gpu-is-10x-faster-than-nvidias-rtx-5090-bolts-first-gpu-coming-in-2026