Free tools Windows power users keep installed
One-click scans. No signup required.
The NVIDIA Grace Hopper Superchip is a single module that pairs an NVIDIA Grace CPU with an NVIDIA Hopper GPU. The two chips are linked by NVLink-C2C, a high-bandwidth, memory-coherent interconnect, so CPU and GPU code can work on a shared memory space. NVIDIA uses the term “superchip” for this building block. Larger products such as DGX GH200 and GH200 NVL2 are systems built from these superchips, not the superchip itself.
What the name means
“Grace” refers to NVIDIA’s Arm-based Grace CPU architecture, and “Hopper” refers to its Hopper GPU architecture. The superchip joins the two in one package. The feature that makes it a superchip rather than two separate processors is the NVLink-C2C link between them.
As an Amazon Associate I earn from qualifying purchases.
NVLink-C2C is a direct, memory-coherent connection. Under the supported programming model, CPU and GPU threads can access system-allocated memory without the developer copying data explicitly between two separate pools. NVIDIA presents this as a way to reduce data movement and to make a larger memory pool available to GPU workloads.
Coherent access does not mean the two memory types behave identically. CPU-attached memory and GPU-attached memory have different capacities and bandwidths, and the GPU’s HBM3 is faster than the CPU’s LPDDR5X. Code that is tuned for the GPU’s local memory still performs differently from code that reaches into CPU memory.
#1 Best Overall
- The world’s fastest gaming processor, built on AMD ‘Zen5’ technology and Next Gen 3D V-Cache.
- 8 cores and 16 threads, delivering +~16% IPC uplift and great power efficiency
- 96MB L3 cache with better thermal performance vs. previous gen and allowing higher clock speeds, up to 5.2GHz
- Drop-in ready for proven Socket AM5 infrastructure
- Cooler not included
Architecture and published maximum figures
NVIDIA’s Technical Blog architecture article, published around 2023, gives the following maximum values for a single Grace Hopper Superchip. These are the upper limits of the architecture. Specific products may ship with less.
| Component | Published maximum | Qualification |
|---|---|---|
| CPU cores | Up to 72 Arm Neoverse V2 cores | Grace CPU maximum per superchip |
| CPU memory | Up to 512 GB LPDDR5X | Grace CPU maximum per superchip |
| CPU memory bandwidth | Up to 546 GB/s | Grace CPU maximum per superchip |
| GPU memory | Up to 96 GB HBM3 | Hopper GPU maximum per superchip |
| GPU memory bandwidth | Up to 3000 GB/s | Hopper GPU maximum per superchip |
| NVLink-C2C bandwidth | Up to 900 GB/s total (450 GB/s in each direction) | Chip-to-chip link between CPU and GPU |
Read these as architecture ceilings. NVIDIA’s guidance is that the values depend on the configuration, so any product sheet or procurement document should be checked before quoting a figure for a specific machine.
Rank #2
- AMD Ryzen 9 9950X3D Gaming and Content Creation Processor
- Max. Boost Clock : Up to 5.7 GHz; Base Clock: 4.3 GHz
- Form Factor: Desktops , Boxed Processor
- Architecture: Zen 5; Former Codename: Granite Ridge AM5
Superchip versus system: DGX GH200 and GH200 NVL2
The word “GH200” is used for more than one thing, and the names are not interchangeable. The superchip is the module. DGX GH200 and GH200 NVL2 are system configurations that contain superchips, and they have different totals.
| Product name | What it contains | Memory figures stated by NVIDIA |
|---|---|---|
| Grace Hopper Superchip (GH200) | One Grace CPU and one Hopper GPU | Architecture maxima listed above: up to 512 GB LPDDR5X and up to 96 GB HBM3 |
| DGX GH200 system | Grace Hopper Superchips connected with the NVLink Switch System | 480 GB LPDDR5 CPU memory and 96 GB HBM3 per superchip in the described configuration |
| GH200 NVL2 | Two Grace CPUs and two Hopper GPUs | Capacities and bandwidths depend on configuration; exact figures not stated in the Grace Performance Tuning Guide excerpt reviewed |
The DGX GH200 figure of 480 GB is LPDDR5, not the LPDDR5X used in the architecture maximum of 512 GB. It describes that system’s configuration only, and it should not be read as a universal GH200 specification.
Rank #3
- Can deliver fast 100 plus FPS performance in the world's most popular games, discrete graphics card required
- 6 Cores and 12 processing threads, bundled with the AMD Wraith Stealth cooler
- 4.2 GHz Max Boost, unlocked for overclocking, 19 MB cache, DDR4-3200 support
- For the advanced Socket AM4 platform
When you quote specifications, name the exact variant. A figure for a DGX GH200 node or an NVL2 server does not describe a standalone superchip, and the reverse is also true.
Intended workloads
NVIDIA positions Grace Hopper for accelerated AI and high-performance computing. For GH200 NVL2, NVIDIA lists these target workloads: single-node large language model inference, retrieval-augmented generation, recommender systems, graph neural networks, HPC, and data processing. These are vendor-described targets. They indicate where the platform is designed to be used, not measured performance for a given application.
Rank #4
- Pure gaming performance with smooth 100+ FPS in the world's most popular games
- 6 Cores and 12 processing threads, based on AMD "Zen 5" architecture
- 5.4 GHz Max Boost, unlocked for overclocking, 38 MB cache, DDR5-5600 support
- For the state-of-the-art Socket AM5 platform, can support PCIe 5.0 on select motherboards
- Cooler not included
How to compare Grace Hopper configurations
If you are evaluating systems rather than reading a definition, compare them on the same five points:
Recommended Free Tools
- The number of Grace CPUs and Hopper GPUs in the system, which separates a single superchip from NVL2 or DGX-class configurations.
- LPDDR5 or LPDDR5X capacity and bandwidth on the CPU side.
- HBM capacity and bandwidth on the GPU side.
- The interconnect and multi-GPU topology, including whether the NVLink Switch System is part of the design.
- Whether the target workload runs on one node or is scaled across several.
Confirm each value in the official system datasheet for the exact model you are considering. Architecture maxima and system configurations change across products and generations.
Best Value
- Processor provides dependable and fast execution of tasks with maximum efficiency.Graphics Frequency : 2200 MHZ.Number of CPU Cores : 8. Maximum Operating Temperature (Tjmax) : 89°C.
- Ryzen 7 product line processor for better usability and increased efficiency
- 5 nm process technology for reliable performance with maximum productivity
- Octa-core (8 Core) processor core allows multitasking with great reliability and fast processing speed
- 8 MB L2 plus 96 MB L3 cache memory provides excellent hit rate in short access time enabling improved system performance
Official description
NVIDIA’s Technical Blog describes the architecture this way: the Grace Hopper Superchip “brings together the groundbreaking performance of the NVIDIA Hopper GPU with the versatility of the NVIDIA Grace CPU, connected with a high bandwidth and memory coherent NVIDIA NVLink Chip-2-Chip (C2C) interconnect in a single superchip, and support for the new NVIDIA NVLink Switch System.” The source does not name an individual author for that sentence, so it should be attributed to NVIDIA Technical Blog.
In short, a Grace Hopper Superchip is one CPU, one GPU, and a coherent link between them, packaged as a building block. Everything larger than that is a system built from it, and its specifications need to be read against that system’s own documentation.
Quick Recap
“
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
