AMD Radeon Instinct MI350P: specs and benchmarks
Contents
Summary
AMD started Radeon Instinct MI350P sales 7 May 2026. This is a CDNA 4.0 architecture desktop card based on 3 nm manufacturing process and primarily aimed at designers. 144 GB of HBM3e memory clocked at 2 GHz are supplied, and together with 8192 Bit memory interface this creates a bandwidth of 8.19 TB/s.
Compatibility-wise, this is a dual-slot graphics card attached via PCIe 5.0 x16 interface. Its manufacturer default version has a length of 267 mm. 1x 16-pin power connector is required, and power consumption is at 600 Watt.
Primary details
Some basic facts about Radeon Instinct MI350P: architecture, market segment, release date etc.
| Place in the ranking | not rated | |
| Place by popularity | not in top-100 | |
| Architecture | CDNA 4.0 (2025−2026) | |
| GPU code name | MI350 128CU | |
| Market segment | Workstation | |
| Release date | 7 May 2026 (less than a year ago) |
Detailed specifications
Radeon Instinct MI350P's specs such as number of shaders, GPU base clock, manufacturing process, texturing and calculation speed. These parameters indirectly speak of Radeon Instinct MI350P's performance, but for precise assessment you have to consider its benchmark and gaming test results.
| Pipelines / CUDA cores | 8192 | of 32768 (Radeon Instinct MI455X) |
| Core clock speed | 1000 MHz | of 2670 MHz (Arc B580) |
| Boost clock speed | 2200 MHz | of 3130 MHz (Radeon RX 9060 XT 16 GB) |
| Number of transistors | 73,000 million | of 320,000 million (Radeon Instinct MI455X) |
| Manufacturing process technology | 3 nm | of 2 nm (Radeon Instinct MI455X) |
| Power consumption (TDP) | 600 Watt | of 2400 Watt (Data Center GPU Max Subsystem) |
| Texture fill rate | 1,126.4 | of 2,553.6 (Radeon Instinct MI300X) |
| Floating-point processing power | 36.04 TFLOPS | of 157.3 (Radeon Instinct MI455X) |
| TMUs | 512 | of 1280 (Data Center GPU Max NEXT) |
| L1 Cache | 2 MB | of 64 MB (Data Center GPU Max 1550) |
| L2 Cache | 16 MB | of 408 MB (Data Center GPU Max 1350) |
| L3 Cache | 128 MB | of 256 MB (Radeon Instinct MI300X) |
Form factor & compatibility
This section provides details about the physical dimensions of Radeon Instinct MI350P and its compatibility with other computer components. This information is useful when selecting a computer configuration or upgrading an existing one. For desktop graphics cards, it includes details about the interface and bus (for motherboard compatibility) and additional power connectors (for power supply compatibility).
| Interface | PCIe 5.0 x16 | |
| Length | 267 mm | |
| Width | 2-slot | |
| Supplementary power connectors | 1x 16-pin |
VRAM capacity and type
Parameters of memory installed on Radeon Instinct MI350P: its type, size, bus, clock and resulting bandwidth. Note that GPUs integrated into processors have no dedicated memory and use a shared part of system RAM instead.
| Memory type | HBM3e | |
| Maximum RAM amount | 144 GB | of 432 GB (Radeon Instinct MI455X) |
| Memory bus width | 8192 Bit | of 24576 Bit (Radeon Instinct MI455X) |
| Memory clock speed | 2000 MHz | of 16000 GB/s (GeForce RTX 4050 Mobile) |
| Memory bandwidth | 8.19 TB/s | of 3,277 GB/s (Radeon Instinct MI250X) |
Connectivity and outputs
Types and number of video connectors present on Radeon Instinct MI350P. As a rule, this section is relevant only for desktop reference graphics cards, since for notebook ones the availability of certain video outputs depends on the laptop model, while non-reference desktop models can (though not necessarily will) bear a different set of video ports.
| Display Connectors | No outputs |
API and SDK support
APIs supported by Radeon Instinct MI350P, sometimes including their particular versions.
| DirectX | N/A | |
| Shader Model | N/A | |
| OpenGL | N/A | of 4.6 (GeForce RTX 5090 D) |
| OpenCL | 3.0 | |
| Vulkan | N/A |
NVIDIA equivalent
According to our data, the closest NVIDIA alternative to Radeon Instinct MI350P is RTX PRO 4000 Blackwell.
Similar GPUs
Here is our recommendation of several graphics cards that are more or less close in performance to the one reviewed.
