AMD Instinct MI300 vs AMD Radeon 8065S Comparison

The Verdict

The AMD Instinct MI300 and AMD Radeon 8065S serve fundamentally different compute roles, and the recorded data makes that separation clear. The Instinct MI300 is a data-center accelerator built around the CDNA 3.0 architecture, delivering 47.87 TFLOPS of FP32 throughput and 47.87 TFLOPS of FP16, paired with 128 GB of HBM3 memory across an 8192-bit bus. The Radeon 8065S is a mobile integrated graphics processor on the RDNA 3.5 architecture, producing 15.36 TFLOPS in both FP32 and FP16, with system-shared memory and a 55 W power envelope.

For compute-heavy workloads that demand large memory capacity and maximum floating-point throughput, the Instinct MI300 is the only choice supported by the data. Its 5.32 TB/s memory bandwidth and 1,496.0 GTexel/s texture rate dwarf the Radeon 8065S, which depends on system-dependent shared memory and delivers 480.0 GTexel/s. The Radeon 8065S, however, is the only one with display outputs, DirectX 12 Ultimate support, Vulkan 1.4, and OpenGL 4.6, making it suitable for graphics-oriented tasks in portable devices. The Instinct MI300 has no display outputs and reports N/A for all graphics APIs.

The percentile ranking for both is 50, with no benchmark scores recorded, so the database does not provide a direct performance comparison. Instead, the architectural specifications define the intended deployment: the MI300 for server racks with a 1000 W suggested power supply, the 8065S for integrated mobile systems that require no external power connectors. Users needing compute density and HBM3 memory should select the MI300; users needing a graphics-capable IGP with modern API support should select the 8065S.

Architecture Differences

The Instinct MI300 uses the Aqua Vanjaram chip on a 5 nm TSMC process, housing 153,000 million transistors on a 1017 mm² die. The transistor density calculates to 150.4M per mm². The Radeon 8065S uses the Gorgon Halo chip on a 4 nm TSMC process with a 308 mm² die size; its transistor count is unknown. The MI300 belongs to the Instinct (MIx) generation and the CDNA 3.0 architecture, while the 8065S belongs to the Navi Mobile (RX 8000M) generation and the RDNA 3.5 architecture.

The MI300 has 14,080 shading units, 880 texture mapping units, and zero raster operation units. It reports 0 MPixel/s pixel rate and no ray tracing cores. The 8065S has 2,560 shading units, 160 TMUs, 64 ROPs, and 40 ray tracing cores. Its pixel rate is 192.0 GPixel/s. The MI300 has no graphics API support: DirectX, OpenGL, and Vulkan all report N/A. The 8065S supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The MI300 is a PCIe 5.0 x16 card with 2x 8-pin power connectors and a 600 W TDP, requiring a 1000 W suggested power supply. It has no display outputs. The 8065S is an integrated graphics processor with a 55 W TDP, no power connectors, and display outputs that are portable device dependent. The MI300 measures 267 mm in length and 111 mm in height; the 8065S has no recorded dimensions. The MI300 uses HBM3 memory with 128 GB capacity, 8192-bit bus width, and 5.32 TB/s bandwidth. The 8065S uses system shared memory with system dependent bandwidth.

FAQ

Q: Which GPU has higher raw FP32 compute?

A: The Instinct MI300 delivers 47.87 TFLOPS FP32, which is 3.1 times the 15.36 TFLOPS of the Radeon 8065S.

Q: Can the Instinct MI300 output video to a display?

A: No. The MI300 reports no display outputs, while the Radeon 8065S provides display outputs that are portable device dependent.

Q: What memory configurations do these two parts use?

A: The MI300 uses 128 GB of HBM3 with an 8192-bit bus and 5.32 TB/s bandwidth. The 8065S uses system shared memory with system dependent bandwidth and bus width.

Q: Which part supports ray tracing?

A: Only the Radeon 8065S has ray tracing cores, specifically 40 of them. The MI300 has no ray tracing cores recorded.

Q: What are the power requirements for each?

A: The MI300 has a 600 W TDP with 2x 8-pin power connectors and a 1000 W suggested power supply. The 8065S has a 55 W TDP with no power connectors and no suggested power supply.

Q: Which part is smaller in die area?

A: The Radeon 8065S has a 308 mm² die, while the Instinct MI300 has a 1017 mm² die, making the MI300 more than three times larger.

Specification Differences

| Specification | AMD Instinct MI300 | AMD Radeon 8065S |

|---|---|---|

| Architecture | CDNA 3.0 | RDNA 3.5 |

| Generation | Instinct (MIx) | Navi Mobile (RX 8000M) |

| Process Node | 5 nm | 4 nm |

| Transistors | 153,000 million | unknown |

| Die Size | 1017 mm² | 308 mm² |

| Transistor Density | 150.4M / mm² | null |

| Base Clock | 1000 MHz | 1295 MHz |

| Boost Clock | 1700 MHz | 3000 MHz |

| Memory Clock | 1300 MHz 5.2 Gbps effective | System Shared |

| Memory Size | 128 GB | System Shared |

| Memory Type | HBM3 | System Shared |

| Memory Bus Width | 8192 bit | System Shared |

| Memory Bandwidth | 5.32 TB/s | System Dependent |

| Shading Units | 14080 | 2560 |

| TMUs | 880 | 160 |

| ROPs | 0 | 64 |

| Ray Tracing Cores | null | 40 |

| Pixel Rate | 0 MPixel/s | 192.0 GPixel/s |

| Texture Rate | 1,496.0 GTexel/s | 480.0 GTexel/s |

| FP32 | 47.87 TFLOPS | 15.36 TFLOPS |

| FP16 | 47.87 TFLOPS (1:1) | 15.36 TFLOPS (1:1) |

| TDP | 600 W | 55 W |

| Slot Width | null | IGP |

| Power Connectors | 2x 8-pin | None |

| Suggested PSU | 1000 W | null |

| Display Outputs | No outputs | Portable Device Dependent |

| DirectX | N/A | 12 Ultimate (12_2) |

| OpenGL | N/A | 4.6 |

| Vulkan | N/A | 1.4 |

| Dimensions | 267 mm x 111 mm | null |

| Production Status | null | Active |

| Release Date | 2023-01-03 | 2025-12-31 |

| Predecessor | Radeon Instinct | Polaris Mobile |

Head-to-Head Benchmarks

The database records no direct head-to-head benchmark scores for these two parts, and neither has an average benchmark score above zero. The comparison therefore rests entirely on the specification sheet. The most decisive numerical gaps appear in compute throughput and memory subsystem.

The Instinct MI300 leads by 32.51 TFLOPS in FP32 performance, reaching 47.87 TFLOPS versus 15.36 TFLOPS for the 8065S. The same margin applies to FP16, where both parts run at 1:1 ratio. This means the MI300 delivers 3.1 times the floating-point throughput of the 8065S in both precisions. For texture work, the MI300 produces 1,496.0 GTexel/s against 480.0 GTexel/s, a 1,016 GTexel/s advantage that reflects its 880 TMUs versus 160 TMUs.

The memory comparison shows an even larger structural difference. The MI300 provides 128 GB of HBM3 across an 8192-bit bus, reaching 5.32 TB/s. The 8065S has system shared memory with a system dependent bandwidth figure, meaning no fixed bandwidth number exists to compare directly. The 8192-bit bus width alone, however, indicates the MI300's memory subsystem is designed for bandwidth-intensive accelerator workloads, while the 8065S relies on the host system's memory architecture.

The Radeon 8065S wins in pixel throughput, producing 192.0 GPixel/s from its 64 ROPs, whereas the MI300 reports 0 MPixel/s and zero ROPs. The 8065S also has 40 ray tracing cores, which the MI300 lacks entirely. Clock speeds favor the 8065S as well: its base clock runs at 1295 MHz and boost reaches 3000 MHz, compared to the MI300's 1000 MHz base and 1700 MHz boost. The 8065S's boost clock is 1,300 MHz higher than the MI300's.

Power consumption differentiates the two sharply. The MI300 draws a 600 W TDP and needs a 1000 W suggested power supply, while the 8065S operates at 55 W TDP with no external power connectors. The MI300 is 267 mm long and 111 mm tall; the 8065S has no recorded dimensions and is classified as an IGP. The MI300 uses 2x 8-pin power connectors, the 8065S uses none.

The release dates place the MI300 in January 2023 and the 8065S at the end of December 2025. The MI300's predecessor is the Radeon Instinct, while the 8065S's predecessor is the Polaris Mobile. The 8065S is marked as active in production; the MI300's production status is not recorded. The MI300 has no successor recorded, and neither does the 8065S.

The manufacturing process differs by one nanometer: the MI300 uses TSMC 5 nm, the 8065S uses TSMC 4 nm. The MI300's transistor count of 153,000 million dwarfs the unknown transistor count of the 8065S, though the die size ratio of 1017 mm² to 308 mm² shows the MI300 is 3.3 times larger in area. The transistor density of the MI300 is 150.4M per mm², while no density figure exists for the 8065S.

Both parts use PCIe 5.0 x16 as the bus interface, so connectivity does not differentiate them. The API support, however, is exclusive to the 8065S: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 all report as N/A on the MI300. This makes the 8065S the only one of the two capable of running standard graphics software stacks, while the MI300 is confined to compute-centric workloads that do not require a display or graphics API.