AMD Instinct MI350P vs AMD Radeon 880M Comparison

AMD Instinct MI350P and AMD Radeon 880M occupy entirely different segments of the GPU spectrum. The MI350P is a dual-slot accelerator built for compute density with 144 GB of HBM3e memory, while the 880M is an integrated graphics processor within a mobile chip. The database records no overlapping benchmark scores for the two, so direct head-to-head measurements are unavailable. Instead, the analysis compares their recorded specifications, architectural designs, and the 880M’s measured performance against its nearest rivals.

Head-to-Head Benchmarks

The database contains no shared benchmark entries between the AMD Instinct MI350P and the AMD Radeon 880M. The MI350P has an empty benchmark array and a percentile ranking of 50 against all GPUs, with an average benchmark score of 0. The 880M, by contrast, has ten recorded benchmark results across several suites.

The 880M’s strongest recorded result is in Geekbench Vulkan, where it scores 40006. Its Geekbench OpenCL score is 31285. In Passmark tests, the 880M scores 7615 in G3D, 3719 in GPU Compute, 969 in G2D, 97 in DirectX 9, 73 in DirectX 11, 32 in DirectX 12, and 31 in DirectX 10. The 3DMark Steel Nomad DX12 test yields a score of 535.

The 880M’s average benchmark score across all recorded tests is 8436, placing it in the 43rd percentile of all GPUs in the database. Its nearest rivals, based on average scores, are the NVIDIA GeForce GTX 675MX with an average score of 8427 (a delta of 0.1 percent), the NVIDIA GeForce MX330 at 8458 (negative 0.3 percent), the AMD Radeon HD 8870M at 8462 (negative 0.3 percent), and the AMD Radeon R9 M375X at 8325 (positive 1.3 percent). The data shows the 880M sits within a tight cluster of older mobile discrete GPUs, with deltas under 1.5 percent in either direction.

The MI350P’s theoretical compute figures are substantially higher. Its FP32 throughput is 36.04 TFLOPS, and its FP16 throughput is also 36.04 TFLOPS with a 1:1 ratio. The 880M delivers 4.454 TFLOPS in both FP32 and FP16. The MI350P’s texture rate is 1,126.4 GTexel/s compared to the 880M’s 139.2 GTexel/s. The 880M records a pixel rate of 46.40 GPixel/s, while the MI350P lists 0 MPixel/s, reflecting its lack of display outputs and rasterization focus.

The MI350P’s memory subsystem is the clearest differentiator. It uses 144 GB of HBM3e on an 8192-bit bus, delivering 8.19 TB/s of bandwidth. The 880M relies on system shared memory, with bandwidth described as system dependent. The MI350P’s memory clocks run at 2000 MHz with 8 Gbps effective, while the 880M has no dedicated memory clock.

FAQ

Q: Which GPU has a higher FP32 compute throughput?

A: The AMD Instinct MI350P records 36.04 TFLOPS in FP32, while the AMD Radeon 880M records 4.454 TFLOPS. The MI350P’s FP32 figure is roughly eight times higher.

Q: What is the average benchmark score for the Radeon 880M, and how does it compare to its closest rivals?

A: The 880M’s average benchmark score is 8436. Its nearest rival, the NVIDIA GeForce GTX 675MX, scores 8427, a difference of 0.1 percent. The NVIDIA GeForce MX330 scores 8458 (negative 0.3 percent), the AMD Radeon HD 8870M scores 8462 (negative 0.3 percent), and the AMD Radeon R9 M375X scores 8325 (positive 1.3 percent).

Q: Does the Instinct MI350P have any display outputs?

A: No. The MI350P lists “No outputs” as its display configuration, and its pixel rate is recorded as 0 MPixel/s. It is designed as a compute accelerator without rasterization output capability.

Q: What memory configuration does the Radeon 880M use?

A: The Radeon 880M uses system shared memory. Its memory size, type, and bus width are all listed as system shared, and its bandwidth is described as system dependent. It has no dedicated VRAM.

Q: What are the recorded benchmark scores for the 880M in DirectX tests?

A: In Passmark tests, the 880M scores 97 in DirectX 9, 73 in DirectX 11, 32 in DirectX 12, and 31 in DirectX 10. In 3DMark Steel Nomad DX12, it scores 535.

Q: What is the transistor density of each GPU?

A: The MI350P has a transistor density of 61.3 million transistors per square millimeter, while the Radeon 880M has a density of 145.9 million transistors per square millimeter. The 880M’s density is higher despite having fewer total transistors.

The Verdict

The data indicates two distinct use cases. The AMD Instinct MI350P is an accelerator with no display outputs, no recorded benchmarks, and no rasterization hardware. Its design centers on massive memory capacity (144 GB HBM3e), extreme bandwidth (8.19 TB/s), and high FP32/FP16 throughput (36.04 TFLOPS). It targets compute workloads where memory capacity and bandwidth dominate, such as large model inference or data processing. The 880M, conversely, is an integrated GPU for mobile systems, with a 15 W TDP, system shared memory, and a full suite of graphics APIs including DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

For users selecting between these two, the choice is determined by the workload. The MI350P suits server or workstation environments where its dual-slot form factor, 600 W TDP, and 1x 16-pin power connector are acceptable. Its PCIe 5.0 x16 interface and 8,192 shading units provide the resources for sustained compute. The 880M suits portable devices where its integrated design, 15 W power draw, and 768 shading units fit within a processor package. Its benchmark performance places it near older discrete mobile GPUs like the GTX 675MX and MX330, with deltas of 0.1 and negative 0.3 percent respectively.

The MI350P’s 50th percentile ranking among all GPUs, combined with zero recorded benchmark scores, suggests the database has not yet gathered workload-specific performance data for it. The 880M’s 43rd percentile ranking is backed by ten measurements. Neither GPU is positioned for the same market segment, and the recorded data does not support a direct performance comparison. The MI350P’s theoretical specifications indicate a compute-oriented device, while the 880M’s measured results indicate a capable integrated graphics solution for mainstream mobile tasks.

Specification Differences

The two GPUs differ across nearly every recorded specification. The MI350P uses the MI350 128CU chip, while the 880M uses the Strix Point chip. The MI350P is built on a 3 nm process at TSMC, while the 880M uses a 4 nm process, also at TSMC. Transistor counts differ substantially: 73,000 million for the MI350P versus 34,000 million for the 880M. Die sizes are 1190 mm² for the MI350P and 233 mm² for the 880M. Transistor density favors the 880M at 145.9 million transistors per square millimeter versus 61.3 million for the MI350P.

Clock speeds show a different balance. The MI350P has a base clock of 1000 MHz and a boost clock of 2200 MHz. The 880M has a base clock of 400 MHz and a boost clock of 2900 MHz. The 880M’s higher boost clock reflects its mobile design, but the MI350P’s larger chip and higher throughput compensate for its lower frequency.

Memory configurations are entirely different. The MI350P has 144 GB of HBM3e with an 8192-bit bus and 8.19 TB/s bandwidth. The 880M uses system shared memory with no dedicated size, type, bus width, or bandwidth figures. The MI350P’s memory clock is 2000 MHz with 8 Gbps effective, while the 880M has no memory clock specified.

Compute unit resources diverge sharply. The MI350P has 8,192 shading units, 512 texture mapping units, and 0 raster operation pipelines. The 880M has 768 shading units, 48 texture mapping units, and 16 raster operation pipelines. The MI350P also has no recorded ray tracing cores, while the 880M has 12. Pixel rates are 0 MPixel/s for the MI350P and 46.40 GPixel/s for the 880M. Texture rates are 1,126.4 GTexel/s and 139.2 GTexel/s respectively.

Power and physical specifications differ completely. The MI350P has a 600 W TDP, is dual-slot, uses a 1x 16-pin power connector, and suggests a 1000 W power supply. The 880M has a 15 W TDP, is an IGP, uses no power connectors, and has no suggested power supply. The MI350P uses PCIe 5.0 x16, while the 880M uses PCIe 4.0 x8. The MI350P has no display outputs; the 880M’s outputs are portable device dependent. The MI350P has dimensions of 267 mm length, 111 mm height, and 40 mm width; the 880M has no recorded dimensions.

Architecture Differences

The MI350P is built on CDNA 4.0 architecture, while the 880M uses RDNA 3.5. These are distinct design families within AMD’s lineup. CDNA 4.0 targets compute acceleration, with a focus on FP32 and FP16 throughput at a 1:1 ratio. RDNA 3.5 targets graphics rendering in integrated and mobile parts, with support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The MI350P lists no API support in the database, consistent with its compute-only role.

The generation names reflect their different lineages. The MI350P is part of the Instinct (MIx) generation, with a predecessor of Radeon Instinct. The 880M belongs to the Navi III IGP generation for Strix Point Mobile, with a predecessor of Navi II IGP. The MI350P has no recorded production status, while the 880M is listed as active. Release dates differ by roughly two years: the 880M was released in July 2024, and the MI350P is dated May 2026.

The MI350P’s CDNA 4.0 architecture omits raster operations pipelines entirely, with a pixel rate of 0 MPixel/s and no display outputs. This confirms a design without graphics output stages. The 880M’s RDNA 3.5 architecture includes 16 ROPs and 12 ray tracing cores, enabling standard graphics pipeline functionality. The 880M’s API support includes DirectX 12 Ultimate (12_2), a feature set that requires hardware ray tracing support.

Process node differences are minimal but present: 3 nm for the MI350P versus 4 nm for the 880M. Both use TSMC as the foundry. The MI350P’s larger die (1190 mm²) accommodates 73,000 million transistors, while the 880M’s smaller die (233 mm²) holds 34,000 million. The 880M’s higher transistor density (145.9M per mm² versus 61.3M per mm²) reflects the tighter integration of a mobile IGP.

The MI350P’s CDNA 4.0 architecture provides no rasterization features, no API exposure, and no display interfaces. Its 8,192 shading units and 512 TMUs feed compute workloads, with memory bandwidth of 8.19 TB/s enabling large data movement. The 880M’s RDNA 3.5 architecture provides a complete graphics stack, from pixel output to ray tracing, within a 15 W power envelope. The architectural split between compute acceleration and integrated graphics is reflected in every recorded metric, from memory type to power connectors to slot width.