AMD Instinct MI350P vs AMD Ryzen Z1 Extreme GPU Comparison

Where Each One Wins

The recorded data places these two AMD GPUs at opposite ends of the compute spectrum. The AMD Instinct MI350P is a data-center accelerator built around the CDNA 4.0 architecture, with 8,192 shading units and 512 texture mapping units. It targets high-throughput workloads where massive memory bandwidth and raw FP32 compute matter most. Its benchmark profile is defined by a FP32 throughput of 36.04 TFLOPS, a texture rate of 1,126.4 GTexel/s, and a memory bandwidth of 8.19 TB/s. These figures indicate that the MI350P wins in dense numerical workloads, large-scale matrix operations, and applications that can saturate an 8192-bit HBM3e interface.

The AMD Ryzen Z1 Extreme GPU, in contrast, is a mobile-grade component using the RDNA 3.0 architecture, with 768 shading units, 48 TMUs, and 32 ROPs. It also includes 12 ray tracing cores, which the MI350P lacks entirely. The Z1 Extreme delivers 8.294 TFLOPS of FP32 performance and 16.59 TFLOPS of FP16 performance using a 2:1 ratio. Its wins come in graphics-oriented tasks: rasterization, ray-traced rendering, and real-time visual workloads. The presence of 32 ROPs gives it a pixel rate of 86.40 GPixel/s, while the MI350P records 0 MPixel/s because it has no ROPs and no display outputs. The data shows that the MI350P is not designed for pixel generation, while the Z1 Extreme is fully equipped for it.

The architectural split is clear. The MI350P wins on raw compute density and memory throughput, while the Z1 Extreme wins on graphics features, ray tracing, and power efficiency. The MI350P has a TDP of 600 W, whereas the Z1 Extreme operates at 30 W. That 20:1 power ratio does not translate into a 20:1 compute ratio; the MI350P's FP32 figure is roughly 4.3 times higher, and its memory bandwidth is about 160 times higher. But the Z1 Extreme's 86.40 GPixel/s pixel rate is infinitely higher than the MI350P's 0 MPixel/s. These are not competing products in the same segment; they win in entirely different usage contexts.

The Verdict

The benchmark data indicates that the AMD Instinct MI350P is the correct choice for server-side compute, AI inference, and scientific simulation. Its 144 GB of HBM3e memory, 8.19 TB/s bandwidth, and 36.04 TFLOPS FP32 performance make it a data-center accelerator with no display outputs and no DirectX, OpenGL, or Vulkan support. It cannot render frames, and its API list is marked as N/A across the board. Any workload that requires graphics output, ray tracing, or a standard consumer graphics API must look to the other card.

The AMD Ryzen Z1 Extreme GPU is the correct choice for portable gaming and graphics-accelerated mobile devices. It supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, and it includes 12 ray tracing cores. Its 16 GB of LPDDR5 memory on a 64-bit bus delivers 51.20 GB/s, which is modest but sufficient for its 30 W power envelope. The Z1 Extreme has a launch MSRP of 699 USD, and it outputs video through a single USB Type-C port. The MI350P has no launch MSRP listed and no display outputs.

The percentile data places both at the 50th percentile among all GPUs, but this is a misleading metric for these two parts. The MI350P's average benchmark score is 0, as is the Z1 Extreme's, because neither has recorded benchmark entries in the database. The percentile is a positional rank, not a performance score. The verdict from the data is structural: the MI350P wins where compute throughput and memory capacity dominate, and the Z1 Extreme wins where graphics features, API compatibility, and low power consumption dominate. There is no overlap in their intended use cases.

Head-to-Head Benchmarks

The head-to-head benchmark list is empty, so the comparison rests on the recorded specification data. The largest win for the MI350P is in memory bandwidth. The MI350P provides 8.19 TB/s over an 8192-bit bus, while the Z1 Extreme provides 51.20 GB/s over a 64-bit bus. That is a 160-fold difference in raw bandwidth. The MI350P's 144 GB of HBM3e also dwarfs the Z1 Extreme's 16 GB of LPDDR5.

The second major win for the MI350P is in texture throughput. Its 1,126.4 GTexel/s rate is about 8.7 times the Z1 Extreme's 129.6 GTexel/s. The FP32 compute gap is narrower but still substantial: 36.04 TFLOPS versus 8.294 TFLOPS, a 4.3x advantage. The MI350P also has more shading units (8,192 versus 768), more TMUs (512 versus 48), and a larger die at 1190 mm² versus 178 mm². Its transistor count is 73,000 million versus 25,390 million.

The Z1 Extreme wins decisively in pixel throughput. Its 86.40 GPixel/s is a functional capability, while the MI350P records 0 MPixel/s. The Z1 Extreme also wins in FP16 compute when measured by its peak ratio: 16.59 TFLOPS at 2:1, whereas the MI350P achieves 36.04 TFLOPS at 1:1. The MI350P's FP16 figure is higher in absolute terms, but the Z1 Extreme's ratio suggests a different optimization for half-precision workloads. The Z1 Extreme also wins in clock speed: its boost clock is 2700 MHz versus the MI350P's 2200 MHz, and its base clock is 800 MHz versus 1000 MHz, so the MI350P has a higher base but the Z1 Extreme has a higher boost.

The Z1 Extreme wins on API support. It lists DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The MI350P lists N/A for all three. The Z1 Extreme has 12 ray tracing cores; the MI350P has none. The Z1 Extreme also has 32 ROPs, which the MI350P lacks entirely. In power efficiency, the Z1 Extreme's 30 W TDP is 20 times lower than the MI350P's 600 W, and it requires no power connectors, whereas the MI350P uses a single 16-pin connector and a suggested 1000 W PSU.

FAQ

Q: Which GPU has higher FP32 compute performance?

A: The AMD Instinct MI350P delivers 36.04 TFLOPS of FP32 performance, which is about 4.3 times the 8.294 TFLOPS of the AMD Ryzen Z1 Extreme GPU.

Q: Does the AMD Instinct MI350P support ray tracing?

A: No. The MI350P has no ray tracing cores listed, while the AMD Ryzen Z1 Extreme GPU includes 12 ray tracing cores.

Q: Which GPU can output video to a display?

A: Only the AMD Ryzen Z1 Extreme GPU can, via a single USB Type-C output. The AMD Instinct MI350P has no display outputs and records a pixel rate of 0 MPixel/s.

Q: How do their memory bandwidths compare?

A: The AMD Instinct MI350P provides 8.19 TB/s over an 8192-bit HBM3e interface. The AMD Ryzen Z1 Extreme GPU provides 51.20 GB/s over a 64-bit LPDDR5 interface. The MI350P's bandwidth is approximately 160 times higher.

Q: What is the difference in power consumption?

A: The AMD Instinct MI350P has a TDP of 600 W and requires a 16-pin power connector with a suggested 1000 W PSU. The AMD Ryzen Z1 Extreme GPU has a TDP of 30 W and uses no power connectors.

Q: Which GPU supports DirectX 12 Ultimate?

A: The AMD Ryzen Z1 Extreme GPU supports DirectX 12 Ultimate (12_2), along with OpenGL 4.6 and Vulkan 1.4. The AMD Instinct MI350P lists N/A for DirectX, OpenGL, and Vulkan.

Architecture Differences

The two GPUs are built on different AMD architectures. The AMD Instinct MI350P uses CDNA 4.0, designed for compute acceleration, while the AMD Ryzen Z1 Extreme GPU uses RDNA 3.0, designed for graphics rendering. This is the fundamental split that explains every other difference in the data.

The manufacturing process differs. The MI350P is fabricated on a 3 nm process at TSMC, while the Z1 Extreme uses a 4 nm process at the same foundry. The MI350P has a much larger die at 1190 mm², compared to 178 mm² for the Z1 Extreme. The MI350P packs 73,000 million transistors, versus 25,390 million for the Z1 Extreme. The transistor density is higher on the Z1 Extreme at 142.6M per mm², versus 61.3M per mm² for the MI350P, which reflects the MI350P's larger, more power-hungry compute-oriented layout.

Memory architecture is entirely different. The MI350P uses 144 GB of HBM3e on an 8192-bit bus, yielding 8.19 TB/s. The Z1 Extreme uses 16 GB of LPDDR5 on a 64-bit bus, yielding 51.20 GB/s. The MI350P's memory clock is 2000 MHz with 8 Gbps effective, while the Z1 Extreme's memory clock is 800 MHz with 6.4 Gbps effective.

Compute resources differ substantially. The MI350P has 8,192 shading units, 512 TMUs, and 0 ROPs. The Z1 Extreme has 768 shading units, 48 TMUs, and 32 ROPs. The MI350P has no ray tracing cores; the Z1 Extreme has 12. The MI350P's texture rate is 1,126.4 GTexel/s, and its pixel rate is 0 MPixel/s. The Z1 Extreme's texture rate is 129.6 GTexel/s, and its pixel rate is 86.40 GPixel/s.

Clock behavior also differs. The MI350P has a base clock of 1000 MHz and a boost clock of 2200 MHz. The Z1 Extreme has a base clock of 800 MHz and a boost clock of 2700 MHz. The MI350P's FP16 performance is 36.04 TFLOPS at a 1:1 ratio with FP32, while the Z1 Extreme's FP16 is 16.59 TFLOPS at a 2:1 ratio, indicating different half-precision strategies.

The physical and interface specs diverge as well. The MI350P is a dual-slot card, 267 mm long, 111 mm high, and 40 mm wide, using PCIe 5.0 x16. The Z1 Extreme is 280 mm long, 111 mm high, and 21 mm wide, with no listed bus interface. The MI350P has no display outputs; the Z1 Extreme has one USB Type-C. The MI350P uses a single 16-pin power connector and suggests a 1000 W PSU, while the Z1 Extreme has no power connectors and no PSU suggestion. The MI350P has no API support listed, while the Z1 Extreme supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The MI350P's release date is recorded as 2026-05-06, while the Z1 Extreme's release date is 2023-06-12.