AMD Instinct MI350P vs AMD Radeon RX 9070 XT Comparison

Head-to-Head Benchmarks

The recorded data presents an unusual comparison: the AMD Instinct MI350P has no benchmark entries in the database, while the AMD Radeon RX 9070 XT has ten measured results. This means the head-to-head analysis is entirely one-sided. The RX 9070 XT shows an average benchmark score of 13,543, placing it at the 55th percentile of all GPUs tracked. Its nearest rivals in the database are the AMD Radeon HD 7770M with a score of 13,536 (a 0.1% difference), the NVIDIA GeForce GTX 570 at 13,515 (0.2% difference), the NVIDIA P106-090 at 13,470 (0.5% difference), and the AMD Radeon Pro 555 at 13,407 (1% difference). The RX 9070 XT sits just ahead of all four, though the margins are narrow, ranging from 0.1% to 1%.

The RX 9070 XT delivers its strongest result in Geekbench Vulkan with a score of 65,879, which is substantially higher than its OpenCL result of 17,428. The Passmark suite shows a G3D score of 26,795, a GPU compute score of 15,864, and a G2D score of 1,328. DirectX tests in Passmark show scores of 353 for DirectX 9, 292 for DirectX 11, 154 for DirectX 10, and 78 for DirectX 12. In 3DMark Steel Nomad DX12, the card scores 7,260.

For the MI350P, the database records no benchmark scores and an average score of zero, with a percentile rank of 50. Without measured results, the only numerical comparison available is the specification sheet, which shows the MI350P with a significantly different performance profile. The FP32 throughput of the MI350P is 36.04 TFLOPS, while the RX 9070 XT reaches 48.66 TFLOPS. The MI350P has more shading units at 8,192 versus 4,096, more texture mapping units at 512 versus 256, and a higher texture rate at 1,126.4 GTexel/s versus 760.3 GTexel/s. However, the MI350P has a pixel rate of 0 MPixel/s, while the RX 9070 XT produces 380.2 GPixel/s. The MI350P also has no display outputs, which is consistent with its accelerator role.

Architecture Differences

The two GPUs come from different architectural families. The MI350P uses CDNA 4.0, built for compute acceleration, while the RX 9070 XT uses RDNA 4.0, designed for graphics rendering. The manufacturing process differs: the MI350P is on a 3 nm node at TSMC, while the RX 9070 XT uses a 4 nm node, also at TSMC. The MI350P integrates 73,000 million transistors on a 1,190 mm² die, giving a transistor density of 61.3 million per mm². The RX 9070 XT packs 53,900 million transistors onto a 357 mm² die, with a much higher density of 151.0 million per mm². The smaller die and higher density of the RX 9070 XT reflects its client graphics focus, while the MI350P's massive die targets data center compute workloads.

Memory architecture is a major divider. The MI350P uses 144 GB of HBM3e on an 8,192-bit bus, delivering 8.19 TB/s of bandwidth. The RX 9070 XT uses 16 GB of GDDR6 on a 256-bit bus, with 644.6 GB/s bandwidth. The MI350P has 8192 shading units and 512 TMUs, but no ROPs, which is why its pixel rate is listed as zero. The RX 9070 XT has 4,096 shading units, 256 TMUs, 128 ROPs, and 64 ray tracing cores. The MI350P lists no ray tracing cores, no tensor cores, and no graphics API support for DirectX, OpenGL, or Vulkan. The RX 9070 XT supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Clock behavior also differs. The MI350P has a base clock of 1000 MHz and a boost clock of 2200 MHz, with memory running at 2000 MHz (8 Gbps effective). The RX 9070 XT has a base clock of 1660 MHz, a boost clock of 2970 MHz, and a game clock of 2400 MHz, with memory at 2518 MHz (20.1 Gbps effective). Power requirements separate the two clearly: the MI350P is rated at 600 W with a suggested 1000 W PSU and a single 16-pin connector, while the RX 9070 XT draws 304 W, suggests a 700 W PSU, and uses two 8-pin connectors.

Where Each One Wins

The data indicates the RX 9070 XT is the only one of the two with measured graphics performance. Its benchmark scores cover DirectX and Vulkan workloads, and its display outputs (1x HDMI 2.1b and 3x DisplayPort 2.1a) allow it to drive monitors directly. The presence of 128 ROPs and a pixel rate of 380.2 GPixel/s confirms it can handle rasterization. The 64 ray tracing cores add hardware support for ray-traced effects, and the FP32 throughput of 48.66 TFLOPS is higher than the MI350P's 36.04 TFLOPS. For any workload that requires rendering, video output, or graphics API support, the RX 9070 XT is the only viable choice in this pair.

The MI350P wins in memory capacity and bandwidth. Its 144 GB of HBM3e with 8.19 TB/s bandwidth is a massive advantage for data sets that exceed 16 GB, which is the entire memory pool of the RX 9070 XT. The MI350P also has more shading units and TMUs, and its texture rate of 1,126.4 GTexel/s outperforms the RX 9070 XT's 760.3 GTexel/s. The lack of ROPs and display outputs means it is not built for graphics presentation, but for compute tasks that saturate memory and texture pipelines. The MI350P's 3 nm process and 73,000 million transistors give it a raw compute scale that the RX 9070 XT does not match in shading unit count, despite the RX 9070 XT's higher FP32 TFLOP rating.

The benchmark database shows no measured wins for the MI350P, simply because none are recorded. The RX 9070 XT holds all ten benchmark entries, with its highest scores in Vulkan and G3D tests. The percentile placement of the RX 9070 XT at 55 versus 50 for the MI350P reflects the available data, not necessarily the compute capability of the accelerator.

FAQ

Q: Which GPU has higher FP32 compute?

A: The RX 9070 XT lists 48.66 TFLOPS, while the MI350P lists 36.04 TFLOPS.

Q: How much memory does each card have?

A: The MI350P has 144 GB of HBM3e, while the RX 9070 XT has 16 GB of GDDR6.

Q: Does the MI350P support DirectX?

A: No. The MI350P lists DirectX as N/A, along with OpenGL and Vulkan. The RX 9070 XT supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: What is the memory bandwidth difference?

A: The MI350P delivers 8.19 TB/s, while the RX 9070 XT delivers 644.6 GB/s.

Q: Can the MI350P output video?

A: No. The MI350P has no display outputs, while the RX 9070 XT has 1x HDMI 2.1b and 3x DisplayPort 2.1a.

Q: What are the power requirements?

A: The MI350P is rated at 600 W with a suggested 1000 W PSU. The RX 9070 XT is rated at 304 W with a suggested 700 W PSU.

Specification Differences

The two cards differ in nearly every major specification. The MI350P uses a 3 nm process, while the RX 9070 XT uses 4 nm. Transistor count is 73,000 million for the MI350P versus 53,900 million for the RX 9070 XT. Die size is 1190 mm² versus 357 mm². Transistor density is 61.3M per mm² versus 151.0M per mm². The MI350P has a base clock of 1000 MHz and boost of 2200 MHz, while the RX 9070 XT has 1660 MHz base, 2970 MHz boost, and a 2400 MHz game clock. Memory clock is 2000 MHz (8 Gbps effective) on the MI350P and 2518 MHz (20.1 Gbps effective) on the RX 9070 XT.

Memory capacity is 144 GB versus 16 GB. Memory type is HBM3e versus GDDR6. Bus width is 8192 bit versus 256 bit. Bandwidth is 8.19 TB/s versus 644.6 GB/s. Shading units are 8192 versus 4096. TMUs are 512 versus 256. ROPs are 0 versus 128. The RX 9070 XT has 64 ray tracing cores; the MI350P has none listed. Pixel rate is 0 MPixel/s versus 380.2 GPixel/s. Texture rate is 1,126.4 GTexel/s versus 760.3 GTexel/s. FP32 is 36.04 TFLOPS versus 48.66 TFLOPS. FP16 is 36.04 TFLOPS (1:1) for both, though the RX 9070 XT also lists 48.66 TFLOPS (1:1).

TDP is 600 W versus 304 W. Power connectors are 1x 16-pin versus 2x 8-pin. Suggested PSU is 1000 W versus 700 W. The MI350P has no display outputs; the RX 9070 XT has 1x HDMI 2.1b and 3x DisplayPort 2.1a. API support is N/A for the MI350P, while the RX 9070 XT supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI350P is 267 mm long, 111 mm high, and 40 mm wide; the RX 9070 XT has no dimensions recorded. Release dates differ: the RX 9070 XT launched on 2025-03-05, while the MI350P is dated 2026-05-06. The RX 9070 XT has a launch MSRP of 599 USD. The MI350P has no recorded MSRP.

The Verdict

The RX 9070 XT is the only card in this pair with any benchmark presence. Ten recorded tests place it at the 55th percentile, with an average score of 13,543 and nearest rivals within 1%. It delivers measured performance in Vulkan, OpenCL, DirectX, and Passmark workloads, plus display outputs and full graphics API support. For rendering, gaming, or any task that needs a screen, the RX 9070 XT is the clear choice. Its higher FP32 throughput and pixel rate reinforce that.

The MI350P is a compute accelerator with no graphics outputs, no API support, and no benchmark scores in the database. Its strengths are structural: 144 GB of HBM3e, 8.19 TB/s bandwidth, 8,192 shading units, and a 3 nm process with 73,000 million transistors. Those specifications serve workloads that need massive memory capacity and bandwidth, not rasterization. The data does not show any measured performance for the MI350P, so its relative standing against the RX 9070 XT in compute tasks cannot be quantified from the database. The recorded numbers, however, indicate that the MI350P is built for memory-bound compute, while the RX 9070 XT is built for graphics. The verdict follows the data: select the RX 9070 XT for graphics workloads with confirmed benchmark results, and consider the MI350P only for compute roles that require its memory configuration, acknowledging that no benchmark data exists to support a numeric performance claim.