AMD Instinct MI350P vs NVIDIA GeForce RTX 5080 Comparison

Head-to-Head Benchmarks

The comparison between the AMD Instinct MI350P and the NVIDIA GeForce RTX 5080 is not a conventional one. The database shows no head-to-head benchmark results between these two accelerators, and the MI350P has no recorded benchmark scores or average score in our measurements. The RTX 5080, by contrast, has a substantial benchmark profile with an average score of 56,083, placing it in the 87th percentile of all GPUs in the database.

The RTX 5080 delivers its strongest result in Geekbench Vulkan with a score of 255,450, followed closely by Geekbench OpenCL at 235,901. These cross-API compute results indicate strong general-purpose throughput. In Passmark G3D, the card records 36,565, while Passmark GPU Compute shows 21,789. DirectX-specific tests are more modest: Passmark DirectX 11 scores 324, DirectX 10 scores 208, DirectX 12 scores 151, and DirectX 9 scores 389. The Passmark G2D score of 1,415 reflects 2D graphics performance.

The MI350P, with zero recorded benchmarks and an average score of 0, cannot be directly compared on raw numbers. Its percentile rank of 50 places it at the median of the database distribution, but this is derived from its empty benchmark set rather than measured performance. The data indicates that no direct head-to-head testing exists between these two products, so any performance comparison must rely on architectural and specification analysis rather than empirical benchmark deltas.

Where Each One Wins

Based on the recorded data, the RTX 5080 wins in every measurable category for which benchmarks exist. Its 56.28 TFLOPS FP32 throughput and matching 56.28 TFLOPS FP16 performance give it a clear compute advantage over the MI350P's 36.04 TFLOPS in both precisions. The RTX 5080 also has a higher boost clock at 2617 MHz versus the MI350P's 2200 MHz, and a base clock of 2295 MHz against 1000 MHz.

The MI350P counters with memory capacity and bandwidth. Its 144 GB of HBM3e memory on an 8192-bit bus delivers 8.19 TB/s bandwidth, dwarfing the RTX 5080's 16 GB of GDDR7 on a 256-bit bus at 960.0 GB/s. This represents a 13.5x advantage in memory bandwidth and a 9x advantage in capacity, figures that point toward large-scale data residency rather than graphics rendering.

The RTX 5080 has a full graphics feature set: DirectX 12 Ultimate (12_2), OpenGL 4.6, Vulkan 1.4, and display outputs including 1x HDMI 2.1b and 3x DisplayPort 2.1b. The MI350P lists no APIs and no display outputs, confirming its compute-only orientation. Pixel rate for the MI350P is 0 MPixel/s, while the RTX 5080 delivers 293.1 GPixel/s. Texture rates are 1,126.4 GTexel/s for the MI350P against 879.3 GTexel/s for the RTX 5080, one of the few metrics where the AMD part leads.

FAQ

Q: Which GPU has higher FP32 compute throughput?

A: The RTX 5080 delivers 56.28 TFLOPS FP32, which is 56.2% higher than the MI350P's 36.04 TFLOPS. The same 56.28 TFLOPS versus 36.04 TFLOPS applies to FP16, as both parts run FP16 at a 1:1 ratio with FP32.

Q: How do the memory subsystems compare?

A: The MI350P uses 144 GB of HBM3e on an 8192-bit bus for 8.19 TB/s bandwidth. The RTX 5080 uses 16 GB of GDDR7 on a 256-bit bus for 960.0 GB/s. The MI350P leads by 8.19 TB/s versus 960.0 GB/s in bandwidth and 144 GB versus 16 GB in capacity.

Q: Does the MI350P support any graphics APIs or display outputs?

A: The database records no DirectX, OpenGL, or Vulkan support for the MI350P, and it has no display outputs. The RTX 5080 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, Vulkan 1.4, and includes 1x HDMI 2.1b plus 3x DisplayPort 2.1b outputs.

Q: What are the thermal design power requirements?

A: The MI350P has a TDP of 600 W with a suggested PSU of 1000 W. The RTX 5080 has a TDP of 360 W with a suggested PSU of 750 W. Both use a single 16-pin power connector and are dual-slot cards.

Q: What is the RTX 5080's benchmark standing relative to its nearest rivals?

A: The RTX 5080's average score of 56,083 places it 0.6% above the AMD Radeon 8060S (55,757), 0.7% above the AMD Radeon RX 6750 GRE 12 GB (55,698), and 2.3% above the AMD Radeon Pro W5700X (54,828). It trails the AMD Radeon RX 9070 GRE (57,367) by 2.2%.

Q: Which card has more shading units and texture mapping units?

A: The RTX 5080 has 10,752 shading units and 336 TMUs. The MI350P has 8,192 shading units and 512 TMUs. The RTX 5080 also has 112 ROPs, 84 RT cores, and 336 tensor cores, while the MI350P has 0 ROPs and no recorded RT or tensor core counts.

Specification Differences

The two cards diverge significantly across nearly every specification field. The MI350P uses a 3 nm process with 73,000 million transistors on a 1190 mm² die, yielding a density of 61.3M transistors per mm². The RTX 5080 uses a 5 nm process with 45,600 million transistors on a 378 mm² die, yielding 120.6M transistors per mm². The MI350P has the larger die by 1190 mm² versus 378 mm², but the RTX 5080 has nearly double the transistor density at 120.6M versus 61.3M per mm².

Clock speeds differ substantially. The MI350P runs at 1000 MHz base and 2200 MHz boost, while the RTX 5080 runs at 2295 MHz base and 2617 MHz boost. Memory clocks also differ: the MI350P uses 2000 MHz with 8 Gbps effective, while the RTX 5080 uses 1875 MHz with 30 Gbps effective.

The MI350P has 8,192 shading units, 512 TMUs, and 0 ROPs. The RTX 5080 has 10,752 shading units, 336 TMUs, and 112 ROPs. The MI350P lists no RT cores or tensor cores, while the RTX 5080 includes 84 RT cores and 336 tensor cores.

Pixel rate is 0 MPixel/s for the MI350P versus 293.1 GPixel/s for the RTX 5080. Texture rate favors the MI350P at 1,126.4 GTexel/s versus 879.3 GTexel/s. The MI350P has no display outputs; the RTX 5080 has 1x HDMI 2.1b and 3x DisplayPort 2.1b. The MI350P lists no API support, while the RTX 5080 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Physical dimensions differ: the MI350P measures 267 mm by 111 mm by 40 mm, while the RTX 5080 measures 304 mm by 137 mm by 40 mm. Both are dual-slot and use PCIe 5.0 x16. TDP is 600 W for the MI350P and 360 W for the RTX 5080; suggested PSU is 1000 W versus 750 W. Both use a single 16-pin power connector.

Release dates differ: the MI350P is dated 2026-05-06, while the RTX 5080 is dated 2025-01-29. The RTX 5080 has a launch MSRP of 999 USD. The MI350P has no launch MSRP recorded. Production status for the MI350P is not listed; the RTX 5080 is marked Active. The MI350P's predecessor is Radeon Instinct, while the RTX 5080's predecessor is GeForce 40 and its successor is GeForce 60.

Architecture Differences

The MI350P is built on CDNA 4.0 architecture with the MI350 128CU chip, part of AMD's Instinct (MIx) generation. The RTX 5080 uses Blackwell 2.0 architecture with the GB203 chip, part of the GeForce 50-series. These are fundamentally different design philosophies: CDNA 4.0 targets compute acceleration with no graphics pipeline, while Blackwell 2.0 is a unified graphics and compute architecture.

The MI350P's memory architecture uses HBM3e with a 8192-bit bus width, enabling 8.19 TB/s of bandwidth. This configuration is designed for data-center workloads where memory capacity and bandwidth dominate. The RTX 5080 uses GDDR7 on a 256-bit bus for 960.0 GB/s, a consumer-oriented memory subsystem optimized for latency and graphics workloads.

The MI350P has no ROPs, no RT cores, and no tensor cores listed, along with no DirectX, OpenGL, or Vulkan support. The RTX 5080 includes a full graphics stack with 112 ROPs, 84 RT cores, and 336 tensor cores, plus support for DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI350P's 512 TMUs versus the RTX 5080's 336 TMUs reflects its focus on texture-heavy compute workloads, though its 0 MPixel/s pixel rate confirms it does not rasterize.

Transistor counts and die sizes reflect different manufacturing targets. The MI350P packs 73,000 million transistors into 1190 mm², while the RTX 5080 packs 45,600 million into 378 mm². Both use TSMC fabrication, but the MI350P uses a 3 nm node versus the RTX 5080's 5 nm node. The MI350P's lower transistor density at 61.3M per mm² suggests a design prioritizing memory and interconnect over logic density, consistent with its HBM3e implementation and 8192-bit bus.

The Verdict

The data indicates two products with almost no overlap in intended use. The RTX 5080 is a fully featured graphics card with display outputs, API support, and a complete benchmark profile. Its 87th percentile ranking and average score of 56,083 show it performs competitively against its nearest rivals, with deltas ranging from 2.3% above the Radeon Pro W5700X to 2.2% below the Radeon RX 9070 GRE.

The MI350P has no recorded benchmarks, no graphics APIs, and no display outputs. Its 50th percentile rank is a placeholder based on an empty benchmark set. What the specification data does show is a compute accelerator with 144 GB of memory, 8.19 TB/s bandwidth, and a 600 W TDP, clearly oriented toward memory-bound data-center workloads rather than graphics rendering.

For users selecting a GPU for graphics, rendering, or any workload requiring display output, the RTX 5080 is the only viable option in this comparison. Its 56.28 TFLOPS FP32, 293.1 GPixel/s pixel rate, and 84 RT cores provide the full graphics feature set, and its benchmark results confirm it operates in the upper percentile of the database.

For users selecting a compute accelerator for large-scale data processing, the MI350P's memory capacity of 144 GB and bandwidth of 8.19 TB/s are the defining characteristics. These specifications exceed the RTX 5080's 16 GB and 960.0 GB/s by wide margins, but without benchmark data, the MI350P's actual performance cannot be validated against any recorded results. The choice between these two cards comes down to workload type: graphics and consumer compute points to the RTX 5080, while memory-bound data-center compute points to the MI350P based on specifications alone.