AMD Instinct MI455X vs NVIDIA GeForce RTX 5090 Mobile Comparison

AMD Instinct MI455X and NVIDIA GeForce RTX 5090 Mobile represent two divergent approaches to GPU design, one aimed at massive compute throughput and the other at mobile graphics efficiency. The recorded data shows a stark contrast in theoretical specifications, yet the benchmark results are entirely one-sided due to the absence of any recorded scores for the AMD accelerator. The following analysis draws exclusively on the database entries for both parts.

Head-to-Head Benchmarks

The head-to-head comparison is defined by a single critical fact: the AMD Instinct MI455X has no recorded benchmark scores in the database, while the NVIDIA GeForce RTX 5090 Mobile has a complete suite of ten benchmark results. Consequently, the AMD part holds zero wins in direct comparisons, and the NVIDIA part holds zero wins as well, because no matching test scores exist for the AMD side to compare against. The database records a win count of 0 for both products, reflecting the absence of overlapping benchmark data rather than an equivalence in performance.

For the NVIDIA GeForce RTX 5090 Mobile, the benchmark suite provides a clear performance profile. In 3DMark Steel Nomad DX12, the mobile GPU scores 5871 points. The Geekbench OpenCL result is 201834 points, while the Geekbench Vulkan score is 198405 points. PassMark tests show a G3D score of 30034, a GPU Compute score of 13401, and legacy DirectX results of 324 in DX9, 269 in DX11, 183 in DX10, and 138 in DX12. The PassMark G2D score is 1057. The average benchmark score across all tests for the RTX 5090 Mobile is 45152, placing it at the 84th percentile of all GPUs in the database.

The nearest rival data for the RTX 5090 Mobile provides context for these averages. The AMD Radeon Pro 5500 XT has an average score of 45384, which is 0.5% higher than the RTX 5090 Mobile. The NVIDIA GeForce RTX 4070 Ti averages 44795, which is 0.8% lower than the mobile part. The Intel Arc A730M averages 45592, putting it 1% higher. The NVIDIA RTX 5880 Ada Generation averages 45972, which is 1.8% higher than the RTX 5090 Mobile. These deltas show the mobile GPU sitting within a narrow band of desktop and mobile competitors, all within roughly 2% of each other in average score.

The absence of AMD Instinct MI455X benchmark data means its percentile rank of 50 is a placeholder, not a measured outcome. The RTX 5090 Mobile, by contrast, has a measured percentile of 84, indicating that it outperforms the majority of recorded GPUs in the database. The AMD accelerator cannot be positioned relative to these rivals without any scores to analyze.

FAQ

Q: Why does the AMD Instinct MI455X have no benchmark scores in the database?

A: The database records an empty benchmarks array for the AMD Instinct MI455X, resulting in an average benchmark score of 0. No head-to-head benchmark entries exist between the two products, so no comparative scores are available for the AMD accelerator.

Q: What is the average benchmark score for the NVIDIA GeForce RTX 5090 Mobile?

A: The RTX 5090 Mobile has an average benchmark score of 45152, derived from its ten recorded tests. This places the GPU at the 84th percentile of all GPUs in the database.

Q: How does the RTX 5090 Mobile compare to its nearest rivals?

A: The RTX 5090 Mobile is 0.5% slower than the AMD Radeon Pro 5500 XT, 0.8% faster than the NVIDIA GeForce RTX 4070 Ti, 1% slower than the Intel Arc A730M, and 1.8% slower than the NVIDIA RTX 5880 Ada Generation. All four rivals sit within a 2% band of the mobile GPU's average score.

Q: What is the highest PassMark score recorded for the RTX 5090 Mobile?

A: The highest PassMark result for the RTX 5090 Mobile is the G3D score of 30034. The GPU Compute score is 13401, and the G2D score is 1057.

Q: Does the AMD Instinct MI455X have any API support listed?

A: The database lists DirectX, OpenGL, and Vulkan as N/A for the AMD Instinct MI455X. The RTX 5090 Mobile, in contrast, supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: What is the transistor density of each GPU?

A: The AMD Instinct MI455X has a transistor density of 107.0 million transistors per square millimeter, while the NVIDIA GeForce RTX 5090 Mobile has a density of 120.6 million per square millimeter.

Where Each One Wins

The NVIDIA GeForce RTX 5090 Mobile wins in every measurable category because it is the only part with recorded benchmark data. Its 3DMark Steel Nomad DX12 score of 5871, Geekbench OpenCL score of 201834, and Geekbench Vulkan score of 198405 provide concrete evidence of graphics and compute performance. The PassMark suite confirms broad capability across legacy DirectX versions, with the DX9 score of 324 being the highest among the DirectX tests. The G3D score of 30034 indicates strong rasterization performance, while the GPU Compute score of 13401 shows compute throughput that is roughly 45% of the G3D score.

The AMD Instinct MI455X has no benchmark wins because no scores exist. Its theoretical specifications, however, suggest a different purpose. The FP32 compute rating of 157.3 TFLOPS and FP16 rating of 157.3 TFLOPS (1:1) are far above the RTX 5090 Mobile's 31.80 TFLOPS in both precisions. The texture rate of 2,457.6 GTexel/s dwarfs the mobile GPU's 496.9 GTexel/s. These figures indicate that the AMD part is designed for raw compute workloads, not for the gaming-oriented benchmark suite where the NVIDIA part is measured. The RTX 5090 Mobile, with its 82 ray tracing cores and 328 tensor cores, is equipped for graphics features that the AMD accelerator lacks entirely, as the MI455X has no RT cores, no tensor cores, and a pixel rate of 0 MPixel/s.

The use-case split is clear from the data. The RTX 5090 Mobile is the choice for any workload represented by the recorded benchmarks, including DirectX gaming, OpenCL compute, Vulkan rendering, and PassMark's general graphics tests. The AMD Instinct MI455X, with no display outputs and no API support, cannot participate in those tests. Its strengths lie in raw compute density, memory capacity, and bandwidth, which the benchmark suite does not measure.

Specification Differences

The two GPUs differ across nearly every specification field. The AMD Instinct MI455X uses an MI450 256CU chip built on a 2 nm process at TSMC, while the NVIDIA GeForce RTX 5090 Mobile uses a GB203 chip on a 5 nm process, also at TSMC. The AMD part contains 320,000 million transistors on a 2990 mm² die, whereas the NVIDIA part has 45,600 million transistors on a 378 mm² die. The transistor density favors NVIDIA at 120.6M per mm² versus 107.0M per mm² for AMD.

Clock speeds show a mixed comparison. The AMD base clock is 1000 MHz with a boost of 2400 MHz, while the NVIDIA base clock is 990 MHz with a boost of 1515 MHz. The AMD boost clock is significantly higher. Memory clocks differ in format: the AMD memory runs at 1900 MHz with 7.6 Gbps effective, while the NVIDIA memory runs at 1750 MHz with 28 Gbps effective.

Memory capacity and bandwidth heavily favor the AMD accelerator. The MI455X has 432 GB of HBM4 memory on a 24576-bit bus, delivering 23.3 TB/s of bandwidth. The RTX 5090 Mobile has 24 GB of GDDR7 memory on a 256-bit bus, delivering 896.0 GB/s of bandwidth. The AMD memory bus is 96 times wider, and the bandwidth is roughly 26 times higher.

The compute unit counts differ substantially. The AMD part has 32768 shading units, 1024 TMUs, and 0 ROPs. The NVIDIA part has 10496 shading units, 328 TMUs, and 112 ROPs. The AMD texture rate of 2,457.6 GTexel/s is roughly 5 times the NVIDIA rate of 496.9 GTexel/s. The pixel rate is 0 MPixel/s for AMD and 169.7 GPixel/s for NVIDIA.

Power and interface specifications diverge sharply. The AMD TDP is 2300 W with a suggested PSU of 2700 W, while the NVIDIA TDP is 95 W with no suggested PSU listed. The AMD slot width is EAM Module, and the NVIDIA slot width is IGP. The AMD bus interface is PCIe 6.0 x16, and the NVIDIA interface is PCIe 5.0 x16. The AMD part has no display outputs, while the NVIDIA outputs are portable device dependent.

Architecture Differences

The architectural split is defined by the GPU generations. The AMD Instinct MI455X uses CDNA 5.0 architecture from the Instinct (MIx) generation, with a release date of July 22, 2026. The NVIDIA GeForce RTX 5090 Mobile uses Blackwell 2.0 architecture from the GeForce 50 Mobile generation, with a release date of March 26, 2025. The AMD predecessor is listed as Radeon Instinct, and the NVIDIA predecessor is GeForce 40 Mobile.

The feature sets reflect their intended markets. The AMD part has no ray tracing cores, no tensor cores, and no supported APIs, with DirectX, OpenGL, and Vulkan all marked N/A. The NVIDIA part includes 82 ray tracing cores and 328 tensor cores, and supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The AMD production status is not listed, while the NVIDIA production status is Active.

The process nodes indicate different design priorities. The 2 nm node for AMD enables a 2990 mm² die with 320,000 million transistors, an enormous chip dedicated to compute density and memory throughput. The 5 nm node for NVIDIA produces a 378 mm² die with 45,600 million transistors, a much smaller chip optimized for mobile power efficiency. The transistor density difference is modest, with NVIDIA at 120.6M per mm² and AMD at 107.0M per mm², suggesting that the AMD chip's advantage comes from its sheer physical size rather than density.

The memory technologies are generationally distinct. The AMD HBM4 implementation with a 24576-bit bus and 23.3 TB/s bandwidth is designed for massive data movement, while the NVIDIA GDDR7 with a 256-bit bus and 896.0 GB/s bandwidth is conventional for mobile graphics. The clock behavior also differs, with the AMD boost clock of 2400 MHz exceeding the NVIDIA boost of 1515 MHz, though the NVIDIA memory runs at a much higher effective data rate of 28 Gbps versus 7.6 Gbps.

The database does not provide a production status for the AMD part, which leaves its availability unconfirmed. The NVIDIA part is marked Active. No launch MSRP is recorded for either product, so no pricing information is available in the database.