AMD Instinct MI355X vs NVIDIA GeForce RTX 5090 Mobile Comparison

The AMD Instinct MI355X and NVIDIA GeForce RTX 5090 Mobile occupy opposite ends of the hardware spectrum. One is a massive accelerator module built for dense compute, the other is a slim integrated graphics processor for laptops. The recorded data shows no direct head-to-head benchmark overlap, so the comparison rests on architectural specifications, raw compute capacity, and the limited benchmark suite available for the RTX 5090 Mobile. The MI355X has no benchmark scores in the database, while the RTX 5090 Mobile has an average score of 45152 and sits at the 84th percentile among all GPUs. The MI355X sits at the 50th percentile, but that figure reflects a lack of recorded tests rather than measured performance.

Head-to-Head Benchmarks

Direct benchmark comparison is impossible because the database contains no overlapping test results for these two products. The MI355X has no recorded benchmark scores. The RTX 5090 Mobile has ten recorded results, and those provide the only quantitative performance data available.

The RTX 5090 Mobile’s strongest recorded result is in Geekbench OpenCL, where it scores 201834. Its Geekbench Vulkan score is 198405, nearly identical. These two results indicate that the mobile GPU delivers consistent compute performance across different API backends. The PassMark G3D score of 30034 is the next largest figure, followed by PassMark GPU Compute at 13401. The lower PassMark scores, such as DirectX 9 at 324, DirectX 11 at 269, DirectX 10 at 183, and DirectX 12 at 138, reflect the PassMark suite’s older workload characteristics rather than the GPU’s peak capability. The G2D score of 1057 is the only 2D result and is not comparable to any 3D or compute figures.

The MI355X has no benchmark entries to place against these numbers. Its theoretical FP32 throughput is 78.64 TFLOPS, which is 2.47 times the RTX 5090 Mobile’s 31.80 TFLOPS. Its FP16 figure is also 78.64 TFLOPS with a 1:1 ratio, meaning it does not double throughput for reduced precision. The RTX 5090 Mobile also runs FP16 at 1:1 with 31.80 TFLOPS. Texture rate favors the MI355X heavily: 2457.6 GTexel/s versus 496.9 GTexel/s, a 4.95x advantage. Pixel rate is the reverse case. The MI355X has 0 MPixel/s because it has no ROPs and no display outputs. The RTX 5090 Mobile has 112 ROPs and a pixel rate of 169.7 GPixel/s.

The nearest rivals for the RTX 5090 Mobile give context for its benchmark average. The AMD Radeon Pro 5500 XT scores 45384, which is 0.5% above the RTX 5090 Mobile’s 45152. The Intel Arc A730M scores 45592, 1% higher. The NVIDIA GeForce RTX 4070 Ti scores 44795, 0.8% lower. The NVIDIA RTX 5880 Ada Generation scores 45972, 1.8% higher. These deltas place the RTX 5090 Mobile in a tight band around similar scoring GPUs, with no rival more than 1.8% ahead or 0.8% behind. This suggests the mobile part performs at a level comparable to desktop and workstation GPUs from the prior generation, despite its 95 W power envelope.

The Verdict

The data supports a clear split. The AMD Instinct MI355X is for compute installations that prioritize raw throughput, memory capacity, and bandwidth above all else. The NVIDIA GeForce RTX 5090 Mobile is for portable systems that need a full API feature set, display output, and a wide range of software support. Neither product replaces the other.

The MI355X delivers 288 GB of HBM3e memory on an 8192-bit bus, producing 8.19 TB/s of bandwidth. The RTX 5090 Mobile has 24 GB of GDDR7 on a 256-bit bus, yielding 896.0 GB/s. The MI355X’s bandwidth is 9.14 times higher. The MI355X also uses a 3 nm process at TSMC versus the RTX 5090 Mobile’s 5 nm process, also from TSMC. The MI355X has 185,000 million transistors on a 2380 mm² die, while the RTX 5090 Mobile has 45,600 million transistors on a 378 mm² die. Transistor density is higher on the mobile part at 120.6M per mm² versus 77.7M per mm², but the MI355X’s absolute transistor count is 4.06 times larger.

The RTX 5090 Mobile is the only one of the two that can function as a graphics output device. Its display outputs are listed as portable device dependent, meaning the laptop manufacturer decides the actual ports. The MI355X has no outputs at all. The RTX 5090 Mobile supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI355X lists N/A for all three APIs, indicating it is not designed for graphics rendering in the traditional sense. The RTX 5090 Mobile also has 82 ray tracing cores and 328 tensor cores. The MI355X has no recorded RT or tensor core counts.

For anyone choosing between these two, the decision is not competitive but categorical. The MI355X is best suited for server racks where power is available, cooling is external, and the workload is massive parallel compute. The RTX 5090 Mobile is the only option for a laptop that must run games, render graphics, and still fit inside a portable chassis.

Architecture Differences

The MI355X uses the CDNA 4.0 architecture, built around the MI350 256CU chip. The RTX 5090 Mobile uses Blackwell 2.0, built around the GB203 chip. These architectures target different workloads. CDNA is optimized for data center compute, while Blackwell is designed for consumer graphics and Tensor operations.

The process nodes differ: the MI355X is on a 3 nm TSMC node, while the RTX 5090 Mobile is on a 5 nm TSMC node. The MI355X’s die is 2380 mm², which is 6.3 times larger than the RTX 5090 Mobile’s 378 mm². The transistor counts reflect that scale: 185,000 million versus 45,600 million. The MI355X has 16384 shading units, while the RTX 5090 Mobile has 10496. Texture mapping units number 1024 on the MI355X and 328 on the RTX 5090 Mobile. The MI355X has no ROPs, but the RTX 5090 Mobile has 112.

Memory architecture is fundamentally different. The MI355X uses HBM3e across an 8192-bit bus, giving 8.19 TB/s. The RTX 5090 Mobile uses GDDR7 across a 256-bit bus, giving 896.0 GB/s. The MI355X’s memory clock is listed as 2000 MHz with 8 Gbps effective, while the RTX 5090 Mobile runs at 1750 MHz with 28 Gbps effective. The effective data rate is higher on the NVIDIA part, but the bus width difference overwhelms that advantage.

Clock speeds also diverge. The MI355X runs at a 1000 MHz base and 2400 MHz boost. The RTX 5090 Mobile runs at 990 MHz base and 1515 MHz boost. The MI355X has a higher boost clock by 885 MHz, but its workload assumptions are entirely different. The MI355X’s TDP is 1400 W with a suggested PSU of 1800 W. The RTX 5090 Mobile has a TDP of 95 W and no suggested PSU, since it is integrated into a laptop. The MI355X is an OAM module, while the RTX 5090 Mobile is an IGP.

The MI355X has no power connectors listed and no display outputs. The RTX 5090 Mobile also has no power connectors, but it does have display outputs that depend on the portable device. Both use PCIe 5.0 x16 for host connectivity. The MI355X measures 102 mm in length and 165 mm in width. The RTX 5090 Mobile has no recorded dimensions.

FAQ

Q: Which GPU has higher FP32 compute?

A: The AMD Instinct MI355X delivers 78.64 TFLOPS of FP32, which is 2.47 times the NVIDIA GeForce RTX 5090 Mobile’s 31.80 TFLOPS.

Q: How much memory does each GPU have?

A: The MI355X has 288 GB of HBM3e, while the RTX 5090 Mobile has 24 GB of GDDR7. The MI355X’s memory bus is 8192 bits wide versus 256 bits for the RTX 5090 Mobile.

Q: Does either GPU support ray tracing?

A: The RTX 5090 Mobile has 82 ray tracing cores. The MI355X has no recorded ray tracing core count and no support for DirectX, OpenGL, or Vulkan.

Q: What is the power requirement for each?

A: The MI355X has a TDP of 1400 W and requires a suggested PSU of 1800 W. The RTX 5090 Mobile has a TDP of 95 W with no suggested PSU listed.

Q: Which GPU is more recent?

A: The MI355X has a release date of 2025-06-11. The RTX 5090 Mobile has a release date of 2025-03-26, making the NVIDIA part earlier in the same year.

Q: Can either GPU output video to a display?

A: The RTX 5090 Mobile has display outputs that are portable device dependent. The MI355X has no display outputs at all.

Where Each One Wins

The MI355X wins on every metric that measures raw compute scale. Its FP32 throughput of 78.64 TFLOPS is more than double the RTX 5090 Mobile’s 31.80 TFLOPS. Its texture rate of 2457.6 GTexel/s is nearly five times the 496.9 GTexel/s of the RTX 5090 Mobile. Its memory bandwidth of 8.19 TB/s is over nine times the 896.0 GB/s available to the NVIDIA part. The MI355X also has more shading units, 16384 versus 10496, and more TMUs, 1024 versus 328. For workloads that fit entirely in GPU memory and require maximum throughput, such as large model inference or simulation, the MI355X is the clear choice.

The RTX 5090 Mobile wins on integration and versatility. It has 112 ROPs and a pixel rate of 169.7 GPixel/s, while the MI355X has 0 MPixel/s. The RTX 5090 Mobile supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, while the MI355X has N/A for all three. The RTX 5090 Mobile has ray tracing cores and tensor cores, features absent from the MI355X’s recorded specifications. The RTX 5090 Mobile also fits in a 95 W power envelope, versus 1400 W for the MI355X. For any application that requires rendering, display output, or portability, the RTX 5090 Mobile is the only option between the two.

The benchmark record reinforces this split. The RTX 5090 Mobile’s average score of 45152 places it at the 84th percentile among all GPUs, with nearest rivals within 1.8% of its score. The MI355X has no benchmark data, so its 50th percentile is not a performance indicator. The presence of ten benchmark entries for the RTX 5090 Mobile and zero for the MI355X indicates that the mobile GPU is tested across a broad software ecosystem, while the MI355X is not subjected to consumer benchmark suites. That distinction is meaningful: the RTX 5090 Mobile is a general-purpose GPU, while the MI355X is a specialized compute accelerator.

Specification Differences

The two GPUs differ in nearly every specification category. The MI355X uses a 3 nm process, the RTX 5090 Mobile uses 5 nm. The MI355X has 185,000 million transistors on a 2380 mm² die, while the RTX 5090 Mobile has 45,600 million on 378 mm². Transistor density favors the mobile part at 120.6M per mm² versus 77.7M per mm².

Clock speeds: the MI355X runs at 1000 MHz base and 2400 MHz boost, the RTX 5090 Mobile at 990 MHz base and 1515 MHz boost. Memory clocks: the MI355X uses 2000 MHz with 8 Gbps effective, the RTX 5090 Mobile uses 1750 MHz with 28 Gbps effective. Memory type and size: HBM3e 288 GB versus GDDR7 24 GB. Bus width: 8192 bits versus 256 bits. Bandwidth: 8.19 TB/s versus 896.0 GB/s.

Compute units: the MI355X has 16384 shading units, 1024 TMUs, and 0 ROPs. The RTX 5090 Mobile has 10496 shading units, 328 TMUs, and 112 ROPs. The MI355X has no RT cores or tensor cores recorded, while the RTX 5090 Mobile has 82 RT cores and 328 tensor cores. Pixel rate is 0 MPixel/s for the MI355X and 169.7 GPixel/s for the RTX 5090 Mobile. Texture rate is 2457.6 GTexel/s versus 496.9 GTexel/s. FP32 and FP16 are each 78.64 TFLOPS for the MI355X and 31.80 TFLOPS for the RTX 5090 Mobile.

Power and form factor: the MI355X has a TDP of 1400 W, a suggested PSU of 1800 W, and an OAM Module slot width. The RTX 5090 Mobile has a TDP of 95 W, no suggested PSU, and an IGP slot width. The MI355X measures 102 mm by 165 mm, while the RTX 5090 Mobile has no recorded dimensions. Display outputs are absent on the MI355X and portable device dependent on the RTX 5090 Mobile. API support is N/A on the MI355X and includes DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 on the RTX 5090 Mobile. Release dates differ by about two and a half months, with the RTX 5090 Mobile arriving first.