AMD Instinct MI350X vs NVIDIA GeForce RTX 4090 D Comparison

Head-to-Head Benchmarks

The recorded data for the AMD Instinct MI350X contains no benchmark entries, while the NVIDIA GeForce RTX 4090 D has three recorded results. The RTX 4090 D achieves an average benchmark score of 178,050, placing it in the 98th percentile of all GPUs in the database. The MI350X holds a 50th percentile ranking with an average benchmark score of 0, reflecting the absence of submitted results rather than a measure of its compute capability.

In the 3DMark Steel Nomad DX12 test, the RTX 4090 D scores 8,587. Its Geekbench OpenCL score reaches 278,621, and its Vulkan score comes in at 246,941. These results place the RTX 4090 D ahead of several notable accelerators in the database. It trails the NVIDIA RTX PRO 5000 Blackwell by 2.2 percent, the NVIDIA A100 SXM4 80 GB by 3.1 percent, the NVIDIA RTX 5000 Ada Generation by 3.6 percent, and the NVIDIA A100 SXM4 40 GB by 4.9 percent. The delta percentages indicate that the RTX 4090 D sits within a tight competitive cluster, with the closest rival, the RTX PRO 5000 Blackwell, holding a 182,109 average score.

The MI350X offers no comparable head-to-head benchmark data. The database records zero wins for the MI350X and zero wins for the RTX 4090 D in direct comparisons. This means a numerical comparison of their measured performance cannot be established. What the data does show is that the RTX 4090 D has verified results across multiple test suites, while the MI350X has none recorded. The MI350X is a newly released accelerator with a production status that remains unlisted, whereas the RTX 4090 D is marked end-of-life. The absence of benchmark results for the MI350X may reflect its specialized role as a compute module rather than a general-purpose graphics card.

The Verdict

Based strictly on the recorded data, the NVIDIA GeForce RTX 4090 D is the only one of the two with measurable benchmark performance. It delivers a 98th percentile standing with an average score of 178,050. The AMD Instinct MI350X has no benchmark results, placing it at the 50th percentile with a zero average score. For any workload where verified performance matters, the RTX 4090 D is the clear choice according to the database.

The MI350X targets a different operational envelope. It draws 1000 W, requires a 1400 W suggested PSU, and mounts as an OAM module with no display outputs. The RTX 4090 D draws 425 W, needs an 800 W PSU, and fits a triple-slot PCIe card with display outputs. The data indicates that the RTX 4090 D is suited for a workstation or desktop environment where users need confirmed performance numbers, graphics output, and API support. The MI350X, with its 288 GB HBM3e memory and 8.19 TB/s bandwidth, is positioned for large-scale compute tasks, but the database currently has no recorded measurements to validate that positioning.

Architecture Differences

The two accelerators share a foundry but diverge in nearly every architectural detail. Both use TSMC fabrication, but the MI350X employs a 3 nm process while the RTX 4090 D uses a 5 nm node. The MI350X packs 185,000 million transistors into a 2380 mm² die, yielding a transistor density of 77.7 million per mm². The RTX 4090 D contains 76,300 million transistors on a 609 mm² die, with a density of 125.3 million per mm². The MI350X has a larger physical die and more transistors, but the RTX 4090 D achieves higher transistor density.

The MI350X uses the CDNA 4.0 architecture with an MI350 256CU chip, while the RTX 4090 D uses the Ada Lovelace architecture with the AD102 chip. Clock speeds differ substantially. The MI350X has a 1000 MHz base clock and a 2200 MHz boost clock. The RTX 4090 D runs at 2280 MHz base and 2520 MHz boost. Memory subsystems are in different classes. The MI350X carries 288 GB of HBM3e across an 8192-bit bus, delivering 8.19 TB/s of bandwidth. The RTX 4090 D has 24 GB of GDDR6X on a 384-bit bus, with 1.01 TB/s bandwidth. The memory clock for the MI350X is 2000 MHz with 8 Gbps effective speed, while the RTX 4090 D runs at 1313 MHz with 21 Gbps effective.

Compute resources also differ. The MI350X has 16,384 shading units, 1,024 texture mapping units, and no ROPs, reporting a pixel rate of 0 MPixel/s. The RTX 4090 D has 14,592 shading units, 456 TMUs, and 176 ROPs, with a pixel rate of 443.5 GPixel/s. Texture rates reflect these counts: the MI350X reaches 2,252.8 GTexel/s, the RTX 4090 D reaches 1,149.1 GTexel/s. FP32 performance is nearly identical at 72.09 TFLOPS for the MI350X and 73.54 TFLOPS for the RTX 4090 D, with both offering 1:1 FP16 performance at the same figures. The RTX 4090 D includes 114 RT cores and 456 tensor cores, while the MI350X lists no RT or tensor core counts.

FAQ

Q: Which card has more memory?

A: The AMD Instinct MI350X has 288 GB of HBM3e memory, while the NVIDIA GeForce RTX 4090 D has 24 GB of GDDR6X.

Q: What is the bandwidth difference?

A: The MI350X delivers 8.19 TB/s across an 8192-bit bus. The RTX 4090 D provides 1.01 TB/s over a 384-bit bus.

Q: Which card has higher FP32 performance?

A: The RTX 4090 D is marginally ahead at 73.54 TFLOPS versus 72.09 TFLOPS for the MI350X.

Q: Does the MI350X support display outputs?

A: No, the MI350X has no display outputs. The RTX 4090 D includes 1x HDMI 2.1 and 3x DisplayPort 1.4a.

Q: What API support does each card offer?

A: The RTX 4090 D supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The MI350X lists N/A for DirectX, OpenGL, and Vulkan.

Q: Which card has a higher thermal design power?

A: The MI350X has a TDP of 1000 W with a suggested PSU of 1400 W. The RTX 4090 D has a TDP of 425 W and a suggested PSU of 800 W.

Where Each One Wins

The NVIDIA GeForce RTX 4090 D wins in any scenario requiring verified benchmark results. Its 178,050 average score, 98th percentile placement, and three recorded test results give it a measurable advantage in the database. It wins on pixel processing with 443.5 GPixel/s versus 0 MPixel/s for the MI350X, reflecting its ROP count of 176 compared to none. It offers full graphics API support with DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, plus display outputs for direct monitor connection. Its boost clock of 2520 MHz exceeds the MI350X boost of 2200 MHz. The RTX 4090 D also has a smaller footprint at 304 mm length, 137 mm height, and 61 mm width, versus the MI350X at 102 mm length and 165 mm width. The RTX 4090 D is production-status end-of-life, meaning it has an established track record in the database.

The AMD Instinct MI350X wins on raw capacity and memory bandwidth. Its 288 GB HBM3e memory dwarfs the 24 GB GDDR6X of the RTX 4090 D. Its 8.19 TB/s bandwidth is roughly eight times the 1.01 TB/s of the RTX 4090 D. The MI350X also leads in texture throughput at 2,252.8 GTexel/s versus 1,149.1 GTexel/s. It uses a more advanced 3 nm process compared to the 5 nm node of the RTX 4090 D. Its 185,000 million transistors and 2380 mm² die size indicate a much larger compute substrate. The MI350X uses PCIe 5.0 x16, while the RTX 4090 D uses PCIe 4.0 x16. For workloads that depend on massive memory capacity and extreme bandwidth, such as large model inference or data-intensive compute, the MI350X offers the necessary resources on paper, though the database has no benchmark scores to confirm real-world performance.

Specification Differences

The two cards differ in nearly every measured specification. The MI350X uses the CDNA 4.0 architecture on a 3 nm node, while the RTX 4090 D uses Ada Lovelace on a 5 nm node. Transistor counts are 185,000 million versus 76,300 million. Die size is 2380 mm² versus 609 mm². Transistor density favors the RTX 4090 D at 125.3M per mm² against 77.7M per mm² for the MI350X.

Base clocks are 1000 MHz for the MI350X and 2280 MHz for the RTX 4090 D. Boost clocks are 2200 MHz and 2520 MHz respectively. Memory size is 288 GB HBM3e versus 24 GB GDDR6X. Bus width is 8192 bit versus 384 bit. Bandwidth is 8.19 TB/s versus 1.01 TB/s. Memory clock is 2000 MHz with 8 Gbps effective for the MI350X, and 1313 MHz with 21 Gbps effective for the RTX 4090 D.

Shading units are 16,384 for the MI350X and 14,592 for the RTX 4090 D. TMU counts are 1,024 versus 456. ROP counts are 0 versus 176. The RTX 4090 D has 114 RT cores and 456 tensor cores; the MI350X lists no such cores. Pixel rate is 0 MPixel/s for the MI350X and 443.5 GPixel/s for the RTX 4090 D. Texture rate is 2,252.8 GTexel/s versus 1,149.1 GTexel/s. FP32 and FP16 performance are 72.09 TFLOPS for the MI350X and 73.54 TFLOPS for the RTX 4090 D.

TDP is 1000 W versus 425 W. The MI350X uses an OAM module form factor with no power connectors, while the RTX 4090 D is triple-slot with a single 16-pin connector. Suggested PSU is 1400 W versus 800 W. Bus interface is PCIe 5.0 x16 versus PCIe 4.0 x16. Display outputs are none for the MI350X and 1x HDMI 2.1 plus 3x DisplayPort 1.4a for the RTX 4090 D. API support is N/A for the MI350X, while the RTX 4090 D supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. Dimensions are 102 mm by 165 mm for the MI350X, and 304 mm by 137 mm by 61 mm for the RTX 4090 D. The RTX 4090 D has a launch MSRP of 1,599 USD. Release dates are June 2025 for the MI350X and December 2023 for the RTX 4090 D. The MI350X lists no production status, while the RTX 4090 D is end-of-life.