AMD Radeon RX 7900 GRE vs NVIDIA GeForce RTX 3090 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon RX 7900 GRE records an average benchmark score of 32,456, while the NVIDIA GeForce RTX 3090 averages 27,565. The AMD card sits at the 77th percentile of all GPUs, compared to the RTX 3090's 73rd percentile.

Q: How do the two cards compare in the 3DMark Steel Nomad DX12 test?

A: The NVIDIA GeForce RTX 3090 wins this test with a score of 5,118 versus 4,814 for the AMD Radeon RX 7900 GRE, a margin of 5.9% in NVIDIA's favor.

Q: Which card is ahead in Vulkan compute performance?

A: The AMD Radeon RX 7900 GRE dominates the Geekbench Vulkan test, scoring 99,850 against 53,927 for the RTX 3090. That is a 85.2% advantage for AMD.

Q: What is the launch MSRP of each card?

A: The AMD Radeon RX 7900 GRE had a launch MSRP of 549 USD, while the NVIDIA GeForce RTX 3090 launched at 1,499 USD.

Q: Which card has more memory and wider memory bus?

A: The NVIDIA GeForce RTX 3090 features 24 GB of GDDR6X on a 384-bit bus, delivering 936.2 GB/s bandwidth. The AMD Radeon RX 7900 GRE has 16 GB of GDDR6 on a 256-bit bus, providing 576.0 GB/s bandwidth.

Q: What is the production status of each GPU?

A: The AMD Radeon RX 7900 GRE is listed as Active in production, whereas the NVIDIA GeForce RTX 3090 is marked End-of-life.

Architecture Differences

The two GPUs represent different architectural generations and manufacturing approaches. The AMD Radeon RX 7900 GRE is built on RDNA 3.0, codenamed Plum Bonito, and uses the Navi 31 chip. It is fabricated on a 5 nm process at TSMC, containing 57,700 million transistors on a 529 mm² die. The transistor density works out to 109.1 million transistors per square millimeter.

The NVIDIA GeForce RTX 3090 uses the Ampere architecture with the GA102 chip. It is manufactured on Samsung's 8 nm process, packing 28,300 million transistors into a larger 628 mm² die. Its transistor density is considerably lower at 45.1 million per square millimeter. This means the AMD chip crams nearly twice the transistor density into a smaller physical area, despite having over twice the total transistor count.

Core configuration differs substantially. The RX 7900 GRE carries 5,120 shading units, 320 texture mapping units, and 160 ROPs. It also includes 80 ray tracing cores and no tensor cores. The RTX 3090 has 10,496 shading units, 328 TMUs, and 112 ROPs, along with 82 RT cores and 328 tensor cores. Despite having roughly half the shading units, the AMD card reports a much higher FP32 throughput of 45.98 TFLOPS versus 35.58 TFLOPS for NVIDIA. FP16 performance also diverges: AMD lists 91.96 TFLOPS with a 2:1 ratio, while NVIDIA lists 35.58 TFLOPS at a 1:1 ratio.

Memory subsystems are another major split. AMD uses 16 GB of GDDR6 across a 256-bit bus, achieving 576.0 GB/s bandwidth. NVIDIA uses 24 GB of GDDR6X across a 384-bit bus, achieving 936.2 GB/s bandwidth. Clock behavior differs as well: the RX 7900 GRE has a base clock of 1287 MHz and boost clock of 2245 MHz, with a game clock of 1880 MHz. The RTX 3090 has a higher base clock of 1395 MHz and a boost clock of 1695 MHz, with no game clock listed. Effective memory clocks are 18 Gbps for AMD and 18 Gbps for NVIDIA (listed as 19.5 Gbps effective).

Power and physical design also contrast sharply. The RX 7900 GRE is rated at 260 W TDP with dual-slot width and dual 8-pin power connectors, suggesting a 600 W PSU. The RTX 3090 is rated at 350 W TDP, triple-slot, with a single 12-pin connector and a 750 W suggested PSU. The NVIDIA card is physically larger at 336 mm length, 140 mm height, and 61 mm width, versus 276 mm, 110 mm, and 51 mm for AMD.

Display output support differs as well. AMD offers 1x HDMI 2.1a, 2x DisplayPort 2.1, and 1x USB Type-C. NVIDIA provides 1x HDMI 2.1 and 3x DisplayPort 1.4a, with no USB Type-C. Both support PCIe 4.0 x16, DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 on both sides.

Head-to-Head Benchmarks

The head-to-head results show a 6 to 4 win split in favor of the AMD Radeon RX 7900 GRE across the ten recorded tests. The largest AMD victory by percentage is the Geekbench Vulkan result, where AMD scores 99,850 against NVIDIA's 53,927, a gap of 85.2% for AMD. This is an enormous margin in a single API test. The DirectX 11 PassMark test also favors AMD heavily: 300 versus 220, a 36.4% difference. DirectX 9 shows AMD ahead at 310 versus 268, a 15.7% lead.

Smaller AMD wins appear in several tests. Geekbench OpenCL shows 175,758 against 172,758, a 1.7% edge. PassMark G3D records 27,089 versus 26645, also a 1.7% difference. PassMark G2D gives AMD at 1,180 compared to 1,060 for NVIDIA, an 11% lead in 2D performance.

NVIDIA's wins are led by 3DMark Steel Nomad DX12, where the RTX 3090 posts 5,118 against 4,814, a 5.9% advantage. PassMark DirectX 10 gives NVIDIA the 182 to 139 win, a 23.6% margin in NVIDIA's favor. DirectX 12 PassMark is close at 110 versus 107 for a 2.7% NVIDIA edge. GPU compute PassMark shows NVIDIA at 15,356 against 15,016 for AMD, a 2.2% win for NVIDIA.

The data shows a pattern: AMD wins consistently in legacy DirectX and compute-oriented APIs, while NVIDIA takes the modern DX12 stress test and some specific compute loads. The absolute scores in several tests are quite close, within 1-3%, while the Vulkan result is a true outlier. The average benchmark scores reflect this split, with AMD at 32,456 and NVIDIA at 27,565, a difference of roughly 16%.

The nearest rivals listed for each card put these numbers in context. The RX 7900 GRE's closest competitors cluster around 32,000 average score, within about 1% either way. The RTX 3090's nearest rivals sit near 27,500, also within roughly 1%. So the 4,891 point gap between the two cards is substantial relative to their respective peer groups.

The Verdict

The recorded data points to the AMD Radeon RX 7900 GRE as the stronger overall performer in this comparison. It wins six of ten head-to-head tests, holds a higher average benchmark score by nearly 4,900 points, and ranks higher in the percentile distribution. Its Vulkan advantage is enormous at 85.2%, and it leads in DirectX 9, DirectX 11, G2D, G3D, and OpenCL tests.

The NVIDIA GeForce RTX 3090 retains specific strengths. It wins the modern 3DMark Steel Nomad DX12 test by 5.9%, which indicates better performance in that particular DX12 workload. It also leads in DirectX 10 by 23.6%, DirectX 12 by 2.7%, and GPU compute by 2.2%. The 24 GB memory capacity and 936.2 GB/s bandwidth are clear advantages for capacity-bound workloads, even if the benchmark suite does not isolate that trait.

Buyers prioritizing Vulkan performance, legacy DirectX titles, or overall average benchmark scores should lean toward the RX 7900 GRE. Those who need the larger memory pool or who focus on the specific DX12 workload represented by Steel Nomad may prefer the RTX 3090. The RX 7900 GRE also has the practical advantages of active production status and lower power draw at 260 W TDP, alongside a smaller physical footprint and a dual-slot design. The RTX 3090 is end-of-life and demands 350 W.

The RTX 3090's higher shading unit count and tensor cores, 328 of them, make it a different kind of compute resource, but the AMD card delivers higher FP32 throughput at 45.98 TFLOPS. In the data recorded here, the RX 7900 GRE is the more balanced winner.

Specification Differences

The two cards differ across most core specification fields.

| Specification | AMD Radeon RX 7900 GRE | NVIDIA GeForce RTX 3090 |

|---|---|---|

| Architecture | RDNA 3.0 | Ampere |

| Process node | 5 nm | 8 nm |

| Foundry | TSMC | Samsung |

| Transistors | 57,700 million | 28,300 million |

| Die size | 529 mm² | 628 mm² |

| Transistor density | 109.1M / mm² | 45.1M / mm² |

| Base clock | 1287 MHz | 1395 MHz |

| Boost clock | 2245 MHz | 1695 MHz |

| Memory clock | 2250 MHz, 18 Gbps effective | 1219 MHz, 19.5 Gbps effective |

| Memory size | 16 GB | 24 GB |

| Memory type | GDDR6 | GDDR6X |

| Memory bus width | 256 bit | 384 bit |

| Memory bandwidth | 576.0 GB/s | 936.2 GB/s |

| Shading units | 5120 | 10496 |

| TMUs | 320 | 328 |

| ROPs | 160 | 112 |

| RT cores | 80 | 82 |

| Tensor cores | None | 328 |

| Pixel rate | 359.2 GPixel/s | 189.8 GPixel/s |

| Texture rate | 718.4 GTexel/s | 556.0 GTexel/s |

| FP32 performance | 45.98 TFLOPS | 35.58 TFLOPS |

| FP16 performance | 91.96 TFLOPS (2:1) | 35.58 TFLOPS (1:1) |

| TDP | 260 W | 350 W |

| Slot width | Dual-slot | Triple-slot |

| Power connectors | 2x 8-pin | 1x 12-pin |

| Suggested PSU | 600 W | 750 W |

| Display outputs | 1x HDMI 2.1a, 2x DisplayPort 2.1, 1x USB Type-C | 1x HDMI 2.1, 3x DisplayPort 1.4a |

| Dimensions | 276 mm x 110 mm x 51 mm | 336 mm x 140 mm x 61 mm |

| Production status | Active | End-of-life |

| Release date | 2023-07-26 | 2020-08-31 |

| Launch MSRP | 549 USD | 1,499 USD |

The AMD card has more ROPs, higher pixel and texture rates, and a faster FP32 rating. The NVIDIA card counters with more shading units, more TMUs, a wider memory bus, and more memory capacity. The two GPUs are built on different process nodes with different foundries, and they differ in every clock speed field except for the shared PCIe 4.0 x16 bus interface. The RTX 3090 has direct predecessor and successor entries in the database, while the RX 7900 GRE lists the same Navi II to Navi IV generation lineage.