AMD Radeon RX 580 vs NVIDIA GeForce GTX 1080 Comparison

The AMD Radeon RX 580 and NVIDIA GeForce GTX 1080 represent two very different approaches to the high-end graphics card market from the same era. The data shows a clear split: the GTX 1080 dominates in traditional DirectX and compute workloads, while the RX 580 posts surprising wins in specific API-level tests. The head-to-head benchmark suite records 9 wins for the NVIDIA card and only 2 for the AMD card, yet the nature of those wins reveals important behavioral differences that go beyond a simple aggregate score.

Head-to-Head Benchmarks

The most lopsided defeat for the AMD Radeon RX 580 comes in the Passmark GPU compute test, where the NVIDIA GeForce GTX 1080 scores 7614 against 3488, a delta of -54.2%. This is the largest percentage gap in the entire comparison, indicating a fundamental advantage in raw parallel throughput for the Pascal card. The GTX 1080 also wins Passmark G3D by a wide margin, scoring 15586 versus 8813, which is a 43.5% advantage. These two results alone establish the NVIDIA card as the stronger performer for generalized 3D rendering tasks.

The DirectX legacy tests follow the same pattern. In Passmark DirectX 11, the GTX 1080 scores 124 against the RX 580’s 60, a 51.6% difference. Passmark DirectX 10 shows a 50.5% gap (93 vs 46), and even DirectX 9, where the RX 580 scores its highest Passmark API result at 124, the GTX 1080 still wins with 211, a 41.2% margin. The newer DirectX 12 Passmark test narrows the gap to 21.8% (55 vs 43), but the NVIDIA card still holds the lead. The 3DMark Steel Nomad DX12 benchmark also favors the GTX 1080, with 1560 points to 1005, a 35.6% deficit for the AMD card.

The OpenCL benchmark reinforces the NVIDIA dominance. The GTX 1080 scores 51204 versus 37453, a 26.9% advantage. Even the 2D performance test, Passmark G2D, goes to NVIDIA by 13.4% (888 vs 769). In total, the GTX 1080 wins every traditional rasterization and compute-oriented test in the suite.

However, the AMD Radeon RX 580 has a remarkable counter-narrative in two specific benchmarks. The Geekbench Metal test shows the RX 580 scoring 45235 against the GTX 1080’s 23824, an enormous 89.9% advantage. This is the single largest win for either card in the entire head-to-head set. The RX 580 also wins Geekbench Vulkan decisively, scoring 45173 versus 30398, a 48.6% margin. These are not minor victories; they are the two highest raw scores the RX 580 produces in any test, and they indicate that the AMD card’s architecture is exceptionally well-optimized for these specific API paths.

FAQ

Q: Which card wins more head-to-head benchmarks?

A: The NVIDIA GeForce GTX 1080 wins 9 of the 11 head-to-head benchmarks, while the AMD Radeon RX 580 wins 2.

Q: What is the largest single benchmark win for either card?

A: The AMD Radeon RX 580’s 89.9% advantage in Geekbench Metal is the largest percentage delta in the entire head-to-head set, scoring 45235 against the GTX 1080’s 23824.

Q: How does the GTX 1080 fare in traditional DirectX performance?

A: The GTX 1080 wins every Passmark DirectX test, with the largest margin being 51.6% in DirectX 11 (124 vs 60) and the smallest being 21.8% in DirectX 12 (55 vs 43).

Q: Is the RX 580 competitive in compute workloads?

A: No, the data shows the GTX 1080 leads by 54.2% in Passmark GPU Compute (7614 vs 3488) and by 26.9% in Geekbench OpenCL (51204 vs 37453).

Q: What is the average benchmark score for each card?

A: The AMD Radeon RX 580 has an average benchmark score of 12928, while the NVIDIA GeForce GTX 1080 has an average score of 11960.

Q: How do the two cards compare in overall GPU percentile ranking?

A: The RX 580 sits at the 53rd percentile of all GPUs, while the GTX 1080 sits at the 51st percentile.

Architecture Differences

The two cards are built on fundamentally different process nodes and architectures. The AMD Radeon RX 580 uses the Polaris 20 chip on a 14 nm process from GlobalFoundries, employing the GCN 4.0 architecture. The NVIDIA GeForce GTX 1080 uses the GP104 chip on a 16 nm process from TSMC, based on the Pascal architecture. The transistor counts reflect this difference: the GTX 1080 packs 7,200 million transistors on a 314 mm² die, while the RX 580 has 5,700 million transistors on a smaller 232 mm² die. Interestingly, the RX 580 achieves a higher transistor density at 24.6M per mm², compared to the GTX 1080’s 22.9M per mm².

The compute resource allocation differs significantly. The GTX 1080 has 2560 shading units, 160 texture mapping units, and 64 ROPs. The RX 580 counters with 2304 shading units, 144 TMUs, and only 32 ROPs. This halving of ROPs is a critical architectural difference that explains the GTX 1080’s dominance in pixel-heavy workloads. The pixel rate for the GTX 1080 is 110.9 GPixel/s, more than double the RX 580’s 42.88 GPixel/s. Texture rate also favors NVIDIA, with 277.3 GTexel/s versus 193.0 GTexel/s.

The FP32 performance tells the same story: the GTX 1080 delivers 8.873 TFLOPS against the RX 580’s 6.175 TFLOPS. The FP16 comparison is starkly different in approach. The RX 580 provides FP16 at a 1:1 ratio with FP32, also at 6.175 TFLOPS, while the GTX 1080 provides only 138.6 GFLOPS at a 1:64 ratio. The RX 580’s 1:1 FP16 support is a notable architectural feature that does not appear to translate into benchmark wins in the tested suite, however.

Specification Differences

The clock speeds differ substantially, with the GTX 1080 running at a 1607 MHz base and 1733 MHz boost, compared to the RX 580’s 1257 MHz base and 1340 MHz boost. Memory configuration is another differentiator: both cards have 8 GB, but the GTX 1080 uses faster GDDR5X memory at 1251 MHz with 10 Gbps effective, achieving a bandwidth of 320.3 GB/s. The RX 580 uses GDDR5 at 2000 MHz with 8 Gbps effective, delivering 256.0 GB/s. The bus width is identical at 256 bits.

The GTX 1080 supports DirectX 12 (12_1) and Vulkan 1.4, while the RX 580 supports DirectX 12 (12_0) and Vulkan 1.3. Both support OpenGL 4.6. The physical dimensions differ, with the GTX 1080 measuring 267 mm in length, 112 mm in height, and 40 mm in width, while the RX 580 is shorter at 241 mm with no listed height or width. Both are dual-slot cards with a single 8-pin power connector and a suggested 450 W PSU. The display outputs differ, with the GTX 1080 offering 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a, while the RX 580 offers 1x HDMI 2.0b and 3x DisplayPort 1.4a. The RX 580 has a TDP of 185 W, slightly higher than the GTX 1080’s 180 W. The release dates are roughly a year apart, with the GTX 1080 launching in May 2016 and the RX 580 in April 2017.

The Verdict

The benchmark data presents a complex picture that defies a simple winner. The NVIDIA GeForce GTX 1080 is the clear victor in 9 of 11 tests, with particularly strong showings in Passmark GPU Compute (54.2% ahead), Passmark DirectX 11 (51.6% ahead), and Passmark G3D (43.5% ahead). Its average benchmark score of 11960, however, is lower than the RX 580’s 12928, and its percentile ranking of 51 is below the RX 580’s 53. The GTX 1080’s nearest rival is the NVIDIA GeForce GTX 960A at -0.3%, while the RX 580’s nearest rival is the NVIDIA GeForce RTX 3050 Ti Mobile at -0.1%. These aggregate statistics suggest the RX 580 performs better across the entire GPU universe, despite losing most head-to-head matchups.

The RX 580’s wins are not trivial. Its 89.9% advantage in Geekbench Metal and 48.6% advantage in Geekbench Vulkan are massive margins that indicate a specific strength in those API environments. The data suggests that the GTX 1080 is the superior card for traditional DirectX gaming and general compute, but the RX 580 has a distinct advantage in Metal and Vulkan workloads. Users who prioritize those specific APIs would find the RX 580 to be the stronger choice, while those running DirectX-based applications would see substantially better performance from the GTX 1080.

Where Each One Wins

The NVIDIA GeForce GTX 1080 wins in every scenario that relies on DirectX, regardless of version. It leads by 50.5% in DirectX 10, 51.6% in DirectX 11, and 21.8% in DirectX 12. Its compute performance is also superior, with a 54.2% lead in Passmark GPU Compute and a 26.9% lead in OpenCL. For 3D rendering as measured by Passmark G3D, the GTX 1080 is 43.5% faster. Even in the 3DMark Steel Nomad DX12 test, which is a modern workload, the GTX 1080 maintains a 35.6% lead. The GTX 1080 also wins the 2D test by 13.4%. This card is the choice for any application that uses DirectX or general-purpose GPU compute.

The AMD Radeon RX 580 wins specifically in Geekbench Metal and Geekbench Vulkan. The Metal score of 45235 is nearly double the GTX 1080’s 23824, and the Vulkan score of 45173 is 48.6% higher than the GTX 1080’s 30398. These are the only two tests where the RX 580 comes out ahead, but they are significant wins. The data indicates that the RX 580 is the superior card for Metal-based applications, which are common in Apple ecosystems, and for Vulkan-based software. Its 1:1 FP16 ratio and higher transistor density suggest an architecture that is well-suited to these specific API paths, even though it loses in the more traditional DirectX and OpenCL benchmarks.