AMD Radeon R9 370X vs NVIDIA GeForce GTX 680 Comparison

Head-to-Head Benchmarks

The benchmark data between the AMD Radeon R9 370X and the NVIDIA GeForce GTX 680 is split sharply by API. The AMD card wins two of the three recorded tests, while the NVIDIA card takes the third by a substantial margin. The overall average benchmark score favors the AMD Radeon R9 370X at 15,862 versus 14,150 for the GeForce GTX 680, a difference of roughly 12%. That aggregate figure, however, hides the fact that the two cards dominate in completely different workloads.

In Geekbench Metal, the R9 370X scores 12,676 against 7,897 for the GTX 680. That is a 60.5% advantage for the AMD card, the largest delta in any test. The Metal result is particularly notable because it is not a close contest; the AMD architecture delivers a performance level that the older NVIDIA card cannot approach in this API. Similarly, in Geekbench OpenCL, the R9 370X posts 25,893 versus 16,906, a 53.2% lead. These two wins are decisive, and they drive the AMD card’s higher average score.

The Geekbench Vulkan test flips the narrative completely. Here, the GTX 680 scores 17,646 against 9,018 for the R9 370X, giving the NVIDIA card a 48.9% advantage. This is a massive reversal. The Vulkan result shows that the GTX 680 is not simply slower across the board; in the right API environment, it outpaces the AMD card by nearly half. The wins tally is 2-1 in favor of AMD, but the magnitude of the NVIDIA Vulkan victory is comparable to the magnitude of the AMD Metal and OpenCL victories.

Putting the scores into context with the database’s nearest-rival data helps clarify where each card sits. The R9 370X’s average score of 15,862 places it within 0.1% of the AMD Radeon RX 7700 (15,852), and it is 1.2% ahead of the AMD Radeon Pro W5500 (15,679). It trails the AMD Radeon RX 9060 by 1% (16,014) and the NVIDIA GeForce RTX 3060 Ti by 1.7% (16,129). The GTX 680’s average of 14,150 sits 0.9% above the NVIDIA Tesla K10 (14,029), but it is 0.9% behind the NVIDIA GeForce GTX 1070 Ti (14,277), 1.2% behind the Intel Iris Xe MAX Graphics (14,315), and 1.4% behind the AMD Radeon Vega 11 (14,352). The percentile rankings confirm the separation: the R9 370X sits at the 58th percentile among all GPUs, while the GTX 680 sits at the 55th percentile.

Architecture Differences

The two cards come from different architectural generations and different design philosophies. The AMD Radeon R9 370X uses the Trinidad chip based on GCN 1.0 architecture, belonging to the Pirate Islands (R9 300) generation. The NVIDIA GeForce GTX 680 uses the GK104 chip based on Kepler architecture, from the GeForce 600 generation. Both are built on the same 28 nm TSMC process, but the transistor counts and die sizes differ considerably. The GTX 680 packs 3,540 million transistors on a 294 mm² die, while the R9 370X has 2,800 million transistors on a smaller 212 mm² die. The transistor density is slightly higher on the AMD card: 13.2M per mm² versus 12.0M per mm² for NVIDIA.

The compute resources are configured differently. The GTX 680 has more shading units (1,536 versus 1,280) and more texture mapping units (128 versus 80), but both cards have 32 ROPs. The clock speeds favor NVIDIA as well: the GTX 680 runs at a base of 1006 MHz and a boost of 1058 MHz, while the R9 370X runs at 980 MHz base and 1030 MHz boost. Memory clocks are close, with the GTX 680 at 1502 MHz (6 Gbps effective) and the R9 370X at 1400 MHz (5.6 Gbps effective). Both use 2 GB of GDDR5 on a 256-bit bus, but the bandwidth differs: 192.3 GB/s for NVIDIA versus 179.2 GB/s for AMD.

These architectural differences show up in the raw throughput numbers. The GTX 680 achieves a pixel rate of 33.86 GPixel/s versus 32.96 GPixel/s for the R9 370X, and a texture rate of 135.4 GTexel/s versus 82.40 GTexel/s. The FP32 compute is also higher on NVIDIA: 3.250 TFLOPS versus 2.637 TFLOPS. The GTX 680 is the more powerful part on paper in every traditional rasterization metric. Yet the benchmark results show AMD winning in Metal and OpenCL. That suggests the GCN 1.0 architecture handles those specific APIs more efficiently, despite having lower peak throughput. The Kepler architecture, meanwhile, excels in Vulkan, where its higher shading unit count and texture rate translate into a 48.9% benchmark advantage.

The API support levels also differ. The R9 370X supports DirectX 12 (11_1), while the GTX 680 supports DirectX 12 (11_0). Both support OpenGL 4.6, but the Vulkan version differs slightly: 1.2.170 for AMD and 1.2.175 for NVIDIA. The power draw is lower on the AMD card at 180 W TDP versus 195 W for NVIDIA, though both use dual-slot coolers and require two 6-pin power connectors, with both suggesting a 450 W power supply. The physical dimensions are similar in height (both 111 mm), but the GTX 680 is longer at 256 mm (10.1 inches) versus 221 mm (8.7 inches) for the R9 370X, and the GTX 680 has a listed width of 38 mm (1.5 inches) while the R9 370X does not have a recorded width.

FAQ

Q: Which card has the higher average benchmark score?

A: The AMD Radeon R9 370X has an average benchmark score of 15,862, which is 12% higher than the NVIDIA GeForce GTX 680’s 14,150.

Q: In which API does the NVIDIA GeForce GTX 680 outperform the AMD Radeon R9 370X?

A: The GTX 680 wins the Geekbench Vulkan test with a score of 17,646 versus 9,018 for the R9 370X, a 48.9% advantage.

Q: How large is the AMD Radeon R9 370X’s lead in Metal and OpenCL?

A: In Metal, the R9 370X scores 12,676 versus 7,897, a 60.5% lead. In OpenCL, it scores 25,893 versus 16,906, a 53.2% lead.

Q: Do both cards use the same memory configuration?

A: Both use 2 GB of GDDR5 on a 256-bit bus, but the GTX 680 has a higher bandwidth of 192.3 GB/s compared to 179.2 GB/s for the R9 370X.

Q: What are the transistor counts for each card?

A: The GTX 680 has 3,540 million transistors on a 294 mm² die, while the R9 370X has 2,800 million transistors on a 212 mm² die.

Q: Which card has a higher percentile ranking among all GPUs?

A: The R9 370X sits at the 58th percentile, while the GTX 680 sits at the 55th percentile.

The Verdict

The data supports a clear split decision based on workload. The AMD Radeon R9 370X is the stronger card in Metal and OpenCL environments, with leads of 60.5% and 53.2% respectively. Its average benchmark score is higher, its percentile ranking is better, and it achieves this with a lower TDP of 180 W versus 195 W. The GTX 680, however, is the definitive choice for Vulkan workloads, where it leads by 48.9%. The GTX 680 also has superior raw specifications: more shading units, more TMUs, higher clocks, higher bandwidth, and higher FP32 throughput.

For a user prioritizing modern compute APIs like Metal or OpenCL, the R9 370X is the clear winner. For a user specifically targeting Vulkan performance, the GTX 680 is the answer. The overall average score favors AMD, but the Vulkan result is too large to ignore. The R9 370X wins two tests decisively, the GTX 680 wins one test decisively, and the average score reflects the 2-1 split. There is no universal winner here; the choice depends entirely on which API the workload uses.

Specification Differences

| Specification | AMD Radeon R9 370X | NVIDIA GeForce GTX 680 |

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

| Chip | Trinidad | GK104 |

| Architecture | GCN 1.0 | Kepler |

| Generation | Pirate Islands (R9 300) | GeForce 600 |

| Transistors | 2,800 million | 3,540 million |

| Die Size | 212 mm² | 294 mm² |

| Transistor Density | 13.2M / mm² | 12.0M / mm² |

| Base Clock | 980 MHz | 1006 MHz |

| Boost Clock | 1030 MHz | 1058 MHz |

| Memory Clock | 1400 MHz (5.6 Gbps effective) | 1502 MHz (6 Gbps effective) |

| Memory Bandwidth | 179.2 GB/s | 192.3 GB/s |

| Shading Units | 1280 | 1536 |

| TMUs | 80 | 128 |

| ROPs | 32 | 32 |

| Pixel Rate | 32.96 GPixel/s | 33.86 GPixel/s |

| Texture Rate | 82.40 GTexel/s | 135.4 GTexel/s |

| FP32 | 2.637 TFLOPS | 3.250 TFLOPS |

| TDP | 180 W | 195 W |

| DirectX | 12 (11_1) | 12 (11_0) |

| Vulkan | 1.2.170 | 1.2.175 |

| Length | 221 mm (8.7 inches) | 256 mm (10.1 inches) |

| Width | Not recorded | 38 mm (1.5 inches) |

| Release Date | 2015-08-26 | 2012-03-21 |

| Launch MSRP | 199 USD | 499 USD |

Where Each One Wins

The AMD Radeon R9 370X wins in Metal and OpenCL. Its Metal score of 12,676 is 60.5% higher than the GTX 680’s 7,897, and its OpenCL score of 25,893 is 53.2% higher than 16,906. These are the two tests where the AMD card demonstrates its strength. The R9 370X also has a lower TDP (180 W versus 195 W), a shorter card length (221 mm versus 256 mm), and a higher transistor density (13.2M per mm² versus 12.0M per mm²). Its higher average benchmark score (15,862 versus 14,150) and better percentile ranking (58th versus 55th) reinforce its standing as the better overall performer in this comparison.

The NVIDIA GeForce GTX 680 wins in Vulkan. Its score of 17,646 is 48.9% higher than the R9 370X’s 9,018. The GTX 680 also holds superiority in raw specifications: more shading units (1,536 versus 1,280), more TMUs (128 versus 80), higher base and boost clocks, higher memory bandwidth (192.3 GB/s versus 179.2 GB/s), higher pixel rate (33.86 GPixel/s versus 32.96 GPixel/s), higher texture rate (135.4 GTexel/s versus 82.40 GTexel/s), and higher FP32 throughput (3.250 TFLOPS versus 2.637 TFLOPS). For any workload that leverages Vulkan, the GTX 680 is the stronger option. The data also shows the GTX 680 has a longer history, with a release date of 2012-03-21 versus 2015-08-26 for the R9 370X, and it carries a higher launch MSRP of 499 USD compared to 199 USD for the AMD card.