AMD Radeon R9 370X vs AMD Radeon RX 460 Comparison

# The Verdict - who should pick which, strictly from the data

The database places the AMD Radeon RX 460 and the AMD Radeon R9 370X in different performance tiers. The RX 460 has an average benchmark score of 18,373 across Metal, OpenCL, and Vulkan tests, putting it at the 62nd percentile of all GPUs. The R9 370X averages 15,862, which places it at the 58th percentile. That 4 point gap in percentiles is modest, but the raw score difference is not trivial: the RX 460 leads by about 2,511 points, or roughly 16% in average compute performance.

For buyers who focus on modern API workloads, the RX 460 is the clear pick. In the Metal benchmark, it scores 17,065 against the R9 370X's 12,676, a 34.6% advantage. The RX 460 also wins the Vulkan test by a wide margin, scoring 20,198 to the older card's 9,018, a 124% lead. The R9 370X has its own territory, however. In the OpenCL test, the older architecture posts 25,893 versus 17,855, a 31% lead. It is a split decision: two wins for the RX 460, one for the R9 370X.

If the priority is the newer API generation, or particularly Vulkan, the RX 460 is the better part. If the workload is primarily OpenCL compute, the R9 370X has the edge. The data does not support the RX 460 as a universal victor.

# Head-to-Head Benchmarks

The recorded benchmark results split cleanly along API lines.

Metal: RX 460 scores 17,065, R9 370X scores 12,676. That is a 34.6% win for the RX 460. This is the largest margin in any test. The RX 460's newer GCN design and higher boost clock (1,090 MHz base, 1,200 MHz boost) are reflected here. The R9 370X runs at 980 MHz base, 1,030 MHz boost.

OpenCL: the tables turn. The R9 370X scores 25,893, the RX 460 scores 17,855. The R9 370X leads by 31%. This is a strong result for the older card. The R9 370X uses a 256-bit memory bus and GDDR5 at 1,400 MHz, giving 179.2 GB/s of bandwidth. The RX 460 has a 128-bit bus and 1,750 MHz memory, for 112.0 GB/s. The wider bus and higher memory bandwidth of the R9 370X is the likely reason for its OpenCL advantage.

Vulkan: the RX 460 comes back strongly, scoring 20,198 to 9,018. That is a 124% lead. Vulkan is a low-level API that rewards raw geometry and driver efficiency. The RX 460's newer architecture (GCN 4.0 versus GCN 1.0) and higher pixel rate (19.20 GPixel/s versus 32.96 GPixel/s, note this is a reversal in the raw specification) let it dominate here. The R9 370X runs at 32.96 GPixel/s, but the RX 460 still wins by a large margin, which suggests that the newer API path is substantially more efficient on this card.

Geekbench Metal: RX 460 wins by 34.6%.

Geekbench OpenCL: R9 370X wins by 31%.

Geekbench Vulkan: RX 460 wins by 124%.

# FAQ

Q: I mostly play Vulkan games and do some OpenCL compute. Which card is better?

A: The RX 460. It wins Vulkan by 124% and is only 0.1% behind the R9 370X in OpenCL. The Metal result (34.6% ahead) also shows that the RX 460 is the stronger card for newer API paths.

Q: I run OpenCL workloads. Which of the two should I buy?

A: The R9 370X is the right choice. It wins OpenCL by 31%, and this is its only benchmark win. Its 256-bit memory bus and 179.2 GB/s bandwidth are the likely reasons.

Q: Which GPU is the better all-rounder?

A: The RX 460. It wins two of the three benchmarks, one by 34.6% and one by 124%. The only test the R9 370X wins is OpenCL, where the R9 370X has a 31% lead.

Q: The RX 460 has higher average and percentile. Why not just pick it?

A: The R9 370X remains competitive in OpenCL, and the RX 460 leads the average score by only 16%. If your specific application is OpenCL-heavy, the R9 370X is the better part. The RX 460's wins are in the newer API tests.

# Specification Differences

The two cards differ in nearly every major internal specification.

| Field | AMD Radeon RX 460 | AMD Radeon R9 370X |

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

| Chip | Baffin | Trinidad |

| Architecture | GCN 4.0 | GCN 1.0 |

| Process node | 14 nm | 28 nm |

| Foundry | GlobalFoundries | TSMC |

| Transistors | 3,000 million | 2,800 million |

| Die size | 123 mm² | 212 mm² |

| Transistor density | 24.4M / mm² | 13.2M / mm² |

| Base clock | 1,090 MHz | 980 MHz |

| Boost clock | 1,200 MHz | 1,030 MHz |

| Memory | 2 GB GDDR5 | 2 GB GDDR5 |

| Memory bus | 128 bit | 256 bit |

| Memory clock | 1,750 MHz | 1,400 MHz |

| Bandwidth | 112.0 GB/s | 179.2 GB/s |

| Shading units | 896 | 1,280 |

| TMUs | 56 | 80 |

| ROPs | 16 | 32 |

| Pixel rate | 19.20 GPixel/s | 32.96 GPixel/s |

| Texture rate | 67.20 GTexel/s | 82.40 GTexel/s |

| FP32 | 2.150 TFLOPS | 2.637 TFLOPS |

| FP16 | 2.150 TFLOPS (1:1) | none |

| TDP | 75 W | 180 W |

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

| Power connectors | None | 2x 6-pin |

| Suggested PSU | 250 W | 450 W |

| Bus interface | PCIe 3.0 x8 | PCIe 3.0 x16 |

| Display outputs | 1x DVI, 1x HDMI 2.0b, 1x DisplayPort 1.4a | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 |

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

| OpenGL | 4.6 | 4.6 |

| Vulkan | 1.3 | 1.2.170 |

The RX 460 is built on a 14 nm process at GlobalFoundries, while the R9 370X is a 28 nm part from TSMC. The RX 460 packs 3,000 million transistors into 123 mm² (24.4M per mm²), the R9 370X has 2,800 million transistors on a 212 mm² die (13.2M per mm²). That is a major density advantage for the RX 460.

Memory is the biggest functional split. Both cards have 2 GB, but the RX 460 uses a 128-bit bus at 1,750 MHz, for 112.0 GB/s. The R9 370X has a 256-bit bus at 1,400 MHz, for 179.2 GB/s. That 67.2 GB/s bandwidth advantage for the R9 370X is the largest specification gap between the two.

The R9 370X also has more compute resources: 1,280 shading units, 80 TMUs, 32 ROPs, versus 896, 56, and 16 for the RX 460. It also has higher peak pixel and texture fill rates. The RX 460 counters with much higher boost clocks (1,200 vs 1,030 MHz), a smaller die, newer architecture, and lower TDP.

Power is a major difference. The RX 460 is a 75 W card with no power connectors. The R9 370X is rated at 180 W and requires two 6-pin connectors. The RX 460 also asks for a 250 W PSU, the R9 370X for 450 W.

# Where Each One Wins

Choose the RX 460 if you are building a small-form-factor system, want to run current and future Vulkan workloads, or need a low-power card that needs no extra power cables. It wins the Metal and Vulkan tests, and takes the average score.

Choose the R9 370X if your work is OpenCL compute, or you want the wider memory bus and higher memory bandwidth. It wins the OpenCL test and has the larger bandwidth by 67.2 GB/s.

The two cards are not close in the data, the choice is a real fork in the road. The newer RX 460 has architectural and API advantages; the older R9 370X has the memory subsystem and OpenCL compute advantage. Pick the card that matches your main use.