AMD Radeon HD 7970M vs NVIDIA GeForce RTX 2070 SUPER Comparison

The NVIDIA GeForce RTX 2070 SUPER and the AMD Radeon HD 7970M represent two very different eras of GPU design. The RTX 2070 SUPER is a desktop-class card from 2019 built on Turing architecture, while the HD 7970M is a mobile GPU from 2012 based on GCN 1.0. The database records a single head-to-head benchmark, showing a massive generational gap. The RTX 2070 SUPER scores 83,358 in Geekbench OpenCL, while the HD 7970M scores 17,019, a delta of 389.8% in favor of the newer card. This page analyzes the architectural divide and what the recorded data means for both products.

FAQ

Q: Which GPU has the higher average benchmark score in the database?

A: The NVIDIA GeForce RTX 2070 SUPER has an average benchmark score of 20,282, while the AMD Radeon HD 7970M has an average score of 17,019.

Q: How large is the performance gap in the only shared benchmark?

A: In Geekbench OpenCL, the RTX 2070 SUPER scores 83,358 compared to the HD 7970M's 17,019, making the NVIDIA card 389.8% faster.

Q: What are the respective process nodes for these GPUs?

A: The RTX 2070 SUPER uses a 12 nm process, while the HD 7970M uses a larger 28 nm process, both fabricated by TSMC.

Q: How do the memory subsystems compare?

A: The RTX 2070 SUPER has 8 GB of GDDR6 on a 256-bit bus delivering 448.0 GB/s, whereas the HD 7970M has 2 GB of GDDR5 on a 256-bit bus delivering 153.6 GB/s.

Q: Which GPU has dedicated ray tracing or tensor cores?

A: Only the RTX 2070 SUPER has them, with 40 RT cores and 320 tensor cores. The HD 7970M has neither.

Q: What is the percentile rank of each card among all GPUs?

A: The RTX 2070 SUPER sits in the 65th percentile, while the HD 7970M is in the 60th percentile.

Architecture Differences

The architectural gap between these two GPUs is fundamental. The RTX 2070 SUPER is built on the Turing architecture, using the TU104 chip, while the HD 7970M uses the Wimbledon chip based on GCN 1.0. The manufacturing process differs significantly: the RTX 2070 SUPER is on a 12 nm node, the HD 7970M on 28 nm, both from TSMC. This translates to a transistor count of 13,600 million for the NVIDIA chip versus 2,800 million for the AMD chip, with die sizes of 545 mm² and 212 mm² respectively. The transistor density is 25.0M per mm² for the RTX 2070 SUPER versus 13.2M per mm² for the HD 7970M.

The compute resources are drastically different. The RTX 2070 SUPER has 2,560 shading units, 160 texture mapping units, and 64 ROPs. The HD 7970M has half the shading units at 1,280, half the TMUs at 80, and half the ROPs at 32. The NVIDIA card also adds dedicated hardware that the AMD card completely lacks: 40 RT cores for ray tracing and 320 tensor cores for AI workloads. These features are absent from the GCN 1.0 architecture, which predates them.

Clock speeds also tell a story. The RTX 2070 SUPER has a base clock of 1605 MHz and a boost clock of 1770 MHz. The HD 7970M has no recorded base or boost clock in the database, only a memory clock of 1200 MHz (4.8 Gbps effective). The memory technology differs as well: GDDR6 at 1750 MHz (14 Gbps effective) on the RTX 2070 SUPER versus GDDR5 at 1200 MHz on the HD 7970M. The memory bandwidth is 448.0 GB/s versus 153.6 GB/s, nearly a threefold difference.

The API support also diverges. The RTX 2070 SUPER supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The HD 7970M supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. The power and physical specifications highlight their different intended uses: the RTX 2070 SUPER has a TDP of 215 W, is a dual-slot card, and requires 1x 6-pin plus 1x 8-pin power connectors with a suggested 550 W power supply. The HD 7970M has a TDP of 100 W, is an MXM module with no power connectors, and relies on the host system for power delivery, typical for a mobile part.

Head-to-Head Benchmarks

The database records only one direct benchmark comparison between these two GPUs: Geekbench OpenCL. The results are decisive. The RTX 2070 SUPER scores 83,358, while the HD 7970M scores 17,019. This yields a delta of 389.8%, meaning the RTX 2070 SUPER is nearly five times faster in this compute workload. This single result accounts for the entire head-to-head record, with one win for the NVIDIA card and zero for the AMD card.

The magnitude of this difference is explained by the architectural advances discussed above. The RTX 2070 SUPER has twice the shading units, more than double the texture rate (283.2 GTexel/s versus 68.00 GTexel/s), and more than four times the pixel rate (113.3 GPixel/s versus 27.20 GPixel/s). Its FP32 throughput of 9.062 TFLOPS dwarfs the HD 7970M's 2.176 TFLOPS. The FP16 performance of 18.12 TFLOPS on the RTX 2070 SUPER has no equivalent on the HD 7970M, which has no recorded FP16 figure.

The average benchmark scores in the database also reflect this gap, though less dramatically because they include different test sets. The RTX 2070 SUPER averages 20,282 across multiple benchmarks, including Passmark tests (G3D score of 18,169, G2D score of 878, GPU compute score of 7,557) and 3DMark Steel Nomad DX12 (1,651). The HD 7970M's average of 17,019 is based solely on its Geekbench OpenCL result, as no other benchmark data is recorded for it.

It is worth noting the nearest rivals for each card. The RTX 2070 SUPER sits within 1.5% of the NVIDIA Quadro M4000M (20,480, delta -1%), the RTX 3070 Mobile (20,534, delta -1.2%), the Intel Arc B570 (20,556, delta -1.3%), and the Intel Arc A750 (20,582, delta -1.5%). The HD 7970M is similarly close to the GTX 690 (17,037, delta -0.1%), the RX 7600 XT (17,083, delta -0.4%), and the Tesla M4 (16,932, delta 0.5%). This suggests that while the RTX 2070 SUPER is a stronger card overall, the HD 7970M holds its own relative to its contemporaries, even if it cannot compete with the newer Turing architecture.

The Verdict

The data points to one clear conclusion: the RTX 2070 SUPER is in a different performance class entirely. Its 389.8% lead in Geekbench OpenCL is not incremental, it is transformative. The RTX 2070 SUPER is the only choice for anyone requiring modern features like ray tracing (40 RT cores) or AI acceleration (320 tensor cores). Its 8 GB of GDDR6 memory with 448.0 GB/s bandwidth also positions it for contemporary workloads, whereas the HD 7970M's 2 GB GDDR5 and 153.6 GB/s bandwidth are dated.

For users with older systems or specific mobile requirements, the HD 7970M still has a role. Its 100 W TDP and MXM form factor make it a low-power, modular solution for portable systems, and its 60th percentile ranking among all GPUs shows it was a capable part in its era. However, the RTX 2070 SUPER's 65th percentile ranking, combined with a higher average score (20,282 versus 17,019), indicates it outperforms the older card even in a broader context.

The production status of both cards is end-of-life, but the RTX 2070 SUPER launched in July 2019 with a successor already in the GeForce 30 series, while the HD 7970M's lineage goes back to 2012 with the Solar System as its successor. The RTX 2070 SUPER supports DirectX 12 Ultimate and Vulkan 1.4, ensuring compatibility with modern APIs, while the HD 7970M tops out at DirectX 12 (11_1) and Vulkan 1.2.170.

The verdict is straightforward: the RTX 2070 SUPER is the superior GPU for any compute or gaming task in the database. The HD 7970M is a historical product, best suited for legacy systems or as a low-power mobile solution where its 100 W TDP and lack of external power connectors are advantages. The recorded data offers no scenario where the HD 7970M edges out the RTX 2070 SUPER in raw performance.

Specification Differences

| Specification | NVIDIA GeForce RTX 2070 SUPER | AMD Radeon HD 7970M |

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

| Architecture | Turing | GCN 1.0 |

| Chip | TU104 | Wimbledon |

| Process Node | 12 nm | 28 nm |

| Transistors | 13,600 million | 2,800 million |

| Die Size | 545 mm² | 212 mm² |

| Transistor Density | 25.0M / mm² | 13.2M / mm² |

| Memory Size | 8 GB | 2 GB |

| Memory Type | GDDR6 | GDDR5 |

| Memory Clock | 1750 MHz (14 Gbps effective) | 1200 MHz (4.8 Gbps effective) |

| Memory Bandwidth | 448.0 GB/s | 153.6 GB/s |

| Shading Units | 2560 | 1280 |

| TMUs | 160 | 80 |

| ROPs | 64 | 32 |

| RT Cores | 40 | None |

| Tensor Cores | 320 | None |

| Pixel Rate | 113.3 GPixel/s | 27.20 GPixel/s |

| Texture Rate | 283.2 GTexel/s | 68.00 GTexel/s |

| FP32 | 9.062 TFLOPS | 2.176 TFLOPS |

| FP16 | 18.12 TFLOPS (2:1) | None |

| TDP | 215 W | 100 W |

| Slot Width | Dual-slot | MXM Module |

| Power Connectors | 1x 6-pin + 1x 8-pin | None |

| Suggested PSU | 550 W | None |

| Bus Interface | PCIe 3.0 x16 | MXM-B (3.0) |

| Display Outputs | 1x HDMI 2.0, 3x DisplayPort 1.4a, 1x USB Type-C | Portable Device Dependent |

| DirectX | 12 Ultimate (12_2) | 12 (11_1) |

| Vulkan | 1.4 | 1.2.170 |

| Dimensions | 267 mm (10.5 inches) x 116 mm (4.6 inches) x 35 mm (1.4 inches) | None recorded |

| Release Date | 2019-07-08 | 2012-04-23 |

| Predecessor | GeForce 10 | Vancouver |

| Successor | GeForce 30 | Solar System |

| Launch MSRP | 499 USD | None |