AMD Radeon RX 9070 XT vs NVIDIA GeForce GTX 670 Comparison

The AMD Radeon RX 9070 XT and NVIDIA GeForce GTX 670 represent two distinct eras of GPU design, separated by over a decade of architectural evolution. The data shows a decisive performance gap, with the RX 9070 XT leading in every shared benchmark, yet the GTX 670 holds historical significance as a 28nm Kepler part. This analysis compares their benchmark scores, specifications, and architectural foundations strictly from the provided facts.

FAQ

Q: How does the AMD Radeon RX 9070 XT compare to the NVIDIA GeForce GTX 670 in the Geekbench OpenCL benchmark?

A: The RX 9070 XT scores 17,428 points, which is 13.6% higher than the GTX 670's 15,341 points. This indicates a clear but not overwhelming lead for the newer AMD card in this compute-oriented test.

Q: What is the most significant performance difference between the two cards?

A: In the Geekbench Vulkan test, the RX 9070 XT scores 65,879 points versus 15,553 for the GTX 670, a 323.6% advantage. This massive delta highlights the RX 9070 XT's modern API support and architectural efficiency compared to the older Kepler design.

Q: Which card has a higher average benchmark score across all tests?

A: The RX 9070 XT has an average benchmark score of 13,543, while the GTX 670 averages 12,773. This places the RX 9070 XT in the 55th percentile of all GPUs, compared to the GTX 670's 52nd percentile.

Q: Are there any benchmark tests where the GTX 670 outperforms the RX 9070 XT?

A: No. In the head-to-head comparison data, the RX 9070 XT wins both available tests (Geekbench OpenCL and Vulkan). The GTX 670 has no recorded wins against the RX 9070 XT.

Q: What is the memory configuration difference between these two cards?

A: The RX 9070 XT features 16 GB of GDDR6 memory on a 256-bit bus, delivering 644.6 GB/s bandwidth. The GTX 670 has 2 GB of GDDR5 memory on a 256-bit bus, providing 192.3 GB/s bandwidth. This represents a significant capacity and bandwidth advantage for the AMD card.

Q: How do their production statuses and release dates differ?

A: The RX 9070 XT is listed as "Active" with a release date of 2025-03-05, while the GTX 670 is "End-of-life" and was released on 2012-05-09. The GTX 670's predecessor is the GeForce 500 series, and its successor is the GeForce 700 series, whereas the RX 9070 XT's predecessor is Navi III with no successor listed.

The Verdict

The benchmark data is unambiguous: the AMD Radeon RX 9070 XT is the superior performer in every measured category. Its average benchmark score of 13,543 exceeds the GTX 670's 12,773 by roughly 6%, and its Vulkan score is over four times higher. For any modern workload, from gaming to compute, the RX 9070 XT is the only viable choice based on performance alone.

The GTX 670, however, occupies a niche for retro or legacy system builders. Its 28nm Kepler architecture, 2 GB GDDR5 memory, and end-of-life status mean it cannot handle contemporary titles or demanding applications. The data shows it remains competitive in OpenCL (15,341 vs 17,428) relative to its age, but it lacks Vulkan efficiency and modern feature support. If a user requires a card for period-accurate 2012-era gaming or has a power constraint (170W TDP vs 304W), the GTX 670 could be considered, but the RX 9070 XT dominates on raw metrics.

Head-to-Head Benchmarks

The head-to-head results show two distinct performance profiles. In Geekbench OpenCL, the RX 9070 XT scores 17,428 against the GTX 670's 15,341, a 13.6% advantage. This test measures general-purpose compute performance, and the delta suggests the RX 9070 XT's 4nm process and RDNA 4.0 architecture provide a solid, but not transformative, lead in this specific workload. The GTX 670's Kepler architecture, while old, still handles OpenCL tasks reasonably well given its 2.634 TFLOPS FP32 throughput.

The Vulkan result is where the generational gap becomes stark. The RX 9070 XT achieves 65,879 points, while the GTX 670 manages only 15,553—a 323.6% difference. Vulkan is a modern low-overhead API, and the GTX 670's support is limited to version 1.2.175, whereas the RX 9070 XT supports Vulkan 1.4. This benchmark reflects not just raw compute power but also driver optimization and hardware design for contemporary APIs. The RX 9070 XT's 64 RT cores and 4096 shading units, compared to the GTX 670's 1344 shading units and no RT cores, explain this massive gap.

Outside the head-to-head, the RX 9070 XT has a broader benchmark portfolio. It scores 7,260 in 3DMark Steel Nomad DX12, 26,795 in Passmark G3D, and 15,864 in Passmark GPU Compute. The GTX 670 only has Geekbench scores (Metal, OpenCL, Vulkan), with no DirectX or Passmark results listed. This lack of data for the older card makes a full comparison impossible, but the available evidence consistently favors the AMD part.

Specification Differences

The specifications reveal fundamental design philosophy changes. The RX 9070 XT uses a 4nm process with 53,900 million transistors on a 357 mm² die, yielding a transistor density of 151.0M per mm². The GTX 670 uses a 28nm process with 3,540 million transistors on a 294 mm² die, for a density of 12.0M per mm². This 12.6x density advantage for the RX 9070 XT is the root cause of its performance lead.

Clock speeds differ substantially. The RX 9070 XT has a base clock of 1660 MHz and a boost of 2970 MHz, with a game clock of 2400 MHz. The GTX 670 runs at 915 MHz base and 980 MHz boost. Memory clocks also diverge: 2518 MHz (20.1 Gbps effective) for the AMD card versus 1502 MHz (6 Gbps effective) for the NVIDIA card.

Memory capacity and bandwidth are major differentiators. The RX 9070 XT offers 16 GB GDDR6 with 644.6 GB/s bandwidth, while the GTX 670 provides 2 GB GDDR5 with 192.3 GB/s. Both use a 256-bit bus, but the newer memory technology and higher clock speed give the RX 9070 XT over 3.3x the bandwidth.

Compute resources scale accordingly. The RX 9070 XT has 4096 shading units, 256 TMUs, and 128 ROPs, with pixel rate of 380.2 GPixel/s and texture rate of 760.3 GTexel/s. The GTX 670 has 1344 shading units, 112 TMUs, and 32 ROPs, with pixel rate of 27.44 GPixel/s and texture rate of 109.8 GTexel/s. FP32 performance is 48.66 TFLOPS for the AMD card versus 2.634 TFLOPS for the NVIDIA card.

Power and connectivity also differ. The RX 9070 XT has a 304W TDP with 2x 8-pin connectors and a 700W suggested PSU. The GTX 670 has a 170W TDP with 2x 6-pin connectors and a 450W suggested PSU. Display outputs favor the newer card: 1x HDMI 2.1b and 3x DisplayPort 2.1a versus 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. Bus interface is PCIe 5.0 x16 for the RX 9070 XT and PCIe 3.0 x16 for the GTX 670.

Architecture Differences

The architectural divide is generational. The RX 9070 XT uses RDNA 4.0 on the Navi 48 chip, part of the Navi IV (RX 9000) series. The GTX 670 uses Kepler on the GK104 chip, from the GeForce 600 generation. Both are manufactured by TSMC, but at vastly different nodes: 4nm versus 28nm.

The RX 9070 XT includes 64 RT cores for hardware-accelerated ray tracing, a feature entirely absent from the GTX 670. It also supports DirectX 12 Ultimate (12_2), while the GTX 670 is limited to DirectX 12 (11_0). API support extends to Vulkan 1.4 on the AMD card versus 1.2.175 on the NVIDIA card, and OpenGL 4.6 on both.

Transistor count is the clearest architectural metric: 53,900 million for the RX 9070 XT versus 3,540 million for the GTX 670, a 15.2x difference. This reflects not just process improvements but also the addition of dedicated RT cores and a more complex shader array. The RX 9070 XT's FP16 performance matches its FP32 at 48.66 TFLOPS (1:1), while the GTX 670 has no listed FP16 capability.

The GTX 670's dimensions are specified at 241 mm length, 111 mm height, and 38 mm width, whereas the RX 9070 XT's dimensions are null in the data. Both are dual-slot cards, but the power delivery systems differ: 2x 8-pin for AMD versus 2x 6-pin for NVIDIA, reflecting the higher TDP of the newer card.

Where Each One Wins

The RX 9070 XT wins in every benchmark category where both cards have data. Its 323.6% Vulkan advantage makes it the clear choice for any modern API-based workload, including current game engines and compute applications that leverage Vulkan. The 13.6% OpenCL lead, while smaller, still favors the AMD card for general-purpose compute. Its 16 GB memory capacity and 644.6 GB/s bandwidth are essential for 4K textures and large datasets, areas where the 2 GB GTX 670 would immediately bottleneck. The RX 9070 XT's 64 RT cores enable ray-traced effects that the GTX 670 cannot render at all.

The GTX 670's only advantages are operational, not performance-based. Its 170W TDP means lower power consumption and heat output, making it suitable for older power supplies or small form factor builds. The 450W suggested PSU requirement versus 700W for the RX 9070 XT allows installation in legacy systems. Its 2x 6-pin connectors are more compatible with older PSUs, and its 2x DVI outputs support older monitors without adapters. For a retro build targeting 2012-era DirectX 11 games, the GTX 670's 2 GB memory and 27.44 GPixel/s pixel rate are adequate, though the RX 9070 XT's 380.2 GPixel/s would still crush it.

In terms of market positioning, the RX 9070 XT sits in the 55th percentile of all GPUs, with rivals like the GTX 570 (delta 0.2%) and HD 7770M (delta 0.1%) nearby, indicating it is a mid-range performer on the global scale. The GTX 670 is in the 52nd percentile, with rivals like the GTX 590 (delta -0.4%) and RX 7600M XT (delta 0.5%). This suggests both cards occupy similar relative tiers in their respective eras, but the absolute scores show the RX 9070 XT is dramatically faster.