AMD Radeon RX 7900 XT vs NVIDIA GeForce GTX 670 Comparison

The data presents a fascinating generational clash: the NVIDIA GeForce GTX 670 from 2012 versus the AMD Radeon RX 7900 XT from 2022. The benchmark results are not a simple sweep, but a complex story of API evolution and raw compute capability. This analysis will dissect the head-to-head scores, architectural chasms, and use-case implications.

Head-to-Head Benchmarks

The direct comparison in the FACT PACK reveals a split decision, with each card claiming one victory. In the Geekbench OpenCL test, the NVIDIA GeForce GTX 670 scores 15,341 against the AMD Radeon RX 7900 XT's 12,756. This gives the GTX 670 a 20.3% lead, a surprising result for a decade-old card. This suggests that the GTX 670's Kepler architecture excels at the specific compute workloads in this older API test, or that the test itself is not optimized for the newer RDNA 3.0 design.

The narrative flips dramatically in the Geekbench Vulkan test. Here, the AMD Radeon RX 7900 XT posts a colossal score of 71,068, while the GTX 670 manages only 15,553. The deltaPct of -78.1% indicates the RX 7900 XT is a massive 78.1% ahead. This is a stark illustration of modern API efficiency. The RX 7900 XT's Vulkan performance is in a completely different league, highlighting the immense architectural and driver advancements over the past decade. While the GTX 670 wins the OpenCL battle, the Vulkan result is so one-sided that it fundamentally defines the performance gap for modern workloads.

Looking at the broader context, the average benchmark scores tell a similar story of overall superiority for the newer card. The RX 7900 XT's average score is 13,745, which places it in the 55th percentile of all GPUs. The GTX 670's average is 12,773, placing it in the 52nd percentile. While the percentile difference seems small, the average score difference of nearly 1,000 points is significant. Furthermore, the RX 7900 XT's nearest rivals in the database include the AMD Radeon RX 9070 XT, where it is only 1.5% behind, whereas the GTX 670's closest competitor, the NVIDIA GeForce GTX 590, is 0.4% behind. This shows the RX 7900 XT is competing in a much higher performance tier, even if its average score is dragged down by less favorable results in certain legacy tests.

Architecture Differences

The gulf in performance is rooted in a complete architectural overhaul. The GTX 670 is built on the GK104 chip using the Kepler architecture, manufactured on a 28 nm process at TSMC. In contrast, the RX 7900 XT uses the Navi 31 chip with the RDNA 3.0 architecture, also from TSMC but on a much more advanced 5 nm node. This process shrink is a primary driver of the performance difference, allowing for a massive increase in transistor count. The GTX 670 has 3,540 million transistors on a 294 mm² die, while the RX 7900 XT packs 57,700 million transistors into a 529 mm² die. This results in a transistor density of 12.0M / mm² for the GTX 670 versus 109.1M / mm² for the RX 7900 XT, an almost 9x increase in density.

The GPU cores themselves have evolved. The GTX 670 has 1,344 shading units, 112 TMUs, and 32 ROPs. The RX 7900 XT scales this up considerably with 5,376 shading units, 336 TMUs, and 192 ROPs. More critically, the RX 7900 XT introduces 84 ray tracing cores, a feature entirely absent from the GTX 670. This is a fundamental capability difference that defines modern gaming. Clock speeds have also increased, with the GTX 670's base/boost of 915 MHz / 980 MHz looking ancient next to the RX 7900 XT's 1387 MHz / 2394 MHz (with a game clock of 2025 MHz). The memory subsystem is another chasm. The GTX 670 uses 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s of bandwidth. The RX 7900 XT uses 20 GB of GDDR6 on a 320-bit bus, delivering 800.0 GB/s. This 4x increase in bandwidth and 10x increase in capacity are critical for high-resolution textures and modern game engines.

Finally, the feature set reflects their respective eras. The GTX 670 supports PCIe 3.0, while the RX 7900 XT supports PCIe 4.0. The older card's API support tops out at DirectX 12 (11_0), whereas the RX 7900 XT supports DirectX 12 Ultimate (12_2). Both support OpenGL 4.6, but the RX 7900 XT supports Vulkan 1.4 compared to the GTX 670's 1.2.175.

Where Each One Wins

Based on the benchmark data, the use cases are clearly divided. The GTX 670 wins in the Geekbench OpenCL test. This could indicate an advantage in specific legacy compute applications or OpenCL-based workloads that do not leverage the newer features of the RX 7900 XT. Its lower power draw of 170W compared to 300W also suggests it might be a more practical option for very old systems with limited power supply capacity, though its performance is far lower overall.

The RX 7900 XT wins everywhere that matters for a modern user. Its 78.1% lead in the Vulkan benchmark is a decisive victory for modern gaming and compute APIs. The presence of 84 ray tracing cores makes it the only card of the two capable of hardware-accelerated ray tracing. Its massive 20 GB frame buffer and 800.0 GB/s of bandwidth make it suitable for 4K gaming, content creation, and AI workloads that require large datasets. The raw compute power is also evident in its 51.48 TFLOPS of FP32 performance versus the GTX 670's 2.634 TFLOPS. The RX 7900 XT also has a much higher pixel rate (459.6 GPixel/s vs 27.44 GPixel/s) and texture rate (804.4 GTexel/s vs 109.8 GTexel/s), making it vastly superior for filling high-resolution frames.

FAQ

Q: Is the NVIDIA GeForce GTX 670 faster than the AMD Radeon RX 7900 XT in any test?

A: Yes. In the Geekbench OpenCL benchmark, the GTX 670 scores 15,341 compared to the RX 7900 XT's 12,756, a lead of 20.3% for the older card.

Q: How much faster is the AMD Radeon RX 7900 XT in the Vulkan benchmark?

A: The RX 7900 XT is drastically faster, scoring 71,068 versus the GTX 670's 15,553. The data shows the RX 7900 XT is 78.1% ahead in this test.

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

A: The most significant difference is the inclusion of 84 ray tracing cores in the RX 7900 XT, a feature completely absent from the GTX 670. Additionally, the RX 7900 XT is built on a 5 nm process with 57,700 million transistors, versus the 28 nm process and 3,540 million transistors of the GTX 670.

Q: How do the memory specifications compare?

A: The RX 7900 XT has a 20 GB GDDR6 frame buffer on a 320-bit bus with 800.0 GB/s bandwidth. The GTX 670 has a 2 GB GDDR5 frame buffer on a 256-bit bus with 192.3 GB/s bandwidth.

Q: Based on average benchmark scores, which card has a higher overall performance rating?

A: The RX 7900 XT has a higher average benchmark score of 13,745 and sits in the 55th percentile of all GPUs. The GTX 670 has an average score of 12,773 and sits in the 52nd percentile.

Q: What are the power and connector requirements for each card?

A: The GTX 670 has a TDP of 170 W and requires 2x 6-pin power connectors with a suggested PSU of 450 W. The RX 7900 XT has a TDP of 300 W, requires 2x 8-pin connectors, and has a suggested PSU of 700 W.

The Verdict

The data paints an unambiguous picture for anyone building or upgrading a system today. The AMD Radeon RX 7900 XT is the superior choice for virtually any modern task. Its overwhelming 78.1% lead in Vulkan, a foundational API for contemporary games and applications, makes it the only viable option for high-end gaming. The inclusion of ray tracing cores and a 20 GB frame buffer future-proofs it for upcoming titles. The GTX 670's single OpenCL victory is an outlier, likely a quirk of legacy software, and does not offset the massive deficit in every other metric. The GTX 670's higher score in that one test would only matter for a niche user running very specific, ancient compute software.

The RX 7900 XT is also the clear winner in raw specifications, with over 15x the FP32 compute power (51.48 TFLOPS vs 2.634 TFLOPS) and over 4x the memory bandwidth. While it is a larger and more power-hungry card, the performance it delivers is on a different plane of existence. The GTX 670, with its 52nd percentile ranking, is a relic of a bygone era. Its 399 USD launch MSRP and 899 USD launch MSRP for the RX 7900 XT reflect their respective positions in the market at release. For any user, the choice is clear: the RX 7900 XT is the only card here that can handle modern gaming and compute workloads with authority.

Specification Differences

| Specification | NVIDIA GeForce GTX 670 | AMD Radeon RX 7900 XT |

| :--- | :--- | :--- |

| Chip | GK104 | Navi 31 |

| Architecture | Kepler | RDNA 3.0 |

| Process Node | 28 nm | 5 nm |

| Transistors | 3,540 million | 57,700 million |

| Die Size | 294 mm² | 529 mm² |

| Transistor Density | 12.0M / mm² | 109.1M / mm² |

| Base Clock | 915 MHz | 1387 MHz |

| Boost Clock | 980 MHz | 2394 MHz |

| Game Clock | N/A | 2025 MHz |

| Memory Size | 2 GB | 20 GB |

| Memory Type | GDDR5 | GDDR6 |

| Memory Bus | 256 bit | 320 bit |

| Bandwidth | 192.3 GB/s | 800.0 GB/s |

| Shading Units | 1344 | 5376 |

| TMUs | 112 | 336 |

| ROPs | 32 | 192 |

| Ray Tracing Cores | N/A | 84 |

| Pixel Rate | 27.44 GPixel/s | 459.6 GPixel/s |

| Texture Rate | 109.8 GTexel/s | 804.4 GTexel/s |

| FP32 Performance | 2.634 TFLOPS | 51.48 TFLOPS |

| FP16 Performance | N/A | 103.0 TFLOPS (2:1) |

| TDP | 170 W | 300 W |

| Power Connectors | 2x 6-pin | 2x 8-pin |

| Suggested PSU | 450 W | 700 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x16 |

| DirectX Support | 12 (11_0) | 12 Ultimate (12_2) |

| Vulkan Support | 1.2.175 | 1.4 |

| Display Outputs | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 | 1x HDMI 2.1a, 2x DisplayPort 2.1, 1x USB Type-C |

| Dimensions (L x H x W) | 241 mm x 111 mm x 38 mm | 276 mm x 110 mm x 51 mm |

| Launch MSRP | 399 USD | 899 USD |