AMD Radeon RX 6600S vs NVIDIA GeForce GTX 670 Comparison

The Verdict

The data presents a clear generational split. The AMD Radeon RX 6600S is the definitive winner in the only shared benchmark, delivering a 76.9% higher score in Geekbench OpenCL. The NVIDIA GeForce GTX 670 holds no benchmark wins in the head-to-head comparison. For any workload that relies on raw compute performance, the RX 6600S is the only logical choice. The GTX 670 is a legacy product, end-of-life since 2012, while the RX 6600S, though also end-of-life, represents a much newer design from 2022. The data suggests the GTX 670 should be considered only for compatibility with older systems, not for performance.

Architecture Differences

The two GPUs come from different architectural eras, and the measurements reflect that gap. The GTX 670 uses the Kepler architecture, built on a 28 nm process at TSMC with 3,540 million transistors on a 294 mm² die. The RX 6600S uses RDNA 2.0, built on a 7 nm process at TSMC, with 11,060 million transistors on a smaller 237 mm² die. This is a massive difference in transistor density: the RX 6600S packs 46.7M transistors per mm², versus 12.0M for the GTX 670. The newer process allows AMD to fit far more hardware in less space.

The compute resources differ substantially. The RX 6600S has 1792 shading units, 112 texture mapping units (TMUs), and 64 render output units (ROPs). The GTX 670 has 1,344 shading units, 112 TMUs, and only 32 ROPs. The RX 6600S also features 28 ray tracing (RT) cores, a capability the GTX 670 completely lacks. Memory technology is another major divide: the RX 6600S uses 4 GB of GDDR6 on a 128-bit bus with 224.0 GB/s of bandwidth, while the GTX 670 uses 2 GB of GDDR5 on a 256-bit bus with 192.3 GB/s. The RX 6600S has less bus width but newer, faster memory.

The API support tells a story of progression. The RX 6600S supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, which enables modern features like ray tracing. The GTX 670 only reaches DirectX 12 (11_0) and Vulkan 1.2.175. The GTX 670 is a PCIe 3.0 x16 card, while the RX 6600S is a PCIe 4.0 x8 device, indicating a mobile-first design. The RX 6600S is listed as an integrated graphics processor (IGP) with no power connectors, while the GTX 670 requires 2x 6-pin connectors and a 450 W power supply, drawing 170 W to the RX 6600S's 80 W.

Head-to-Head Benchmarks

The only direct comparison in the database is the Geekbench OpenCL test. The RX 6600S scores 66,435, while the GTX 670 scores 15,341. This is a difference of 51,094 points, or a 76.9% lead for the AMD card. The RX 6600S is not just ahead; it is in a different performance class. This single result shows a comprehensive victory for the newer architecture, likely due to its higher FP32 output of 7.168 TFLOPS versus 2.634 TFLOPS, and its texture rate of 224.0 GTexel/s versus 109.8 GTexel/s.

The GTX 670 has no benchmark wins in the database. Its only scores are in Geekbench Metal (7,424), OpenCL (15,341), and Vulkan (15,553). The RX 6600S has a wider test suite, including Passmark tests, but the only comparable test is OpenCL. In that single point of comparison, the RX 6600S is the clear winner. The recorded data shows no area where the GTX 670 comes out ahead.

Specification Differences

The two cards differ in nearly every physical and performance category.

• Process Node: The GTX 670 is on 28 nm, while the RX 6600S is on 7 nm.
• Transistors: 3,540 million for the GTX 670 vs. 11,060 million for the RX 6600S.
• Die Size: 294 mm² for the GTX 670 vs. 237 mm² for the RX 6600S.
• Base Clock: 915 MHz for the GTX 670 vs. 1,700 MHz for the RX 6600S.
• Boost Clock: 980 MHz for the GTX 670 vs. 2,000 MHz for the RX 6600S.
• Memory Clock: 1,502 MHz (6 Gbps effective) for the GTX 670 vs. 1,750 MHz (14 Gbps effective) for the RX 6600S.
• Memory Size: 2 GB for the GTX 670 vs. 4 GB for the RX 6600S.
• Memory Type: GDDR5 for the GTX 670 vs. GDDR6 for the RX 6600S.
• Memory Bus: 256 bit for the GTX 670 vs. 128 bit for the RX 6600S.
• Shading Units: 1,344 for the GTX 670 vs. 1,792 for the RX 6600S.
• TMUs: 112 for both cards (no difference).
• ROPs: 32 for the GTX 670 vs. 64 for the RX 6600S.
• RT Cores: None for the GTX 670 vs. 28 for the RX 6600S.
• Pixel Rate: 27.44 GPixel/s for the GTX 670 vs. 128.0 GPixel/s for the RX 6600S.
• Texture Rate: 109.8 GTexel/s for the GTX 670 vs. 224.0 GTexel/s for the RX 6600S.
• FP32: 2.634 TFLOPS for the GTX 670 vs. 7.168 TFLOPS for the RX 6600S.
• FP16: Not listed for the GTX 670 vs. 14.34 TFLOPS (2:1) for the RX 6600S.
• TDP: 170 W for the GTX 670 vs. 80 W for the RX 6600S.
• Slot: Dual-slot for the GTX 670 vs. IGP for the RX 6600S.
• Power Connectors: 2x 6-pin for the GTX 670 vs. None for the RX 6600S.
• Bus Interface: PCIe 3.0 x16 for the GTX 670 vs. PCIe 4.0 x8 for the RX 6600S.
• Display Outputs: 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 for the GTX 670 vs. Portable Device Dependent for the RX 6600S.
• DirectX: 12 (11_0) for the GTX 670 vs. 12 Ultimate (12_2) for the RX 6600S.
• Vulkan: 1.2.175 for the GTX 670 vs. 1.4 for the RX 6600S.
• Release Date: 2012-05-09 for the GTX 670 vs. 2022-01-03 for the RX 6600S.
• Dimensions: The GTX 670 is 241 mm long, 111 mm high, and 38 mm wide; the RX 6600S has no recorded dimensions.

FAQ

Q: Which GPU performs better in the Geekbench OpenCL test?

A: The AMD Radeon RX 6600S scores 66,435, which is 76.9% higher than the NVIDIA GeForce GTX 670's score of 15,341.

Q: Does the GTX 670 have any benchmark wins?

A: No. In the recorded head-to-head data, the GTX 670 has zero wins, while the RX 6600S has one win.

Q: What is the power consumption difference?

A: The GTX 670 has a TDP of 170 W, while the RX 6600S has a TDP of 80 W. The RX 6600S is significantly more power-efficient.

Q: Which card has more memory?

A: The RX 6600S has 4 GB of GDDR6, while the GTX 670 has 2 GB of GDDR5.

Q: Does the RX 6600S support ray tracing?

A: Yes, it has 28 ray tracing cores. The GTX 670 has no ray tracing hardware.

Q: What is the FP32 compute difference?

A: The RX 6600S achieves 7.168 TFLOPS, while the GTX 670 achieves 2.634 TFLOPS, a 2.72x gap.

Where Each One Wins

AMD Radeon RX 6600S wins on compute, features, and efficiency. The OpenCL score is the only direct test, and it is a dominant victory. The RX 6600S also has more shading units, more ROPs, higher clocks (2,000 MHz boost vs. 980 MHz), and a faster memory with higher bandwidth despite a narrower bus. It supports DirectX 12 Ultimate and Vulkan 1.4, enabling modern graphics APIs. It includes 28 ray tracing cores for future titles that use them. Its 7.168 TFLOPS FP32 and 14.34 TFLOPS FP16 make it far more capable for compute-heavy tasks. Its 80 W power draw is less than half the GTX 670's 170 W, and it does not need auxiliary power connectors, making it ideal for compact systems.

NVIDIA GeForce GTX 670 wins in no recorded benchmark. Its only advantages are historical or physical. It has a wider 256-bit memory bus, though the slower GDDR5 memory means lower bandwidth. It is a dual-slot discrete card with a specific length of 241 mm, so it fits in a desktop expansion slot. It has a lower transistor density, but that is not a performance metric. It supports PCIe 3.0 x16, which is older than the RX 6600S's PCIe 4.0 x8. The GTX 670's sole strength in the data is that it exists as a compatible part for older systems, but it is not a performance winner in any recorded test.