AMD Radeon HD 7770M vs NVIDIA GeForce GTX 670 Comparison

The AMD Radeon HD 7770M and NVIDIA GeForce GTX 670 are two very different products from the same era, and the benchmark data reflects that starkly. In the single head-to-head benchmark available, the AMD mobile part delivers a decisive victory, but the NVIDIA desktop card offers a broader feature set and a different performance profile. This analysis breaks down the data to show which GPU wins where and why.

Head-to-Head Benchmarks

The only direct benchmark comparison in the data is the Geekbench Metal test, and the result is a landslide victory for the AMD Radeon HD 7770M. The AMD card scores 13536, while the NVIDIA GeForce GTX 670 scores 7424. This translates to an 82.3% performance advantage for the AMD part, a massive gap that defines the entire head-to-head comparison. The AMD Radeon HD 7770M wins the sole benchmark contest, giving it a clean 1 win against 0 for the GTX 670.

This result is particularly striking because the GTX 670 is a desktop card with significantly more raw hardware on paper. The data shows that in this specific Metal-based workload, the AMD architecture is overwhelmingly superior. The AMD Radeon HD 7770M’s score of 13536 places it in the 54th percentile of all GPUs, while the GTX 670’s 7424 score in this same test puts it in a much lower position, despite its overall average benchmark score being 12773 across multiple tests.

It is important to note that the GTX 670 has a broader benchmark profile, with additional scores in Geekbench OpenCL (15341) and Geekbench Vulkan (15553), which are not available for the AMD card. While the AMD card wins the Metal test decisively, the NVIDIA card demonstrates competitive performance in other API environments, suggesting that the 82.3% delta is specific to Metal rather than a universal performance rule.

Architecture Differences

The two GPUs are built on fundamentally different architectures from rival manufacturers. The AMD Radeon HD 7770M uses the GCN 1.0 architecture, built on the "Chelsea" chip, while the NVIDIA GeForce GTX 670 uses the Kepler architecture, built on the "GK104" chip. Both are manufactured by TSMC on a 28 nm process node, which is their only major manufacturing similarity.

The transistor counts and die sizes reveal a significant scale difference. The AMD chip contains 1,500 million transistors on a 123 mm² die, resulting in a transistor density of 12.2M / mm². The NVIDIA chip is substantially larger, packing 3,540 million transistors onto a 294 mm² die, with a slightly lower density of 12.0M / mm². This means the GTX 670 has over twice the transistor count and a much larger physical footprint.

The memory subsystems are also vastly different. The AMD Radeon HD 7770M has 1024 MB of GDDR5 memory on a 128 bit bus, delivering 64.00 GB/s of bandwidth. The GTX 670 doubles the memory capacity to 2 GB and doubles the bus width to 256 bit, resulting in a much higher bandwidth of 192.3 GB/s, which is three times higher than the AMD part.

Core configurations further differentiate the two. The AMD card has 512 shading units, 32 texture mapping units (TMUs), and 16 render output units (ROPs). The NVIDIA card has 1344 shading units, 112 TMUs, and 32 ROPs, which are all significantly higher numbers. This translates to higher theoretical fill rates for the NVIDIA card, with a pixel rate of 27.44 GPixel/s and a texture rate of 109.8 GTexel/s, compared to the AMD’s 10.80 GPixel/s and 21.60 GTexel/s. The floating-point performance follows suit, with the GTX 670 offering 2.634 TFLOPS of FP32 compute versus the AMD’s 691.2 GFLOPS.

Power consumption is another major architectural difference. The AMD Radeon HD 7770M is a mobile part with a TDP of just 32 W, while the NVIDIA GeForce GTX 670 is a desktop card with a TDP of 170 W. The GTX 670 also requires two 6-pin power connectors and a 450 W suggested power supply, while the AMD card’s power connectors are listed as "Portable Device Dependent."

Where Each One Wins

The benchmark data points to a clear split in use cases. The AMD Radeon HD 7770M wins in the Geekbench Metal test, which is indicative of performance in Apple’s Metal graphics API. With a score of 13536 and an 82.3% lead over the GTX 670, this card is the clear choice for workloads that are optimized for Metal, such as certain creative applications and games on macOS platforms. Its low 32 W TDP also makes it suitable for thin and light portable devices, where power efficiency is paramount.

The NVIDIA GeForce GTX 670 wins in the broader sense of raw compute and traditional desktop workloads. Its higher scores in Geekbench OpenCL (15341) and Geekbench Vulkan (15553) show that it is a more versatile performer across different APIs. The data shows it has a higher average benchmark score of 12773 across all its tests, compared to the AMD’s single-test average of 13536. However, this average is skewed by the AMD’s single strong Metal result. The GTX 670’s massive advantage in memory bandwidth (192.3 GB/s), texture fill rate (109.8 GTexel/s), and pixel rate (27.44 GPixel/s) suggests it is better suited for high-resolution gaming and heavy 3D rendering tasks where these resources are critical.

The NVIDIA card also has a higher transistor count (3,540 million) and more shading units (1344), indicating a greater capacity for parallel processing in non-Metal workloads. For a desktop user with a power supply capable of handling 170 W and a need for multi-API support, the GTX 670 is the more robust choice. The AMD card, on the other hand, is the winner for mobile users who require strong Metal performance without the power draw.

FAQ

Q: Which GPU has the higher Geekbench Metal score?

A: The AMD Radeon HD 7770M has a significantly higher Geekbench Metal score of 13536, compared to the NVIDIA GeForce GTX 670’s score of 7424, giving the AMD card an 82.3% lead.

Q: Does the NVIDIA GeForce GTX 670 have more memory bandwidth?

A: Yes. The GTX 670 has a memory bandwidth of 192.3 GB/s, which is exactly three times the 64.00 GB/s of the AMD Radeon HD 7770M.

Q: What is the transistor count difference between the two GPUs?

A: The NVIDIA GeForce GTX 670 has 3,540 million transistors, while the AMD Radeon HD 7770M has 1,500 million transistors. The GTX 670 has over twice as many.

Q: Are both GPUs built on the same manufacturing process?

A: Yes, both the AMD Radeon HD 7770M and the NVIDIA GeForce GTX 670 are built on a 28 nm process node by TSMC.

Q: Which GPU has a higher TDP and what does this imply?

A: The NVIDIA GeForce GTX 670 has a TDP of 170 W, while the AMD Radeon HD 7770M has a TDP of 32 W. This implies the GTX 670 is a desktop card requiring substantial cooling and power, whereas the AMD card is a highly efficient mobile solution.

Q: What is the average benchmark score for each GPU?

A: The AMD Radeon HD 7770M has an average benchmark score of 13536, while the NVIDIA GeForce GTX 670 has an average score of 12773.

The Verdict

The data presents a clear verdict for different audiences. The AMD Radeon HD 7770M is the winner for anyone prioritizing Metal API performance. Its benchmark score of 13536 is 82.3% higher than the GTX 670’s in this specific test, and its low 32 W TDP makes it the only viable choice for portable devices. The data shows it is an end-of-life mobile chip that still holds a strong position in the 54th percentile of all GPUs, largely due to this Metal performance.

The NVIDIA GeForce GTX 670 is the winner for desktop users who need a multi-API workhorse. While it loses the Metal test, its superior scores in OpenCL (15341) and Vulkan (15553) show broader compatibility. Its massive resources, including 3,540 million transistors, 2 GB of memory, and 192.3 GB/s of bandwidth, make it a more capable card for traditional gaming and compute tasks that do not rely on Metal. The GTX 670’s average score of 12773 is lower than the AMD’s, but this is because it is being measured across three different tests, whereas the AMD only has one benchmark result.

In short, pick the AMD Radeon HD 7770M for a power-efficient, Metal-centric mobile setup. Pick the NVIDIA GeForce GTX 670 for a desktop system that needs to handle a variety of graphics APIs with high raw throughput. The data does not support a single overall winner; it supports two winners for two distinct scenarios.

Specification Differences

The following specifications differ between the AMD Radeon HD 7770M and the NVIDIA GeForce GTX 670.

• Chip: Chelsea (AMD) vs GK104 (NVIDIA)
• Architecture: GCN 1.0 (AMD) vs Kepler (NVIDIA)
• Generation: London (HD 7700M) (AMD) vs GeForce 600 (NVIDIA)
• Transistors: 1,500 million (AMD) vs 3,540 million (NVIDIA)
• Die Size: 123 mm² (AMD) vs 294 mm² (NVIDIA)
• Transistor Density: 12.2M / mm² (AMD) vs 12.0M / mm² (NVIDIA)
• Memory Clocks: 1000 MHz / 4 Gbps effective (AMD) vs 1502 MHz / 6 Gbps effective (NVIDIA)
• Base Clock: Not specified (AMD) vs 915 MHz (NVIDIA)
• Boost Clock: Not specified (AMD) vs 980 MHz (NVIDIA)
• Memory Size: 1024 MB (AMD) vs 2 GB (NVIDIA)
• Memory Bus Width: 128 bit (AMD) vs 256 bit (NVIDIA)
• Memory Bandwidth: 64.00 GB/s (AMD) vs 192.3 GB/s (NVIDIA)
• Shading Units: 512 (AMD) vs 1344 (NVIDIA)
• TMUs: 32 (AMD) vs 112 (NVIDIA)
• ROPs: 16 (AMD) vs 32 (NVIDIA)
• Pixel Rate: 10.80 GPixel/s (AMD) vs 27.44 GPixel/s (NVIDIA)
• Texture Rate: 21.60 GTexel/s (AMD) vs 109.8 GTexel/s (NVIDIA)
• FP32 Performance: 691.2 GFLOPS (AMD) vs 2.634 TFLOPS (NVIDIA)
• TDP: 32 W (AMD) vs 170 W (NVIDIA)
• Slot Width: Not specified (AMD) vs Dual-slot (NVIDIA)
• Power Connectors: Not specified (AMD) vs 2x 6-pin (NVIDIA)
• Suggested PSU: Not specified (AMD) vs 450 W (NVIDIA)
• Bus Interface: PCIe 2.0 x16 (AMD) vs PCIe 3.0 x16 (NVIDIA)
• Display Outputs: Portable Device Dependent (AMD) vs 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 (NVIDIA)
• DirectX Version: 12 (11_1) (AMD) vs 12 (11_0) (NVIDIA)
• Vulkan Version: 1.2.170 (AMD) vs 1.2.175 (NVIDIA)
• Dimensions: Not specified (AMD) vs 241 mm length, 111 mm height, 38 mm width (NVIDIA)
• Release Date: 2012-04-23 (AMD) vs 2012-05-09 (NVIDIA)
• Predecessor: Vancouver (AMD) vs GeForce 500 (NVIDIA)
• Successor: Solar System (AMD) vs GeForce 700 (NVIDIA)
• Launch MSRP: Not specified (AMD) vs 399 USD (NVIDIA)