GPU Comparison

The AMD Radeon HD 6770 and NVIDIA GeForce 920MX represent two very different approaches to graphics processing, separated by five years of architectural evolution. The data shows a clear performance leader in the GeForce 920MX, which edges out the older card in the single available benchmark, but the comparison reveals a deeper story about how efficiency and feature support can outweigh raw silicon advantages. The HD 6770 was built for desktop power, while the 920MX was designed for mobile endurance, and their specifications reflect these divergent philosophies.

FAQ

Q: Which GPU scores higher in the Geekbench OpenCL test?

A: The NVIDIA GeForce 920MX scores 4068 points, which is 10.3% higher than the AMD Radeon HD 6770’s score of 3649 points. This makes the 920MX the winner in the only head-to-head benchmark available.

Q: How does the transistor density differ between these two chips?

A: The NVIDIA GeForce 920MX packs 1,020 million transistors into a 77 mm² die, resulting in a density of 13.2M / mm². The AMD Radeon HD 6770 has 1,040 million transistors spread across a much larger 166 mm² die, giving it a density of only 6.3M / mm², less than half the density of the NVIDIA chip.

Q: What is the thermal design power (TDP) for each card?

A: The AMD Radeon HD 6770 has a TDP of 108 W and requires a 1x 6-pin power connector with a suggested power supply of 300 W. The NVIDIA GeForce 920MX has a dramatically lower TDP of 16 W and requires no power connectors, making it suitable for MXM modules in portable devices.

Q: Which GPU supports more modern graphics APIs?

A: The NVIDIA GeForce 920MX supports Vulkan 1.4, DirectX 12 (11_0), and OpenGL 4.6. The AMD Radeon HD 6770 is limited to DirectX 11.2 (11_0) and OpenGL 4.4, with no Vulkan support listed in the data.

Q: How do the memory bandwidth figures compare between the two cards?

A: The AMD Radeon HD 6770 has a 128-bit memory bus with GDDR5 memory delivering 76.80 GB/s of bandwidth. The NVIDIA GeForce 920MX uses a 64-bit bus with DDR3 memory, providing only 14.40 GB/s, a substantial 62.40 GB/s difference in favor of the AMD card.

Q: What are the nearest rivals for each GPU based on average benchmark scores?

A: The HD 6770’s nearest rival is the NVIDIA GeForce GTX 1050, which scores 3629 (a 0.6% delta), while the 920MX’s closest competitor is the NVIDIA GeForce GT 545 at 3594 (a -1.8% delta). Both cards sit near the 22nd and 21st percentiles of all GPUs, respectively.

Architecture Differences

The architectural divide between these two GPUs is stark. The AMD Radeon HD 6770 uses the Juniper chip based on TeraScale 2 architecture, fabricated on a 40 nm process at TSMC. This older design employs 800 shading units, 40 texture mapping units (TMUs), and 16 raster operation pipelines (ROPs). It is part of the Northern Islands (HD 6700) generation, released in January 2011, with a predecessor in Evergreen and successor in Southern Islands. The chip is physically large at 166 mm², reflecting the less dense manufacturing technology of its era.

In contrast, the NVIDIA GeForce 920MX uses the GM108S chip built on Maxwell architecture, manufactured on a 28 nm process also at TSMC. This newer design has 256 shading units, 24 TMUs, and 8 ROPs, fewer of each compared to the AMD card. It belongs to the GeForce 900M generation, released in March 2016, with the GeForce 800M as predecessor and GeForce 10 Mobile as successor. The die size is just 77 mm², and while it has slightly fewer transistors (1,020 million vs. 1,040 million), it achieves a far higher transistor density of 13.2M / mm² compared to 6.3M / mm² on the older chip.

The feature sets also diverge significantly. The 920MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4, whereas the HD 6770 offers only DirectX 11.2 (11_0) and OpenGL 4.4 with no Vulkan support. The bus interface differs too: the HD 6770 uses PCIe 2.0 x16, while the 920MX uses PCIe 3.0 x8. Display outputs tell the story of their intended use, the HD 6770 has dual DVI, HDMI 1.3a, and two mini-DisplayPort 1.1 outputs, while the 920MX’s outputs are listed as "Portable Device Dependent," emphasizing its mobile laptop orientation.

Head-to-Head Benchmarks

The only direct benchmark comparison available is the Geekbench OpenCL test, and the results are intriguing. The NVIDIA GeForce 920MX scores 4068 points, while the AMD Radeon HD 6770 scores 3649 points. This represents a 10.3% advantage for the NVIDIA card, which is notable given the HD 6770’s substantial hardware advantages in several areas. The deltaPct of -10.3% (from the AMD card’s perspective) indicates a decisive win for the 920MX in compute workloads.

This result is surprising when considering the raw specifications. The HD 6770 has more than three times the shading units (800 vs. 256), a wider memory bus (128-bit vs. 64-bit), and over five times the memory bandwidth (76.80 GB/s vs. 14.40 GB/s). Its pixel rate of 13.60 GPixel/s and texture rate of 34.00 GTexel/s also exceed the 920MX’s 7.944 GPixel/s and 23.83 GTexel/s. The FP32 compute throughput of the HD 6770 is 1,360.0 GFLOPS, compared to just 508.4 GFLOPS for the 920MX, a 2.7x advantage for the older card.

Yet the benchmark tells a different story. The 920MX’s Maxwell architecture, with its newer instruction set and better compute efficiency, compensates for its smaller hardware footprint. The average benchmark score data shows the 920MX at 3528 points, while the HD 6770 sits at 3649 points, though the Geekbench OpenCL run reverses this relationship. The percentile ranks are nearly identical, the HD 6770 sits at the 22nd percentile of all GPUs, while the 920MX is at the 21st percentile, meaning both are entry-level performers by modern standards.

The Verdict

The data presents a nuanced picture. The NVIDIA GeForce 920MX wins the only head-to-head benchmark by 10.3%, and this victory is the single most important data point for determining the compute winner. However, the AMD Radeon HD 6770 remains competitive in specific hardware categories, particularly raw bandwidth and fill rates.

For users prioritizing modern API support and energy efficiency, the 920MX is clearly the better choice. It draws only 16 W compared to 108 W, requires no power connectors, and supports Vulkan 1.4 and DirectX 12. The 2 GB of DDR3 memory also doubles the HD 6770’s 1 GB frame buffer, which can help in texture-heavy workloads despite the lower bandwidth.

For users who need maximum memory bandwidth or higher pixel and texture fill rates, the HD 6770 retains advantages. Its 76.80 GB/s bandwidth is more than five times the 920MX’s 14.40 GB/s, and its 34.00 GTexel/s texture rate exceeds the 920MX’s 23.83 GTexel/s. The HD 6770 also has a 198 mm (7.8 inch) length and dual-slot design, indicating a desktop card with more physical presence.

The production status of both is end-of-life, so neither represents a future-proofing option. The HD 6770’s launch MSRP was 129 USD, but the 920MX has no listed launch price. The Geekbench OpenCL results suggest that if you must choose between these two for general compute tasks, the 920MX is the safer pick, but the HD 6770’s specific strengths should not be dismissed in bandwidth-limited scenarios.

Specification Differences

The two GPUs differ in nearly every measurable specification. The AMD Radeon HD 6770 uses a 40 nm process with 1,040 million transistors on a 166 mm² die, while the NVIDIA GeForce 920MX uses a 28 nm process with 1,020 million transistors on a 77 mm² die. The HD 6770 has 800 shading units, 40 TMUs, and 16 ROPs, versus the 920MX’s 256 shading units, 24 TMUs, and 8 ROPs.

Memory configurations are completely different: the HD 6770 has 1024 MB of GDDR5 on a 128-bit bus providing 76.80 GB/s, while the 920MX has 2 GB of DDR3 on a 64-bit bus providing 14.40 GB/s. Clock speeds show the HD 6770’s memory at 1200 MHz (4.8 Gbps effective), while the 920MX has a base clock of 965 MHz, boost clock of 993 MHz, and memory at 900 MHz (1800 Mbps effective).

Power requirements are polar opposites: the HD 6770 has a 108 W TDP, dual-slot width, a 1x 6-pin power connector, and a 300 W suggested PSU, while the 920MX has a 16 W TDP, MXM module form factor, no power connectors, and no suggested PSU. The bus interface also differs, with PCIe 2.0 x16 on the HD 6770 versus PCIe 3.0 x8 on the 920MX.

API support favors the newer card: the 920MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4, while the HD 6770 supports only DirectX 11.2 (11_0) and OpenGL 4.4. Display outputs are another area of divergence, the HD 6770 has 2x DVI, 1x HDMI 1.3a, and 2x mini-DisplayPort 1.1, while the 920MX’s outputs are "Portable Device Dependent." Finally, the HD 6770 has a length of 198 mm (7.8 inches), while the 920MX has no listed dimensions.

Where Each One Wins

The NVIDIA GeForce 920MX wins in compute benchmarks, energy efficiency, and modern software support. Its 4068 OpenCL score beats the HD 6770’s 3649 by 10.3%, and its 16 W TDP makes it far more suitable for battery-powered devices. The Vulkan 1.4 and DirectX 12 support ensure compatibility with newer applications, while the 2 GB memory capacity doubles the HD 6770’s allocation. The 920MX also wins on transistor density, with 13.2M / mm² versus 6.3M / mm², indicating a more efficient design.

The AMD Radeon HD 6770 wins in raw memory bandwidth, fill rates, and compute throughput. Its 76.80 GB/s bandwidth dwarfs the 920MX’s 14.40 GB/s, and its pixel rate of 13.60 GPixel/s is nearly double the 920MX’s 7.944 GPixel/s. The texture rate of 34.00 GTexel/s surpasses the 920MX’s 23.83 GTexel/s, and the FP32 performance of 1,360.0 GFLOPS is more than 2.5 times higher. The HD 6770 also has a wider memory bus (128-bit vs. 64-bit) and a larger physical footprint with dual-slot cooling, which may indicate better sustained performance in desktop environments.

For use cases, the 920MX is the obvious pick for thin-and-light laptops where power draw is critical and modern API support is needed. The HD 6770 would be better suited for legacy desktop systems where memory bandwidth and fill rate are the limiting factors, such as older games or applications that rely heavily on texture streaming. The benchmark data, however, tilts toward the 920MX for general-purpose compute, and this single OpenCL result is the most reliable indicator of real-world performance in the comparison.