NVIDIA GeForce GTX 550 Ti vs NVIDIA GeForce GTX 670MX Comparison

The NVIDIA GeForce GTX 550 Ti and the NVIDIA GeForce GTX 670MX are both end-of-life graphics processors, but they represent distinct design philosophies from different mobile and desktop generations. The GTX 550 Ti is a desktop card from the GeForce 500 series, built on the Fermi 2.0 architecture with a 40 nm process, while the GTX 670MX is a mobile part from the GeForce 600M series, using the Kepler architecture on a 28 nm process. In the sole head-to-head benchmark available—Geekbench OpenCL—the GTX 670MX scores 6.4% higher than the GTX 550 Ti (6125 vs 5731). However, their average benchmark scores across all recorded tests are nearly identical, with the GTX 550 Ti at 5731 and the GTX 670MX at 5721, a difference of just 0.2%. This close overall parity, despite the GTX 670MX’s single-test victory, makes the choice between them less about raw performance and more about the context of their intended use cases.

The Verdict

From the data, the GTX 670MX is the clear winner in the only direct benchmark comparison, posting a 6.4% higher OpenCL score (6125 vs 5731). This suggests that for compute-oriented workloads that leverage OpenCL, the Kepler-based mobile chip holds a distinct advantage. The GTX 670MX also offers a significantly larger memory pool of 3 GB compared to the GTX 550 Ti’s 1 GB, which could be decisive for applications that require large frame buffers or datasets, even if its memory bandwidth is lower (67.20 GB/s vs 98.50 GB/s). However, the overall average scores tell a different story: the GTX 550 Ti’s average of 5731 is actually 0.2% higher than the GTX 670MX’s 5721, indicating that the desktop card is not universally slower. Therefore, the GTX 670MX is the pick for users prioritizing raw compute throughput and memory capacity, while the GTX 550 Ti might be preferred in scenarios where its higher memory bandwidth and desktop-oriented design (with a 210 mm length and dual-slot cooler) are more relevant. Given that the GTX 670MX is a mobile part with no power connectors and a 75 W TDP, versus the GTX 550 Ti’s 116 W TDP and 1x 6-pin connector, the GTX 670MX is also the more power-efficient option, making it suitable for laptops where thermal and power constraints are tighter.

FAQ

Q: Which GPU has a higher score in the Geekbench OpenCL benchmark?

A: The NVIDIA GeForce GTX 670MX wins the Geekbench OpenCL test with a score of 6125, which is 6.4% higher than the GeForce GTX 550 Ti’s score of 5731.

Q: How do the two GPUs compare in their average benchmark scores?

A: The GeForce GTX 550 Ti has an average benchmark score of 5731, while the GeForce GTX 670MX has an average score of 5721. This makes the GTX 550 Ti 0.2% faster on average, a negligible difference.

Q: What are the key differences in memory specifications?

A: The GTX 670MX has triple the memory capacity (3 GB vs 1 GB) but lower bandwidth (67.20 GB/s vs 98.50 GB/s). Both use GDDR5 memory and have a 192-bit bus width.

Q: Which GPU is more power-efficient according to the TDP figures?

A: The GTX 670MX has a TDP of 75 W, which is significantly lower than the GTX 550 Ti’s 116 W TDP. The GTX 670MX also requires no power connectors, whereas the GTX 550 Ti needs a 1x 6-pin connector.

Q: Are these GPUs from the same architecture family?

A: No, they are not. The GTX 550 Ti uses the Fermi 2.0 architecture with the GF116 chip, while the GTX 670MX uses the Kepler architecture with the GK104 chip.

Q: What is the difference in their launch dates and process nodes?

A: The GTX 550 Ti was released on 2011-03-14 and uses a 40 nm process, while the GTX 670MX was released later on 2012-09-30 and uses a smaller 28 nm process.

Architecture Differences

The two GPUs are built on fundamentally different architectures, which explains many of their performance and feature disparities. The GTX 550 Ti is based on the Fermi 2.0 architecture, utilizing the GF116 chip, which is a mature design from the GeForce 500 series. In contrast, the GTX 670MX employs the Kepler architecture with the GK104 chip, representing a newer generation from the GeForce 600M series. This architectural shift is evident in the transistor counts: the GK104 packs 3,540 million transistors on a 294 mm² die, while the GF116 has only 1,170 million transistors on a 238 mm² die. The Kepler chip achieves a much higher transistor density of 12.0M / mm² compared to Fermi’s 4.9M / mm², thanks to the move from a 40 nm to a 28 nm process node at TSMC.

The Kepler architecture also brings a significant increase in shading units, with the GTX 670MX featuring 960 shading units and 80 texture mapping units (TMUs), versus the GTX 550 Ti’s 192 shading units and 32 TMUs. Both GPUs have 24 ROPs, but the GTX 670MX’s higher clock speeds and more numerous cores allow it to achieve higher pixel and texture rates: 12.02 GPixel/s and 48.08 GTexel/s, respectively, compared to the GTX 550 Ti’s 7.200 GPixel/s and 28.80 GTexel/s. In terms of API support, the GTX 670MX adds Vulkan 1.2.175 support, which the GTX 550 Ti lacks, although both support DirectX 12 (11_0) and OpenGL 4.6. The GTX 670MX also supports PCIe 3.0 x16, while the GTX 550 Ti is limited to PCIe 2.0 x16.

Specification Differences

The specification sheets for the two cards reveal several key differences beyond the architecture. The most obvious is the memory configuration: the GTX 670MX has 3 GB of GDDR5, while the GTX 550 Ti has only 1024 MB. Both have a 192-bit memory bus, but the GTX 550 Ti’s memory runs at 1026 MHz (4.1 Gbps effective), yielding a bandwidth of 98.50 GB/s, whereas the GTX 670MX’s memory runs at 700 MHz (2.8 Gbps effective), resulting in a lower bandwidth of 67.20 GB/s. The clock speeds also differ, with the GTX 670MX having a base and boost clock of 601 MHz, while the GTX 550 Ti has no listed base or boost clock, only a memory clock.

The physical and power characteristics are starkly contrasted. The GTX 550 Ti is a dual-slot desktop card with a length of 210 mm (8.3 inches), requiring a 1x 6-pin power connector and a 300 W suggested PSU. Its TDP is 116 W. The GTX 670MX, being a mobile part, has no listed dimensions or slot width, uses no power connectors, and has a lower TDP of 75 W. The GTX 550 Ti offers display outputs of 2x DVI and 1x mini-HDMI 1.3a, while the GTX 670MX’s outputs are described as "Portable Device Dependent." Finally, the GTX 550 Ti has a launch MSRP of 149 USD, while the GTX 670MX has no listed MSRP.

Head-to-Head Benchmarks

The only direct benchmark comparison available is the Geekbench OpenCL test, and the results are decisively in favor of the GTX 670MX. The GTX 670MX scores 6125, while the GTX 550 Ti scores 5731, giving the mobile Kepler part a 6.4% advantage. This is a substantial lead, especially considering that the two GPUs have nearly identical average scores across all benchmarks. The GTX 670MX’s victory in OpenCL is likely driven by its far higher number of shading units (960 vs 192), which are critical for parallel compute workloads. The deltaPct of -6.4% in the head-to-head data indicates that the GTX 550 Ti is behind by that margin.

Interestingly, the nearestRivals data shows that the GTX 550 Ti’s average score of 5731 is 0.2% higher than the GTX 670MX’s average of 5721. This means that while the GTX 670MX wins the OpenCL test outright, the overall performance picture is much closer. The GTX 550 Ti also has a 0.3% edge over the Intel Iris Pro Graphics P6300 (5712) and a 0.7% edge over the AMD Radeon HD 8790M (5691), while the AMD Radeon R7 M465 (5841) is 1.9% faster. For the GTX 670MX, its nearest rival list shows it trailing the GTX 550 Ti by 0.2% in average score, but it leads the Intel Iris Pro Graphics P6300 by 0.2% and the AMD Radeon HD 8790M by 0.5%. The NVIDIA Quadro M500M (5604) is 2.1% behind the GTX 670MX.

Where Each One Wins

Based strictly on the benchmark data, the GTX 670MX wins in the compute-heavy OpenCL workload, making it the better choice for tasks that rely on general-purpose GPU computing, such as OpenCL-accelerated applications or certain content creation tools. Its 6.4% higher score in this test, combined with its 3 GB memory capacity, suggests it can handle larger datasets and more complex parallel operations than the GTX 550 Ti. The GTX 670MX also wins on power efficiency, with a 75 W TDP versus 116 W, making it the superior option for laptop users who need sustained performance without excessive heat or battery drain.

The GTX 550 Ti, while losing the OpenCL head-to-head, has its own strengths. Its average benchmark score of 5731 is slightly higher than the GTX 670MX’s 5721, indicating that in other, non-OpenCL workloads, it may perform on par or better. The GTX 550 Ti also has significantly higher memory bandwidth (98.50 GB/s vs 67.20 GB/s), which can be advantageous in scenarios that are bandwidth-bound rather than compute-bound, such as certain gaming or texture-heavy workloads. Additionally, the GTX 550 Ti is a desktop card with a dedicated 6-pin power connector and a larger physical footprint, which might make it more suitable for users building a desktop system with a 300 W PSU, as opposed to the mobile-oriented GTX 670MX. The GTX 550 Ti also has a lower transistor density (4.9M / mm²) and a smaller die, but these are architectural facts rather than performance wins. Ultimately, the GTX 670MX is the winner for OpenCL compute and power-sensitive applications, while the GTX 550 Ti holds its ground in average performance and memory bandwidth.