AMD Radeon HD 8790M vs NVIDIA GeForce GTX 550 Ti Comparison

# NVIDIA GeForce GTX 550 Ti vs AMD Radeon HD 8790M

The NVIDIA GeForce GTX 550 Ti and AMD Radeon HD 8790M represent two very different approaches to GPU design, separated by two years of architectural evolution. The GTX 550 Ti, a desktop card from 2011 built on Fermi 2.0, squares off against the HD 8790M, a 2013 mobile module built on GCN 1.0. Benchmark data shows the GTX 550 Ti holds a 0.7% lead in average score (5731 vs 5691), but the HD 8790M counters with a Vulkan score the NVIDIA card cannot match, since the GTX 550 Ti has no Vulkan support listed. Both cards sit at the 33rd percentile among all GPUs, placing them in the same performance tier despite their architectural differences.

Where Each One Wins

The GTX 550 Ti wins the only shared benchmark outright. In Geekbench OpenCL, it scores 5731 against the HD 8790M's 5017, a 14.2% advantage. That is not a marginal gap; it is a decisive margin in compute workloads that scale with shading units and memory bandwidth. The GTX 550 Ti's 192 shading units, 32 TMUs, and 24 ROPs, paired with a 192-bit memory bus delivering 98.50 GB/s, give it a structural edge in OpenCL tasks that stress raw throughput. The HD 8790M, despite having 384 shading units (exactly double), cannot translate that count into an OpenCL win because its 128-bit bus and 64.00 GB/s bandwidth bottleneck data movement.

The HD 8790M's win comes in API coverage rather than raw scores. It supports Vulkan 1.2.170 and scores 6365 in Geekbench Vulkan, a test the GTX 550 Ti cannot run due to no Vulkan support in its feature set. The HD 8790M also supports DirectX 12 (11_1) versus the GTX 550 Ti's DirectX 12 (11_0), and both cards support OpenGL 4.6. For modern workloads that leverage Vulkan, the AMD card is the only option between these two. The GTX 550 Ti's OpenCL win is substantial, but it is a single-test victory; the HD 8790M's Vulkan capability opens a whole API surface the NVIDIA card simply does not have.

The HD 8790M also wins on memory capacity with 2 GB versus 1024 MB, which matters for texture-heavy workloads and larger frame buffers. The GTX 550 Ti wins on bandwidth by 34.50 GB/s, but the AMD card has double the capacity. In GPU compute, the OpenCL result favors NVIDIA, but in modern gaming or compute APIs, the AMD card's Vulkan support and larger memory pool give it a functional edge that no single benchmark in the data can fully capture.

Architecture Differences

The two GPUs come from different foundry nodes and design philosophies. The GTX 550 Ti uses the GF116 chip on a 40 nm TSMC process, packing 1,170 million transistors into a 238 mm² die, yielding a transistor density of 4.9M per mm². The HD 8790M uses the Mars chip on a 28 nm TSMC process, with 950 million transistors in just 77 mm², achieving 12.3M transistors per mm². The smaller node gives AMD a density advantage of 7.4M transistors per mm², allowing more logic in less space, but the NVIDIA chip has 220 million more transistors overall.

Clock behavior differs significantly. The GTX 550 Ti has no base or boost clock listed, only a memory clock of 1026 MHz (4.1 Gbps effective). The HD 8790M runs a base clock of 850 MHz with a boost clock of 900 MHz, and memory at 1000 MHz (4 Gbps effective). The AMD card's boost capability is a feature the NVIDIA card lacks entirely, and its memory clock is slightly lower in effective Gbps (4.0 vs 4.1).

The compute architectures are generations apart. The GTX 550 Ti uses Fermi 2.0, while the HD 8790M uses GCN 1.0. Fermi 2.0 relies on 192 shading units, 32 TMUs, and 24 ROPs, while GCN 1.0 uses 384 shading units but only 24 TMUs and 8 ROPs. The pixel rates are identical at 7.200 GPixel/s, but the texture rates differ: the GTX 550 Ti hits 28.80 GTexel/s versus the HD 8790M's 21.60 GTexel/s. FP32 performance is also identical at 691.2 GFLOPS, a coincidence driven by the GTX 550 Ti's higher clock offsetting the HD 8790M's extra shading units.

Memory subsystems reflect their target markets. The GTX 550 Ti uses a 192-bit bus with 1024 MB GDDR5 and 98.50 GB/s bandwidth; the HD 8790M uses a 128-bit bus with 2 GB GDDR5 and 64.00 GB/s. The NVIDIA card was designed for desktop PCIe 2.0 x16 slots with a 116 W TDP and a 6-pin power connector; the AMD card is an MXM-A (3.0) module with no power connectors, drawing power through the portable device interface. Display outputs also differ: the GTX 550 Ti offers 2x DVI and 1x mini-HDMI 1.3a, while the HD 8790M's outputs are listed as "Portable Device Dependent."

The Verdict

The data points to a clear split: the GTX 550 Ti is the better OpenCL performer, and the HD 8790M is the better modern-API card. In the only head-to-head benchmark, the GTX 550 Ti beats the HD 8790M by 14.2% in Geekbench OpenCL, a substantial margin that reflects its superior texture rate (28.80 vs 21.60 GTexel/s) and bandwidth (98.50 vs 64.00 GB/s). If your workload is OpenCL compute, the NVIDIA card is the data-backed choice.

The HD 8790M, however, supports Vulkan 1.2.170 and scores 6365 in Geekbench Vulkan, a test the GTX 550 Ti cannot participate in because it has no Vulkan support listed. The AMD card also offers DirectX 12 (11_1) versus the GTX 550 Ti's 11_0, and doubles the memory to 2 GB. For users running Vulkan-based applications or needing a larger frame buffer, the HD 8790M is the only viable choice between these two.

The nearest-rival data shows how close these cards are in aggregate. The GTX 550 Ti's nearest rivals include the GTX 670MX (5721, 0.2% behind), the Intel Iris Pro Graphics P6300 (5712, 0.3% behind), and the HD 8790M itself (5691, 0.7% behind). The HD 8790M's rivals mirror this: the Iris Pro P6300 (5712, 0.4% ahead), the GTX 670MX (5721, 0.5% ahead), and the GTX 550 Ti (5731, 0.7% ahead). Both cards sit within 1% of each other in average score, but the GTX 550 Ti's OpenCL win is the single largest delta in the head-to-head data. Pick the GTX 550 Ti for OpenCL compute; pick the HD 8790M for Vulkan and modern API support.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 550 Ti has an average benchmark score of 5731, while the AMD Radeon HD 8790M scores 5691, giving NVIDIA a 0.7% lead.

Q: Does the AMD Radeon HD 8790M support Vulkan?

A: Yes, the HD 8790M supports Vulkan 1.2.170 and scores 6365 in Geekbench Vulkan. The GTX 550 Ti has no Vulkan support listed.

Q: What is the difference in OpenCL performance between the two?

A: The GTX 550 Ti scores 5731 in Geekbench OpenCL versus the HD 8790M's 5017, a 14.2% advantage for NVIDIA.

Q: How much memory does each GPU have?

A: The GTX 550 Ti has 1024 MB of GDDR5 memory, while the HD 8790M has 2 GB of GDDR5 memory.

Q: Which GPU has higher memory bandwidth?

A: The GTX 550 Ti has 98.50 GB/s bandwidth on a 192-bit bus, compared to the HD 8790M's 64.00 GB/s on a 128-bit bus.

Q: Are both GPUs in the same performance percentile?

A: Yes, both the GTX 550 Ti and the HD 8790M sit at the 33rd percentile among all GPUs.

Head-to-Head Benchmarks

The only shared benchmark in the data is Geekbench OpenCL, and it is a decisive NVIDIA win. The GTX 550 Ti scores 5731 against the HD 8790M's 5017, a 14.2% delta. That margin comes from the NVIDIA card's architectural advantages: 32 TMUs versus 24, 24 ROPs versus 8, and 98.50 GB/s bandwidth versus 64.00 GB/s. The HD 8790M's 384 shading units are double the GTX 550 Ti's 192, yet the AMD card still loses by double digits, demonstrating that shading unit count alone does not determine OpenCL performance.

The GTX 550 Ti's nearest-rival data reinforces this win. Its closest competitor is the GTX 670MX at 5721 (0.2% behind), followed by the Intel Iris Pro P6300 at 5712 (0.3% behind), and then the HD 8790M at 5691 (0.7% behind). The HD 8790M's closest rival is the Quadro M500M at 5604 (1.6% behind), with the Iris Pro P6300 (0.4% ahead), GTX 670MX (0.5% ahead), and GTX 550 Ti (0.7% ahead) all clustered above it. The GTX 550 Ti's 5731 OpenCL score puts it at the top of this peer group, while the HD 8790M's 5017 OpenCL score is not even close to its own average of 5691, which is pulled up by the Vulkan result.

The HD 8790M's Vulkan score of 6365 is its standout number. It exceeds its OpenCL score by 1348 points (26.9% higher), and it represents the only benchmark where the AMD card posts a higher figure than the GTX 550 Ti's OpenCL score. Since the GTX 550 Ti cannot run Vulkan, this is not a head-to-head comparison but rather a capability gap. The GTX 550 Ti wins the only test both can run, but the HD 8790M has access to an entire API the NVIDIA card lacks, making the overall comparison closer than the 14.2% OpenCL delta suggests. In aggregate, the GTX 550 Ti leads by 0.7% in average score, but that lead rests entirely on the single OpenCL test.