AMD Radeon Pro 455 vs NVIDIA GeForce GTX 660 Ti Comparison

The Verdict

The benchmark data clearly favors the NVIDIA GeForce GTX 660 Ti in direct comparison. It wins both recorded head-to-head tests, with a 46.2% advantage in Geekbench OpenCL and a 22.6% advantage in Geekbench Vulkan. The average benchmark score reinforces this: the GTX 660 Ti averages 15,063 across its two tests against the Radeon Pro 455's 12,831 across three. The GTX 660 Ti also sits at the 57th percentile of all GPUs in the database, while the Radeon Pro 455 lands at the 52nd.

The GTX 660 Ti is the choice for raw compute performance in OpenCL and Vulkan workloads. Its nearest rivals in the database include the AMD Radeon RX 7600 at 0.7% higher average score and the NVIDIA GeForce RTX 3050 OEM at 0.9% higher, placing it in respectable company. The Radeon Pro 455, by contrast, sits just 0.1% above the NVIDIA GeForce GTX 590 and 0.5% above the GeForce GTX 670 in average score, a much older performance tier.

However, the Radeon Pro 455 is not without its own argument. Its single Geekbench Metal score of 15,916 exceeds any single score recorded for the GTX 660 Ti. The GTX 660 Ti has no Metal benchmark entry in the database, meaning the AMD part is the only option with proven Metal performance. For users locked into Apple's Metal ecosystem, the Radeon Pro 455's 15,916 Metal result is the decisive data point, despite losing the OpenCL and Vulkan comparisons.

The verdict splits cleanly: the GTX 660 Ti for OpenCL and Vulkan compute, the Radeon Pro 455 for Metal-bound environments. The GTX 660 Ti's 46.2% OpenCL lead over the Radeon Pro 455 is the largest margin in either direction, and its Vulkan lead of 22.6% is substantial. The Radeon Pro 455's only recorded win category is Metal, where the GTX 660 Ti has no recorded score at all.

Architecture Differences

The two GPUs come from different foundries, process nodes, and architectural generations. The NVIDIA GeForce GTX 660 Ti uses the GK104 chip on TSMC's 28 nm process, built on the Kepler architecture. It packs 3,540 million transistors into a 294 mm² die, yielding a transistor density of 12.0 million transistors per mm². The AMD Radeon Pro 455 uses the Baffin chip on GlobalFoundries' 14 nm process, built on GCN 4.0. It contains 3,000 million transistors on a much smaller 123 mm² die, achieving a transistor density of 24.4 million per mm², more than double the GTX 660 Ti's density.

The compute resources differ sharply. The GTX 660 Ti has 1,344 shading units, 112 texture mapping units, and 24 ROPs. The Radeon Pro 455 has 768 shading units, 48 TMUs, and 16 ROPs. These counts translate into a large throughput gap: the GTX 660 Ti delivers 2.634 TFLOPS of FP32 compute against the Radeon Pro 455's 1,313.3 GFLOPS (roughly 1.31 TFLOPS). The GTX 660 Ti is roughly twice as fast in raw FP32 throughput. The Radeon Pro 455 does match its FP32 rate with FP16 at 1,313.3 GFLOPS, a 1:1 ratio, while the GTX 660 Ti has no recorded FP16 figure.

Pixel and texture rates follow the same pattern. The GTX 660 Ti reaches 27.44 GPixel/s and 109.8 GTexel/s, while the Radeon Pro 455 reaches 13.68 GPixel/s and 41.04 GTexel/s. The GTX 660 Ti is roughly double the pixel rate and roughly 2.7 times the texture rate.

Memory configurations also diverge. Both have 2 GB of GDDR5, but the GTX 660 Ti uses a 192-bit bus with 144.2 GB/s of bandwidth, while the Radeon Pro 455 uses a 128-bit bus with 81.28 GB/s. Clock rates show the same story: the GTX 660 Ti runs its memory at 1502 MHz (6 Gbps effective), while the Radeon Pro 455 runs at 1270 MHz (5.1 Gbps effective). The GTX 660 Ti also has base and boost clocks of 915 MHz and 980 MHz respectively, while the Radeon Pro 455 has no recorded base or boost clock.

Power and form factor differences are stark. The GTX 660 Ti has a 150 W TDP, is dual-slot, requires two 6-pin power connectors, and needs a 450 W suggested PSU. The Radeon Pro 455 has a 35 W TDP, uses an MXM module form factor, requires no power connectors, and has no suggested PSU listed. The bus interface also differs: PCIe 3.0 x16 for the GTX 660 Ti, PCIe 3.0 x8 for the Radeon Pro 455.

API support favors the Radeon Pro 455 in two areas. It supports DirectX 12 (12_0) and Vulkan 1.3, while the GTX 660 Ti supports DirectX 12 (11_0) and Vulkan 1.2.175. Both support OpenGL 4.6. The GTX 660 Ti has no recorded RT cores or tensor cores, and neither does the Radeon Pro 455.

Head-to-Head Benchmarks

The Geekbench OpenCL test is the clearest separation point. The GTX 660 Ti scores 15,113 against the Radeon Pro 455's 10,336, a 46.2% lead. This is the largest delta between the two in any recorded test. The GTX 660 Ti's OpenCL score is within 0.1% of the AMD Radeon Pro 560X (15,082) and within 0.7% of the AMD Radeon RX 7600 (15,171), showing it performs at a much more modern level in this workload. The Radeon Pro 455's 10,336 OpenCL score is far below any of its nearest rivals, which cluster around 12,830.

The Geekbench Vulkan test also goes to the GTX 660 Ti, with a score of 15,012 against 12,240 for the Radeon Pro 455, a 22.6% margin. The GTX 660 Ti's Vulkan score trails its OpenCL score by only 101 points, indicating consistent performance across both APIs. The Radeon Pro 455's Vulkan score of 12,240 is 1,904 points higher than its OpenCL score, showing a smaller but still significant gap.

The Radeon Pro 455's only competitive result is its Geekbench Metal score of 15,916. This single result exceeds the GTX 660 Ti's best recorded score (15,113 OpenCL) by 803 points. However, the GTX 660 Ti has no Metal score in the database, so there is no direct comparison. The Radeon Pro 455's Metal score also exceeds its own Vulkan score by 3,676 points and its OpenCL score by 5,580 points, indicating Metal is clearly its strongest API.

Average scores reinforce the GTX 660 Ti's overall lead. The GTX 660 Ti averages 15,063 across two benchmark entries, landing at the 57th percentile. The Radeon Pro 455 averages 12,831 across three entries, landing at the 52nd percentile. The Radeon Pro 455's average is dragged down by its 10,336 OpenCL score, which is its lowest recorded result.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 660 Ti averages 15,063 across its two recorded tests, while the AMD Radeon Pro 455 averages 12,831 across three recorded tests. The GTX 660 Ti also sits at the 57th percentile of all GPUs, versus the 52nd percentile for the Radeon Pro 455.

Q: Does the Radeon Pro 455 win any benchmark category?

A: Yes. The Radeon Pro 455 records a Geekbench Metal score of 15,916, which is its highest result and exceeds the GTX 660 Ti's best recorded score of 15,113 in OpenCL. The GTX 660 Ti has no Metal score in the database.

Q: What is the largest performance gap between the two?

A: The largest gap is in Geekbench OpenCL, where the GTX 660 Ti scores 15,113 against the Radeon Pro 455's 10,336, a 46.2% lead. The Vulkan gap is smaller at 22.6%, with scores of 15,012 and 12,240 respectively.

Q: How do their power requirements compare?

A: The GTX 660 Ti has a 150 W TDP, is dual-slot, requires two 6-pin power connectors, and lists a 450 W suggested PSU. The Radeon Pro 455 has a 35 W TDP, uses an MXM module form factor, and requires no power connectors.

Q: Which GPU supports a newer DirectX version?

A: The Radeon Pro 455 supports DirectX 12 (12_0), while the GTX 660 Ti supports DirectX 12 (11_0). The Radeon Pro 455 also supports Vulkan 1.3, while the GTX 660 Ti supports Vulkan 1.2.175.

Q: What are their memory bandwidth figures?

A: The GTX 660 Ti has 144.2 GB/s of bandwidth from a 192-bit bus with GDDR5 at 1502 MHz (6 Gbps effective). The Radeon Pro 455 has 81.28 GB/s from a 128-bit bus with GDDR5 at 1270 MHz (5.1 Gbps effective).

Where Each One Wins

The GTX 660 Ti wins in OpenCL and Vulkan compute. Its 46.2% OpenCL lead and 22.6% Vulkan lead over the Radeon Pro 455 make it the stronger choice for any workload that relies on those APIs. Its FP32 throughput of 2.634 TFLOPS is double the Radeon Pro 455's 1,313.3 GFLOPS, and its texture rate of 109.8 GTexel/s is roughly 2.7 times the Radeon Pro 455's 41.04 GTexel/s. Pixel rate also favors the GTX 660 Ti at 27.44 GPixel/s versus 13.68 GPixel/s.

The GTX 660 Ti also wins on memory bandwidth. Its 144.2 GB/s is 77% higher than the Radeon Pro 455's 81.28 GB/s, and its 192-bit bus provides 50% more width than the 128-bit bus. The higher memory clock of 1502 MHz versus 1270 MHz further widens the gap.

The Radeon Pro 455 wins in the Metal API category, where its 15,916 score stands as the only Metal result between the two. It also wins on efficiency and integration: a 35 W TDP versus 150 W, no power connectors versus two 6-pin, and an MXM module form factor versus a dual-slot card. Its 14 nm process and higher transistor density of 24.4 million per mm² versus 12.0 million per mm² indicate a more modern manufacturing approach.

The Radeon Pro 455 also wins on API version support. Its DirectX 12 (12_0) support is a full feature level above the GTX 660 Ti's DirectX 12 (11_0), and its Vulkan 1.3 support is newer than the GTX 660 Ti's Vulkan 1.2.175. For users requiring the latest API feature sets, the Radeon Pro 455 is the only option.

Specification Differences

The two GPUs differ across nearly every specification category. The GTX 660 Ti uses the GK104 chip on a 28 nm TSMC process, while the Radeon Pro 455 uses the Baffin chip on a 14 nm GlobalFoundries process. Transistor counts are 3,540 million versus 3,000 million, with die sizes of 294 mm² versus 123 mm². Transistor density is 12.0 million per mm² versus 24.4 million per mm².

Clock speeds differ in availability: the GTX 660 Ti has a base clock of 915 MHz and boost of 980 MHz, while the Radeon Pro 455 has no recorded base or boost clock. Memory clocks are 1502 MHz (6 Gbps effective) versus 1270 MHz (5.1 Gbps effective). Both use 2 GB of GDDR5, but bus widths are 192-bit versus 128-bit, and bandwidth is 144.2 GB/s versus 81.28 GB/s.

Compute unit counts favor the GTX 660 Ti: 1,344 shading units versus 768, 112 TMUs versus 48, and 24 ROPs versus 16. Pixel rates are 27.44 GPixel/s versus 13.68 GPixel/s, and texture rates are 109.8 GTexel/s versus 41.04 GTexel/s. FP32 throughput is 2.634 TFLOPS versus 1,313.3 GFLOPS. The Radeon Pro 455 has an FP16 rate of 1,313.3 GFLOPS (1:1), while the GTX 660 Ti has no recorded FP16 figure.

Power and physical specifications diverge completely. The GTX 660 Ti has a 150 W TDP, dual-slot width, two 6-pin power connectors, a 450 W suggested PSU, and a length of 241 mm (9.5 inches). The Radeon Pro 455 has a 35 W TDP, MXM module form factor, no power connectors, no suggested PSU, and no recorded dimensions. The bus interface is PCIe 3.0 x16 for the GTX 660 Ti and PCIe 3.0 x8 for the Radeon Pro 455.

Display outputs also differ: the GTX 660 Ti has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the Radeon Pro 455 is listed as "Portable Device Dependent." API support shows the Radeon Pro 455 with DirectX 12 (12_0) and Vulkan 1.3, versus DirectX 12 (11_0) and Vulkan 1.2.175 for the GTX 660 Ti; both support OpenGL 4.6. The GTX 660 Ti has a launch MSRP of 299 USD. The Radeon Pro 455 has no launch MSRP recorded. Release dates are 2012-08-15 for the GTX 660 Ti and 2016-10-29 for the Radeon Pro 455, and the GTX 660 Ti has a predecessor in the GeForce 500 series and a successor in the GeForce 700 series, while the Radeon Pro 455 lists neither.