AMD Radeon Pro 455 vs NVIDIA GeForce GTX 670 Comparison

The AMD Radeon Pro 455 and NVIDIA GeForce GTX 670 occupy the same percentile rank among all GPUs, at 52, but they achieve that position through entirely different performance profiles. The Radeon Pro 455 averages a benchmark score of 12831, while the GeForce GTX 670 sits just 0.5% behind at 12773. The data reveals a split decision: the AMD card dominates in one API, while the NVIDIA card wins the other two tested workloads.

Head-to-Head Benchmarks

The most dramatic result in the comparison is in the Geekbench Metal test, where the AMD Radeon Pro 455 scores 15916 against the GTX 670's 7424. That is a delta of 114.4%, meaning the AMD card delivers more than double the performance in this specific workload. This is the single largest margin between the two cards across all three benchmarks, and it is a decisive victory for the GCN 4.0 architecture. The scale of this win is substantial enough that it alone accounts for the Pro 455's overall average score being higher than its rival's.

The picture flips in the OpenCL benchmark. Here, the NVIDIA GeForce GTX 670 scores 15341, while the Radeon Pro 455 manages only 10336. The delta is -32.6% from the AMD card's perspective, indicating the GTX 670 is roughly a third faster in this test. Unlike the Metal result, which is a blowout, this is a clear but less extreme margin. The GTX 670's Kepler architecture, with its 1344 shading units, appears better suited to the OpenCL workload than the Pro 455's 768 shading units.

The Vulkan benchmark follows a similar pattern to OpenCL, though the margin narrows. The GTX 670 scores 15553, and the Radeon Pro 455 scores 12240, resulting in a -21.3% delta. While the NVIDIA card wins again, the gap is smaller than in OpenCL. Notably, the GTX 670's Vulkan score of 15553 is its highest of the three tests, while the Radeon Pro 455's Vulkan score sits between its Metal and OpenCL results. This suggests the NVIDIA card scales better across heterogeneous compute APIs, while the AMD card is heavily optimized for Apple's Metal framework.

Looking at the rival context from the FACT PACK, the Radeon Pro 455's nearest rival, the NVIDIA GeForce GTX 590, scores 12830, which is a 0% delta — essentially a tie. The GTX 670's nearest rival, the AMD Radeon RX 7600M XT, scores 12710, putting the GTX 670 0.5% ahead. Both cards are clustered tightly with their peers, meaning neither offers a significant overall performance advantage in aggregate, despite the wide swings in individual API tests.

The Verdict

The data supports a clear split based on workload environment. The AMD Radeon Pro 455 is the superior choice for any application that relies on Metal, given its 114.4% advantage in that benchmark. This is not a marginal lead; it is a doubling of performance. For users working within ecosystems where Metal is the primary graphics API, the Pro 455 is categorically the better performer.

The NVIDIA GeForce GTX 670 is the better option for Vulkan and OpenCL workloads. It leads by 21.3% in Vulkan and 32.6% in OpenCL. While these margins are smaller than the AMD card's Metal advantage, they are still significant. The GTX 670 also offers higher raw throughput in terms of pixel rate (27.44 GPixel/s vs 13.68 GPixel/s) and texture rate (109.8 GTexel/s vs 41.04 GTexel/s), which may translate to better performance in traditional rasterization tasks that do not rely on Metal.

For a general-purpose user who does not know which API their applications will use, the aggregate scores are nearly identical, with the Pro 455 holding a 0.5% edge in average benchmark score. However, the GTX 670's wins come in the more widely supported cross-platform APIs (OpenCL and Vulkan), while the Pro 455's win is in a proprietary API. Benchmark results indicate that the GTX 670 is the safer choice for cross-platform compatibility, while the Pro 455 is the specialist choice for Metal-centric environments.

FAQ

Q: Which card has the higher average benchmark score?

A: The AMD Radeon Pro 455 has an average benchmark score of 12831, while the NVIDIA GeForce GTX 670 scores 12773. This puts the Pro 455 0.5% ahead of the GTX 670, according to the nearestRivals data.

Q: How much faster is the AMD Radeon Pro 455 in Metal?

A: In the Geekbench Metal test, the Radeon Pro 455 scores 15916 versus the GTX 670's 7424. This represents a delta of 114.4%, meaning the AMD card is more than twice as fast in this specific benchmark.

Q: Does the NVIDIA GeForce GTX 670 win any benchmarks?

A: Yes, the GTX 670 wins two of the three head-to-head tests. It scores 15341 in Geekbench OpenCL (a 32.6% lead) and 15553 in Geekbench Vulkan (a 21.3% lead) over the Radeon Pro 455.

Q: What is the difference in transistor density between the two cards?

A: The AMD Radeon Pro 455 has a transistor density of 24.4M per mm², based on 3,000 million transistors on a 123 mm² die. The NVIDIA GeForce GTX 670 has a transistor density of 12.0M per mm², based on 3,540 million transistors on a 294 mm² die.

Q: Which card has a higher memory bandwidth?

A: The NVIDIA GeForce GTX 670 has a memory bandwidth of 192.3 GB/s, which is significantly higher than the AMD Radeon Pro 455's 81.28 GB/s. The GTX 670 also uses a 256-bit memory bus, while the Pro 455 uses a 128-bit bus.

Q: What is the launch MSRP of the NVIDIA GeForce GTX 670?

A: The launch MSRP of the NVIDIA GeForce GTX 670 is 399 USD. The AMD Radeon Pro 455 has no launch MSRP listed in the data.

Specification Differences

The two cards differ substantially in their core configurations. The NVIDIA GeForce GTX 670 has 1344 shading units, 112 texture mapping units, and 32 ROPs. The AMD Radeon Pro 455 has 768 shading units, 48 texture mapping units, and 16 ROPs. This means the GTX 670 has nearly double the shading units, more than double the TMUs, and exactly double the ROPs.

Clock speeds also differ, though the AMD card does not list a base or boost clock. The GTX 670 runs at a base clock of 915 MHz and a boost clock of 980 MHz. Memory clocks are closer: the GTX 670 runs its GDDR5 memory at 1502 MHz (6 Gbps effective), while the Pro 455 runs at 1270 MHz (5.1 Gbps effective).

The memory bus width is a major differentiator, with the GTX 670 using a 256-bit interface versus the Pro 455's 128-bit interface. This contributes to the bandwidth gap: 192.3 GB/s for NVIDIA versus 81.28 GB/s for AMD. Both cards have 2 GB of GDDR5 memory.

Power and physical specifications diverge sharply. The AMD Radeon Pro 455 has a TDP of 35 W and uses an MXM Module slot width with no power connectors. The NVIDIA GeForce GTX 670 has a TDP of 170 W, is a dual-slot card, requires two 6-pin power connectors, and has a suggested PSU of 450 W. The GTX 670's dimensions are 241 mm in length, 111 mm in height, and 38 mm in width.

The bus interface also differs: the Pro 455 uses PCIe 3.0 x8, while the GTX 670 uses PCIe 3.0 x16. Display outputs are "Portable Device Dependent" for the AMD card, while the GTX 670 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

Architecture Differences

The architectural split is generational and fundamental. The AMD Radeon Pro 455 is built on the Baffin chip using GCN 4.0 architecture, fabricated on a 14 nm process by GlobalFoundries. The NVIDIA GeForce GTX 670 uses the GK104 chip with Kepler architecture, fabricated on a 28 nm process by TSMC. The process node difference explains the transistor density gap: the Pro 455 packs 24.4M transistors per mm², while the GTX 670 manages only 12.0M per mm².

The AMD card's FP32 performance is listed as 1,313.3 GFLOPS, and it supports FP16 at a 1:1 ratio (1,313.3 GFLOPS). The NVIDIA card has FP32 performance of 2.634 TFLOPS, but no FP16 data is listed. This indicates the GTX 670 has higher single-precision throughput, while the Pro 455 offers half-precision capabilities that the Kepler card lacks.

API support differs in important ways. The Radeon Pro 455 supports DirectX 12 (12_0) and Vulkan 1.3, while the GTX 670 supports DirectX 12 (11_0) and Vulkan 1.2.175. Both support OpenGL 4.6. The higher Vulkan version on the AMD card may explain its relatively stronger Vulkan performance, though it still loses to the GTX 670 in that benchmark.

The release dates are separated by over four years: the GTX 670 launched on 2012-05-09, and the Radeon Pro 455 launched on 2016-10-29. The NVIDIA card's predecessor is the GeForce 500 series, and its successor is the GeForce 700 series. The AMD card has no listed predecessor or successor in the data. Both cards are marked as end-of-life production status.

Where Each One Wins

The AMD Radeon Pro 455 wins exclusively in the Metal benchmark, with a 114.4% advantage. This makes it the clear choice for systems and applications that prioritize Metal performance. The card's lower TDP of 35 W and MXM Module form factor also suggest it is designed for portable or compact systems, where power efficiency is paramount. Its higher transistor density (24.4M / mm²) and newer 14 nm process indicate a more modern manufacturing approach.

The NVIDIA GeForce GTX 670 wins in OpenCL and Vulkan, with leads of 32.6% and 21.3%, respectively. Its higher memory bandwidth (192.3 GB/s), larger memory bus (256-bit), and greater number of shading units (1344) give it a raw compute advantage that shows up in these cross-platform APIs. The card's higher TDP of 170 W and dual-slot design indicate it is intended for desktop systems where power and space are less constrained.

The overall wins tally is 1 for the AMD card and 2 for the NVIDIA card. However, the magnitude of the AMD card's single win (114.4%) is larger than either of the NVIDIA card's wins individually. The data suggests that for Metal-specific workloads, the Radeon Pro 455 is transformative, while for general compute, the GTX 670 is more consistent. Users should select based on the dominant API in their workflow, as the aggregate performance difference of 0.5% is negligible.