AMD Radeon Pro 575 vs NVIDIA GeForce GTX 1060 5 GB Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon Pro 575 records an average benchmark score of 39,555, which places it in the 82nd percentile among all GPUs. The NVIDIA GeForce GTX 1060 5 GB averages 33,694, sitting in the 78th percentile.

Q: How do the two cards compare in OpenCL performance?

A: In the Geekbench OpenCL test, the AMD Radeon Pro 575 scores 34,596, while the NVIDIA GeForce GTX 1060 5 GB scores 33,397. The AMD card wins by 3.6%.

Q: What about Vulkan performance?

A: In the Geekbench Vulkan test, the AMD Radeon Pro 575 scores 37,878 versus 33,991 for the NVIDIA card. This is a wider margin, with AMD leading by 11.4%.

Q: Which card has more memory bandwidth?

A: The AMD Radeon Pro 575 features a 256-bit memory bus with a bandwidth of 217.0 GB/s, compared to the NVIDIA GeForce GTX 1060 5 GB's 160-bit bus and 160.2 GB/s. The AMD card also has a smaller memory capacity of 4 GB versus 5 GB on the NVIDIA.

Q: Are there any benchmark results where the NVIDIA card wins?

A: No. In the recorded head-to-head data, the AMD Radeon Pro 575 wins both available tests (OpenCL and Vulkan). The NVIDIA GeForce GTX 1060 5 GB has zero wins in the comparison.

Q: How does the AMD card's closest rival compare to it?

A: The AMD Radeon Pro 575's nearest rival by average score is the NVIDIA RTX A500 Mobile, which scores 39,568, a delta of 0%. The AMD card is also 1.1% ahead of the AMD Radeon Pro 575X and 1.9% ahead of the AMD Radeon Pro 580.

The Verdict

The data is unambiguous: the AMD Radeon Pro 575 is the stronger performer in every recorded benchmark. It wins both head-to-head tests, with its largest advantage coming in Vulkan where it leads by 11.4%. The average benchmark scores reinforce this, showing the AMD card ahead by roughly 17% overall (39,555 versus 33,694).

For users prioritizing raw compute performance in OpenCL and Vulkan workloads, the AMD Radeon Pro 575 is the clear choice. Its 82nd percentile ranking versus the NVIDIA card's 78th percentile further confirms its position in the broader GPU landscape.

However, the NVIDIA GeForce GTX 1060 5 GB is not without merit. It offers a larger 5 GB memory pool, which could be relevant for workloads that require more capacity than the AMD card's 4 GB. It also carries a lower TDP of 120 W versus 150 W, and it has a higher pixel rate of 68.36 GPixel/s compared to 35.07 GPixel/s on the AMD.

The verdict depends on workload type. For general compute and Vulkan-based applications, pick the AMD Radeon Pro 575. For scenarios where memory capacity or raw pixel throughput is more critical, the NVIDIA card's 5 GB frame buffer and higher pixel rate may serve better, despite losing the benchmark race.

Head-to-Head Benchmarks

The recorded head-to-head data covers two tests, and the AMD Radeon Pro 575 wins both. In Geekbench OpenCL, the AMD card scores 34,596 against 33,397 for the NVIDIA GeForce GTX 1060 5 GB, a 3.6% advantage. This is a modest but consistent lead.

The gap widens considerably in Geekbench Vulkan. The AMD Radeon Pro 575 posts 37,878, while the NVIDIA card manages 33,991. That is an 11.4% difference in favor of AMD. Notably, the AMD card's Vulkan score is also higher than its OpenCL score by 9.5%, whereas the NVIDIA card's Vulkan score is only 1.8% higher than its OpenCL result. This suggests the AMD architecture handles Vulkan workloads particularly well.

Looking at the broader benchmark context, the AMD card's average score of 39,555 is driven by its strong Metal performance of 46,192, a test the NVIDIA card does not have recorded. This Metal result is 19.5% above the AMD card's own average, indicating that Apple's Metal API is a particularly strong environment for this GPU.

The NVIDIA GeForce GTX 1060 5 GB, by contrast, has no recorded Metal benchmark. Its two available scores (OpenCL and Vulkan) are relatively close, with a spread of only 594 points. This consistency might appeal to users who expect similar performance across different APIs, but it does not overcome the AMD card's outright wins.

Specification Differences

The two cards differ across nearly every major specification. The AMD Radeon Pro 575 uses a 14 nm process from GlobalFoundries, while the NVIDIA GeForce GTX 1060 5 GB uses a 16 nm process from TSMC. The AMD chip, Ellesmere, packs 5,700 million transistors on a 232 mm² die, giving a transistor density of 24.6M per mm². The NVIDIA chip, GP106, has 4,400 million transistors on a 200 mm² die, for a density of 22.0M per mm².

Memory configurations diverge significantly. The AMD card has 4 GB of GDDR5 on a 256-bit bus, delivering 217.0 GB/s of bandwidth. The NVIDIA card has 5 GB of GDDR5 on a 160-bit bus, yielding 160.2 GB/s. The AMD card's memory runs at 1695 MHz (6.8 Gbps effective), while the NVIDIA card's memory runs at 2002 MHz (8 Gbps effective).

Compute resources also differ. The AMD Radeon Pro 575 has 2,048 shading units, 128 TMUs, and 32 ROPs. The NVIDIA GeForce GTX 1060 5 GB has 1,280 shading units, 80 TMUs, and 40 ROPs. Despite having fewer shading units, the NVIDIA card achieves a higher pixel rate (68.36 GPixel/s versus 35.07 GPixel/s) due to its higher clock speeds. The texture rates are nearly identical: 140.3 GTexel/s for AMD and 136.7 GTexel/s for NVIDIA.

Clock speeds are only fully specified for the NVIDIA card, with a base of 1506 MHz and boost of 1709 MHz. The AMD card's base and boost clocks are not recorded in the database. The AMD card's FP32 throughput is 4.489 TFLOPS, marginally ahead of the NVIDIA card's 4.375 TFLOPS. The FP16 comparison is stark: the AMD card delivers 4.489 TFLOPS (1:1 ratio), while the NVIDIA card manages only 68.36 GFLOPS (1:64 ratio).

Power and physical characteristics differ as well. The AMD card has a TDP of 150 W and uses an MXM Module slot, with no power connectors. The NVIDIA card has a TDP of 120 W, uses a dual-slot design, requires a single 6-pin power connector, and suggests a 300 W power supply. The NVIDIA card is 250 mm (9.8 inches) long, while the AMD card's dimensions are not recorded. Display outputs also differ, with the NVIDIA card offering 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a, while the AMD card's outputs are listed as portable device dependent.

Architecture Differences

The AMD Radeon Pro 575 is built on GCN 4.0 architecture, using the Ellesmere chip. It belongs to the Radeon Pro Mac (500 Series) generation. The NVIDIA GeForce GTX 1060 5 GB uses the Pascal architecture with the GP106 chip, part of the GeForce 10-series.

The manufacturing process differs, with AMD using 14 nm at GlobalFoundries and NVIDIA using 16 nm at TSMC. The AMD card has a higher transistor count (5,700 million versus 4,400 million) and a larger die (232 mm² versus 200 mm²), resulting in a higher transistor density (24.6M per mm² versus 22.0M per mm²).

Compute architecture varies substantially. The AMD card has 2,048 shading units and 128 TMUs, while the NVIDIA card has 1,280 shading units and 80 TMUs. However, the NVIDIA card has more ROPs (40 versus 32). The FP16 capability highlights a major architectural difference: the AMD card supports FP16 at a 1:1 ratio with FP32 (4.489 TFLOPS), while the NVIDIA card's FP16 is severely limited at a 1:64 ratio (68.36 GFLOPS).

API support shows differences in DirectX and Vulkan versions. The AMD card supports DirectX 12 (12_0) and Vulkan 1.3, while the NVIDIA card supports DirectX 12 (12_1) and Vulkan 1.4. Both support OpenGL 4.6. Neither card has ray tracing cores or tensor cores.

The memory architecture differs in bus width, with the AMD card using a wider 256-bit bus versus the NVIDIA card's 160-bit bus. Despite the NVIDIA card having faster memory clocks (8 Gbps effective versus 6.8 Gbps), the AMD card's wider bus gives it significantly higher bandwidth (217.0 GB/s versus 160.2 GB/s).

Release timing also differs, with the AMD card released on June 4, 2017, and the NVIDIA card released on December 25, 2017. The NVIDIA card has a recorded predecessor (GeForce 900) and successor (GeForce 20), while the AMD card has neither recorded.

Where Each One Wins

The AMD Radeon Pro 575 wins on raw compute performance across all recorded benchmarks. Its OpenCL score of 34,596 beats the NVIDIA card's 33,397 by 3.6%. Its Vulkan score of 37,878 beats the NVIDIA card's 33,991 by 11.4%. The AMD card also has a recorded Metal score of 46,192, which is its strongest benchmark and a test the NVIDIA card does not have.

The AMD card's FP32 and FP16 throughput are both higher, with 4.489 TFLOPS in both cases versus the NVIDIA card's 4.375 TFLOPS FP32 and 68.36 GFLOPS FP16. The AMD card also provides higher memory bandwidth (217.0 GB/s versus 160.2 GB/s) and a wider memory bus (256-bit versus 160-bit). Its 82nd percentile ranking versus 78th for the NVIDIA card further underscores its stronger overall standing.

The NVIDIA GeForce GTX 1060 5 GB wins on memory capacity, offering 5 GB versus 4 GB on the AMD card. This could matter for workloads that are capacity-sensitive rather than bandwidth-sensitive. The NVIDIA card also has a higher pixel rate (68.36 GPixel/s versus 35.07 GPixel/s), which suggests it may handle fill-rate-bound tasks more efficiently. Its texture rate is marginally lower (136.7 GTexel/s versus 140.3 GTexel/s), but the difference is small.

The NVIDIA card has a lower TDP of 120 W versus 150 W, and it uses a standard dual-slot form factor with a 6-pin power connector, which may be easier to integrate in a desktop system. The AMD card's MXM Module form factor and lack of power connectors suggest it is designed for portable or proprietary systems. The NVIDIA card also offers more display outputs (DVI, HDMI 2.0, and three DisplayPort 1.4a) compared to the AMD card's portable-device-dependent outputs.

In direct comparison, the AMD Radeon Pro 575 is the benchmark winner in every test where both are measured. The NVIDIA GeForce GTX 1060 5 GB offers advantages in memory capacity, pixel rate, power consumption, and connectivity, but it cannot match the AMD card's compute performance. Users who prioritize benchmark scores should choose the AMD card; those who need the extra 1 GB of memory or lower power draw might prefer the NVIDIA card despite its benchmark deficit.