AMD Radeon Pro 560X vs NVIDIA T400 4 GB Comparison

The Verdict

The recorded benchmark data splits decisively by workload. The NVIDIA T400 4 GB wins the OpenCL compute test by a massive margin, scoring 17,320 versus 9,663 for the AMD Radeon Pro 560X, a 79.2% advantage. Conversely, the AMD Radeon Pro 560X takes the Vulkan test, scoring 16,819 against 16,263, a 3.3% lead. Each GPU claims one head-to-head win, but the scale of the victories is not equal. The NVIDIA card's OpenCL dominance is overwhelming, while the AMD card's Vulkan edge is narrow. Users prioritizing compute workloads in OpenCL should choose the NVIDIA T400 4 GB without hesitation. Users whose primary API is Vulkan, particularly in macOS environments where this AMD part is designed to operate, will find the Radeon Pro 560X marginally better, but the practical difference in Vulkan is small. The NVIDIA card also holds a higher overall percentile rank, sitting at 60 versus 57 for the AMD part. Its average benchmark score of 16,792 far exceeds the AMD card's 15,082. For raw average performance across the recorded tests, the NVIDIA T400 4 GB is the stronger product.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA T400 4 GB, with an average benchmark score of 16,792, compared to 15,082 for the AMD Radeon Pro 560X.

Q: How large is the OpenCL performance gap between the two cards?

A: The NVIDIA T400 4 GB scores 17,320 in Geekbench OpenCL, while the AMD Radeon Pro 560X scores 9,663. This represents a 79.2% advantage for the NVIDIA card.

Q: Does the AMD Radeon Pro 560X win any benchmark?

A: Yes, it wins the Geekbench Vulkan test, scoring 16,819 versus 16,263 for the NVIDIA T400 4 GB, a 3.3% lead.

Q: What is the memory bandwidth difference?

A: The AMD Radeon Pro 560X has a memory bandwidth of 94.08 GB/s, while the NVIDIA T400 4 GB has 80.00 GB/s. The AMD card has a 128-bit bus, and the NVIDIA card has a 64-bit bus.

Q: How do the shading unit counts compare?

A: The AMD Radeon Pro 560X has 1,024 shading units, while the NVIDIA T400 4 GB has 384 shading units.

Q: Which card has a higher FP32 performance figure?

A: The AMD Radeon Pro 560X, with 2.056 TFLOPS, compared to 1,094.4 GFLOPS for the NVIDIA T400 4 GB.

Architecture Differences

The NVIDIA T400 4 GB is built on the Turing architecture, specifically using the TU117 chip. This is a 12 nm design fabricated by TSMC, containing 4,700 million transistors on a 200 mm² die. The transistor density is 23.5M per mm². The AMD Radeon Pro 560X uses the GCN 4.0 architecture with the Polaris 21 chip. It is built on a 14 nm process at GlobalFoundries, with 3,000 million transistors on a 123 mm² die, yielding a slightly higher density of 24.4M per mm². The NVIDIA part is a discrete, single-slot card designed for workstation use, while the AMD part is an integrated GPU (IGP) intended for Mac portable systems. The NVIDIA card supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The AMD card supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. Notably, the AMD part has a 1:1 FP16 to FP32 ratio, both at 2.056 TFLOPS, while the NVIDIA part has a 2:1 ratio, with FP16 at 2.189 TFLOPS and FP32 at 1,094.4 GFLOPS. The NVIDIA card has no dedicated RT or tensor cores, and neither does the AMD card.

Specification Differences

The two cards differ across nearly every specification field. The NVIDIA T400 4 GB has a base clock of 420 MHz and a boost clock of 1425 MHz, while the AMD Radeon Pro 560X lists no base or boost clock in the database. Memory clocks also differ: the NVIDIA card runs at 1250 MHz, or 10 Gbps effective, while the AMD card runs at 1470 MHz, or 5.9 Gbps effective. Memory type is GDDR6 for NVIDIA and GDDR5 for AMD. Both have 4 GB of memory, but the bus widths differ: 64 bit for NVIDIA versus 128 bit for AMD. Bandwidth is 80.00 GB/s for NVIDIA and 94.08 GB/s for AMD. The shading unit count is 384 for NVIDIA and 1,024 for AMD. Texture mapping units are 24 versus 64. Raster output units are 16 on both. Pixel rate is 22.80 GPixel/s for NVIDIA and 16.06 GPixel/s for AMD. Texture rate is 34.20 GTexel/s for NVIDIA and 64.26 GTexel/s for AMD. The NVIDIA card has a TDP of 30 W, while the AMD card has a TDP of 75 W. The NVIDIA card is single-slot with no power connectors and a suggested PSU of 200 W. The AMD card is an IGP with no power connectors and no suggested PSU. Bus interface is PCIe 3.0 x16 for NVIDIA and PCIe 3.0 x8 for AMD. Display outputs are 3x mini-DisplayPort 1.4a for NVIDIA and "Portable Device Dependent" for AMD. The NVIDIA card was released on 2021-05-05, while the AMD card was released on 2018-07-15. Both are end-of-life. The NVIDIA card has a predecessor in Quadro Volta and a successor in Workstation Ampere. The AMD card has no recorded predecessor or successor.

Head-to-Head Benchmarks

The Geekbench OpenCL test is the defining metric in this comparison. The NVIDIA T400 4 GB delivers 17,320 points, against 9,663 for the AMD Radeon Pro 560X. This is a 79.2% delta, a commanding lead that dwarfs any other difference between the two cards. In raw compute terms, the NVIDIA card is nearly twice as fast in this API. The AMD card's higher FP32 rating of 2.056 TFLOPS does not translate into OpenCL performance, as the NVIDIA card's 1,094.4 GFLOPS still wins decisively. This suggests the NVIDIA architecture is far more efficient in OpenCL execution. The Vulkan test tells a different story. The AMD Radeon Pro 560X scores 16,819, edging out the NVIDIA T400 4 GB's 16,263. The delta is 3.3%, a modest margin that falls within a range where driver optimization or test variance could matter. Still, the data records a win for the AMD part. The NVIDIA card's nearest rivals in the database include the AMD Radeon RX 7600S at 16,696, the NVIDIA Tesla M4 at 16,932, the AMD Radeon HD 7970M at 17,019, and the NVIDIA GeForce GTX 690 at 17,037. The NVIDIA T400 4 GB sits just above the RX 7600S and below the Tesla M4. The AMD Radeon Pro 560X's nearest rivals include the NVIDIA GeForce GTX 660 Ti at 15,063, the AMD Radeon RX 7600 at 15,171, the NVIDIA GeForce RTX 3050 OEM at 15,199, and the AMD Radeon 680M at 15,270. The AMD card is within 0.1% of the GTX 660 Ti and within 1.2% of the Radeon 680M. These rival comparisons place the NVIDIA card in a higher performance tier overall.

Where Each One Wins

The NVIDIA T400 4 GB wins in OpenCL compute. With a 79.2% lead over the AMD Radeon Pro 560X, this is the clear choice for any workload that leverages OpenCL, such as general-purpose GPU computing, certain scientific applications, or rendering pipelines that use this API. The NVIDIA card also wins on average benchmark score, 16,792 versus 15,082, and holds a higher percentile rank at 60 versus 57. Its lower TDP of 30 W versus 75 W makes it more power-efficient, and its PCIe 3.0 x16 interface offers double the bandwidth of the AMD card's x8 link. The NVIDIA card also has a higher pixel rate, 22.80 GPixel/s versus 16.06 GPixel/s, which can benefit 2D and display output tasks. Its display outputs are fixed as 3x mini-DisplayPort 1.4a, making it suitable for multi-monitor workstation setups.

The AMD Radeon Pro 560X wins in Vulkan. Its 3.3% lead in that test, while modest, is a recorded victory. The AMD card also has a higher memory bandwidth at 94.08 GB/s versus 80.00 GB/s, thanks to its 128-bit bus versus 64-bit. It has far more shading units, 1,024 versus 384, and more texture mapping units, 64 versus 24. Its texture rate of 64.26 GTexel/s is nearly double the NVIDIA card's 34.20 GTexel/s. The AMD card also has a higher FP32 throughput at 2.056 TFLOPS. For workloads that scale with shader count or texture fill, such as certain Vulkan-based games or compute tasks that use Vulkan, the AMD card has the theoretical advantage. Its status as an IGP means it is designed for portable Mac systems, where the NVIDIA card's discrete slot-based design would not apply. Users with a Mac portable system should favor the AMD card, as it is the only one of the two that fits that form factor.