AMD Radeon PRO V620 vs NVIDIA Tesla P100 PCIe 16 GB Comparison

# AMD Radeon PRO V620 vs NVIDIA Tesla P100 PCIe 16 GB

The AMD Radeon PRO V620 and NVIDIA Tesla P100 PCIe 16 GB represent two different eras of professional GPU design. The database shows a single head-to-head benchmark, Geekbench OpenCL, where the AMD Radeon PRO V620 scores 128,580 against the Tesla P100's 79,605, a 61.5% advantage for the AMD part. Beyond that headline result, the two cards diverge sharply in architecture, memory configuration, and API support, making them suited to different workloads despite both being datacenter-oriented accelerator boards without display outputs.

Where Each One Wins

The AMD Radeon PRO V620 wins the only directly comparable benchmark in the database. Its Geekbench OpenCL score of 128,580 places it 61.5% ahead of the Tesla P100's 79,605. The AMD card also records a Geekbench Vulkan score of 144,364, a test the Tesla P100 does not appear in at all. This Vulkan result indicates the V620 has a functional advantage in compute workloads using that API, which is consistent with its newer architecture and broader API feature set.

The NVIDIA Tesla P100's position is more nuanced. While it loses the OpenCL comparison decisively, it still holds a respectable 92nd percentile ranking among all GPUs in the database, compared to the V620's 96th percentile. The P100's average benchmark score of 79,605 places it within 0.3% of the Tesla P100 PCIe 12 GB (79,396), within 0.8% of the AMD Radeon RX 6850M XT (78,940), and only 1.7% behind the AMD Radeon Pro Vega 64X (80,959). The V620, by contrast, sits within 0.5% of the AMD Radeon Pro W6800X Duo (135,774), 0.8% of the AMD Radeon PRO W6800 (135,396), and 0.9% of both the NVIDIA A10M (135,230) and NVIDIA RTX 4000 Ada Generation (135,218).

For workloads where raw OpenCL throughput is the primary concern, the V620 is the clear winner. For environments that rely on older Pascal-era compute stacks or require HBM2 memory characteristics, the Tesla P100 retains relevance, but the benchmark data shows it is outclassed by the newer AMD part in raw compute scores.

Architecture Differences

The two GPUs come from different architectural generations and manufacturing processes. The AMD Radeon PRO V620 uses the Navi 21 chip built on RDNA 2.0 architecture, fabricated on TSMC's 7 nm process. It contains 26,800 million transistors on a 520 mm² die, yielding a transistor density of 51.5 million per mm². The NVIDIA Tesla P100 uses the GP100 chip based on Pascal architecture, fabricated on TSMC's 16 nm process. It contains 15,300 million transistors on a larger 610 mm² die, with a lower transistor density of 25.1 million per mm².

Clock speeds differ substantially. The V620 runs at a base clock of 1825 MHz and boosts to 2200 MHz. The P100 operates at a much lower 1190 MHz base and 1329 MHz boost. This clock advantage contributes significantly to the V620's higher compute throughput.

Memory configurations are fundamentally different. The V620 uses 32 GB of GDDR6 memory on a 256-bit bus, delivering 512.0 GB/s of bandwidth. The P100 uses 16 GB of HBM2 memory on a 4096-bit bus, delivering 732.2 GB/s of bandwidth. Despite having half the memory capacity, the P100 has 43% more memory bandwidth, which can benefit memory-bound workloads. However, the V620 has twice the capacity, which is critical for large datasets that must fit in GPU memory.

Compute resources also differ. The V620 has 4608 shading units, 288 texture mapping units, and 128 ROPs, plus 72 ray tracing cores. The P100 has 3584 shading units, 224 TMUs, and 96 ROPs, with no ray tracing cores. The V620's pixel rate is 281.6 GPixel/s versus 127.6 GPixel/s for the P100, and its texture rate is 633.6 GTexel/s versus 297.7 GTexel/s. Floating-point performance follows the same pattern: the V620 delivers 20.28 TFLOPS FP32 and 40.55 TFLOPS FP16 (2:1), while the P100 delivers 9.526 TFLOPS FP32 and 19.05 TFLOPS FP16 (2:1).

The V620 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The P100 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. This means the V620 has a more recent DirectX feature level and a newer Vulkan version, which matters for compatibility with modern compute frameworks.

Power and physical specifications also differ. The V620 has a TDP of 300 W, requires two 8-pin power connectors, and a suggested 700 W PSU. The P100 has a TDP of 250 W, requires one 8-pin connector, and a suggested 600 W PSU. Both are dual-slot cards with no display outputs, and both measure 267 mm (10.5 inches) in length. The V620 is 120 mm (4.7 inches) tall and 50 mm (2 inches) wide; the P100's height and width are not recorded.

The Verdict

The benchmark data supports a clear performance hierarchy. The AMD Radeon PRO V620 outperforms the NVIDIA Tesla P100 PCIe 16 GB by 61.5% in Geekbench OpenCL, the only direct comparison available. It also offers double the memory capacity, a newer architecture, higher clock speeds, more compute units, and support for newer API versions. For any workload that relies on OpenCL compute performance, the V620 is the stronger choice.

The Tesla P100 retains advantages in memory bandwidth (732.2 GB/s versus 512.0 GB/s) and lower power consumption (250 W versus 300 W). Its HBM2 memory on a 4096-bit bus may benefit certain memory-bandwidth-sensitive workloads, and its lower TDP makes it easier to cool and power. However, these advantages do not overcome the V620's substantial compute lead in the recorded benchmarks.

For users with legacy Pascal-specific software stacks or those needing maximum memory bandwidth per watt, the P100 could still serve. For everyone else, the data points firmly toward the AMD Radeon PRO V620 as the higher-performing accelerator.

FAQ

Q: Which GPU has a higher Geekbench OpenCL score?

A: The AMD Radeon PRO V620 scores 128,580, which is 61.5% higher than the NVIDIA Tesla P100 PCIe 16 GB's score of 79,605.

Q: Does the NVIDIA Tesla P100 support Vulkan?

A: Yes, it supports Vulkan 1.3, while the AMD Radeon PRO V620 supports the newer Vulkan 1.4.

Q: Which card has more memory?

A: The AMD Radeon PRO V620 has 32 GB of GDDR6 memory, while the NVIDIA Tesla P100 PCIe 16 GB has 16 GB of HBM2 memory.

Q: Which GPU has higher memory bandwidth?

A: The NVIDIA Tesla P100 PCIe 16 GB delivers 732.2 GB/s over a 4096-bit HBM2 interface, compared to the AMD Radeon PRO V620's 512.0 GB/s over a 256-bit GDDR6 interface.

Q: What is the transistor count difference?

A: The AMD Radeon PRO V620 contains 26,800 million transistors on a 520 mm² die, while the NVIDIA Tesla P100 PCIe 16 GB contains 15,300 million transistors on a 610 mm² die.

Q: Do either of these cards have display outputs?

A: No, both the AMD Radeon PRO V620 and the NVIDIA Tesla P100 PCIe 16 GB have no display outputs, as they are designed for accelerator workloads.

Head-to-Head Benchmarks

The database contains one direct comparison between these two GPUs. In Geekbench OpenCL, the AMD Radeon PRO V620 scores 128,580 against the NVIDIA Tesla P100 PCIe 16 GB's 79,605. This represents a 61.5% advantage for the AMD card. To put that in context, the P100's score is 0.3% higher than the Tesla P100 PCIe 12 GB (79,396) and 0.8% higher than the AMD Radeon RX 6850M XT (78,940), but 1.7% lower than the AMD Radeon Pro Vega 64X (80,959). The V620's score is 0.5% lower than the AMD Radeon Pro W6800X Duo (135,774), 0.8% lower than the AMD Radeon PRO W6800 (135,396), and 0.9% lower than both the NVIDIA A10M (135,230) and NVIDIA RTX 4000 Ada Generation (135,218).

The V620 also has a Geekbench Vulkan score of 144,364, which is even higher than its OpenCL result. The Tesla P100 has no recorded Vulkan benchmark in the database, so no direct comparison is possible for that API. The V620's Vulkan score indicates strong performance in modern compute workloads that leverage this API.

In terms of compute throughput, the V620 delivers 20.28 TFLOPS FP32 and 40.55 TFLOPS FP16, while the P100 delivers 9.526 TFLOPS FP32 and 19.05 TFLOPS FP16. The V620's FP32 throughput is more than double that of the P100, consistent with its OpenCL benchmark advantage. The V620 also has higher pixel and texture rates: 281.6 GPixel/s and 633.6 GTexel/s versus 127.6 GPixel/s and 297.7 GTexel/s for the P100.

The only area where the P100 leads is memory bandwidth, where its 732.2 GB/s exceeds the V620's 512.0 GB/s by roughly 43%. This is a consequence of the P100's wide 4096-bit HBM2 interface versus the V620's 256-bit GDDR6 bus. For workloads that stream large amounts of data through memory, this bandwidth advantage could partially offset the V620's compute lead, but the benchmark results show that in OpenCL compute, the V620's overall performance is decisively higher.

Specification Differences

The two cards differ in nearly every major specification category. The AMD Radeon PRO V620 uses the Navi 21 chip with RDNA 2.0 architecture on a 7 nm TSMC process, while the NVIDIA Tesla P100 uses the GP100 chip with Pascal architecture on a 16 nm TSMC process. The V620 has 26,800 million transistors on a 520 mm² die; the P100 has 15,300 million transistors on a 610 mm² die. Transistor density is 51.5M per mm² for the V620 and 25.1M per mm² for the P100.

Clock speeds favor the V620: 1825 MHz base and 2200 MHz boost versus 1190 MHz base and 1329 MHz boost for the P100. Memory differs fundamentally: the V620 has 32 GB GDDR6 on a 256-bit bus at 2000 MHz (16 Gbps effective), while the P100 has 16 GB HBM2 on a 4096-bit bus at 715 MHz (1430 Mbps effective). Bandwidth favors the P100 at 732.2 GB/s versus 512.0 GB/s.

Compute unit counts favor the V620: 4608 shading units, 288 TMUs, 128 ROPs, and 72 ray tracing cores, versus 3584 shading units, 224 TMUs, and 96 ROPs with no ray tracing cores for the P100. Pixel rate is 281.6 GPixel/s versus 127.6 GPixel/s, and texture rate is 633.6 GTexel/s versus 297.7 GTexel/s. FP32 performance is 20.28 TFLOPS versus 9.526 TFLOPS, and FP16 is 40.55 TFLOPS versus 19.05 TFLOPS, both at 2:1 ratios.

Power requirements differ: the V620 has a 300 W TDP with two 8-pin connectors and a suggested 700 W PSU, while the P100 has a 250 W TDP with one 8-pin connector and a suggested 600 W PSU. Both are dual-slot cards with no display outputs and 267 mm (10.5 inches) length. API support favors the V620 with DirectX 12 Ultimate (12_2) and Vulkan 1.4, versus DirectX 12 (12_1) and Vulkan 1.3 for the P100; both support OpenGL 4.6. The V620 uses PCIe 4.0 x16, while the P100 uses PCIe 3.0 x16. The V620 was released in November 2021 and is end-of-life; the P100 was released in June 2016, is also end-of-life, and had a launch MSRP of 5,699 USD.