NVIDIA Quadro P6000 vs NVIDIA Tesla P100 PCIe 16 GB Comparison
NVIDIA Quadro P6000
Tesla P100 PCIe 16 GB
PERFORMANCE BENCHMARKS
Analysis: NVIDIA Quadro P6000 vs NVIDIA Tesla P100 PCIe 16 GB
The NVIDIA Tesla P100 PCIe 16 GB and the NVIDIA Quadro P6000 are both Pascal-architecture workstation cards from the same generation, but they are engineered for fundamentally different tasks. The data available shows a single head-to-head benchmark result, yet the specification sheets reveal deep divergences in memory technology, compute focus, and display capabilities. This analysis walks through the hard numbers to determine where each card stands.
Head-to-Head Benchmarks
The only direct comparison point in the data is the Geekbench OpenCL test, and it yields a decisive result. The Tesla P100 PCIe 16 GB scores 79,605 points, while the Quadro P6000 scores 66,382 points. That is a 19.9% advantage for the Tesla P100. This is not a marginal lead; it is a substantial gap in raw compute throughput for this particular workload.
The significance of this win becomes clearer when placed against each card’s broader benchmark standing. The Tesla P100’s average benchmark score is 79,605, placing it in the 92nd percentile of all GPUs. Its nearest rival, the Tesla P100 PCIe 12 GB, scores 79,396, meaning the 16 GB model is 0.3% ahead. The GeForce RTX 5090 scores 79,842, which is 0.3% ahead of the P100, and the Radeon RX 6850M XT is 0.8% behind at 78,940. The P100 is clearly operating at the top tier of compute performance.
The Quadro P6000, by contrast, has an average benchmark score of 69,986, which is markedly lower than the Tesla’s 79,605. This places it in the 90th percentile, two points lower than the P100. Its nearest rival, the AMD Radeon Pro WX 8200, scores 69,870, a negligible 0.2% difference. The NVIDIA RTX A3000 Mobile is 0.2% ahead at 70,140, and the AMD Radeon RX 6600 LE is 1.2% ahead at 70,829. The P6000 sits in a competitive cluster, but that cluster is well below the performance tier occupied by the P100.
Interpreting the numbers, the Tesla P100 is not merely faster; it is in a different performance class. A 19.9% lead in OpenCL is a significant margin that suggests a fundamental advantage in compute-heavy tasks. The P6000’s score, while respectable, places it closer to mid-range competitors than to the top of the stack. The data does not show any benchmark where the P6000 wins; the sole head-to-head result is a clean sweep for the Tesla.
Architecture Differences
Both cards are built on NVIDIA’s Pascal architecture, fabricated by TSMC on a 16 nm process node, and both have the same transistor density of 25.1M per mm². However, the silicon underneath is completely different. The Tesla P100 uses the GP100 chip, which is a massive die measuring 610 mm² and containing 15,300 million transistors. The Quadro P6000 uses the GP102 chip, which is smaller at 471 mm² and packs 11,800 million transistors. The P100’s die is 29.5% larger in area and holds 29.7% more transistors, but both maintain the same density.
The memory systems are radically different, and this is where the biggest architectural divergence appears. The Tesla P100 is equipped with 16 GB of HBM2 memory on a 4096-bit bus, delivering a bandwidth of 732.2 GB/s. The Quadro P6000 has 24 GB of GDDR5X memory on a 384-bit bus, yielding a bandwidth of 432.8 GB/s. The P100 has 69.2% more memory bandwidth, which is a critical advantage for data-intensive workloads. The P6000 counters with 50% more memory capacity, which helps in scenarios where fitting larger datasets into VRAM matters more than raw throughput.
Compute resources also differ. The Tesla P100 has 3,584 shading units, 224 texture mapping units, and 96 ROPs. The Quadro P6000 has more of each: 3,840 shading units, 240 TMUs, and 96 ROPs. This gives the P6000 a 7.1% advantage in shading units and TMUs. Clock speeds reinforce this: the P6000 runs at a base of 1506 MHz and boosts to 1645 MHz, while the P100 runs at 1190 MHz base and 1329 MHz boost. That is a 26.5% higher base clock and a 23.8% higher boost clock for the P6000.
Despite these advantages, the P6000’s FP32 throughput is 12.63 TFLOPS, compared to the P100’s 9.526 TFLOPS, a 32.6% difference. However, the FP16 story is inverted: the P100 delivers 19.05 TFLOPS (2:1 ratio), while the P6000 delivers only 197.4 GFLOPS (1:64 ratio). This tells the real story. The P6000 is built for raw FP32 and rasterization, while the P100 is designed for compute acceleration, particularly where FP16 throughput matters. The P100’s pixel rate is 127.6 GPixel/s, and its texture rate is 297.7 GTexel/s. The P6000’s rates are higher at 157.9 GPixel/s and 394.8 GTexel/s, confirming its rasterization focus.
Where Each One Wins
The data points to a clear split. The Tesla P100 PCIe 16 GB wins decisively in compute-oriented benchmarks. Its 19.9% lead in OpenCL, combined with its 732.2 GB/s memory bandwidth and 19.05 TFLOPS FP16 performance, makes it the clear choice for workloads that depend on parallel compute and memory throughput. The 2:1 FP16 ratio is particularly telling; it indicates hardware designed for mixed-precision compute, which is common in scientific simulations and machine learning inferencing. The P100’s position in the 92nd percentile, competing directly with the GeForce RTX 5090, reinforces that it is a top-tier compute part.
The Quadro P6000 wins where the P100 cannot compete: display output and memory capacity. The P6000 has 1x DVI and 4x DisplayPort 1.4a outputs, while the P100 has no display outputs at all. This makes the P6000 functional as a workstation card for driving monitors. Its 24 GB of GDDR5X memory is 50% larger than the P100’s 16 GB, which is a significant advantage for tasks that require loading large textures or datasets that do not fit in 16 GB. The P6000 also has higher FP32 throughput at 12.63 TFLOPS, and higher pixel and texture rates, making it better suited for traditional 3D rendering and viewport work.
For a direct comparison, consider the benchmark scores. The P100’s average score of 79,605 is 13.7% higher than the P6000’s 69,986. In the only head-to-head test, the P100 wins by 19.9%. There is no benchmark in the data where the P6000 comes out ahead. However, the P6000’s higher clock speeds and more shading units suggest it would win in pure rasterization throughput, even though that is not captured in the provided benchmark data.
The Verdict
The data makes the choice straightforward for compute-focused buyers. The NVIDIA Tesla P100 PCIe 16 GB is the superior card for any workload that leverages OpenCL, FP16, or high memory bandwidth. Its 19.9% win in the head-to-head benchmark, its 92nd percentile standing, and its 732.2 GB/s bandwidth make it the clear pick for scientific computing, data analysis, and any task that stresses raw compute throughput. Users who need FP16 performance will find the P100’s 19.05 TFLOPS versus the P6000’s 197.4 GFLOPS to be a deciding factor; that is a 95.6x advantage in FP16 throughput. The P100 is end-of-life, but its performance class is still competitive, sitting within 0.3% of the GeForce RTX 5090.
The NVIDIA Quadro P6000 is the correct choice for professionals who need a functional workstation GPU. Its display outputs are essential for driving monitors, which the P100 cannot do. The 24 GB memory capacity is a real advantage for large-scale rendering or machine learning models that exceed 16 GB. Its higher FP32 throughput (12.63 TFLOPS) and higher pixel rate (157.9 GPixel/s) indicate stronger performance in traditional 3D rendering pipelines. The P6000’s 90th percentile ranking is still strong, and its nearest rivals are within 1.4%, meaning it sits in a competitive but lower tier.
From a pure performance-per-dollar perspective, the data does not support the P6000 over the P100 for compute tasks. The P100 is faster, has more bandwidth, and a higher percentile ranking. The P6000’s advantages are capacity and connectivity, not speed. If the workload requires display output or more than 16 GB of memory, the P6000 is the only option. Otherwise, the P100 is the better performer.
FAQ
Q: Which card is faster in the Geekbench OpenCL test?
A: The NVIDIA Tesla P100 PCIe 16 GB scores 79,605, which is 19.9% higher than the Quadro P6000’s 66,382.
Q: How does the memory bandwidth compare between the two cards?
A: The Tesla P100 has 732.2 GB/s of bandwidth using HBM2 on a 4096-bit bus, while the Quadro P6000 has 432.8 GB/s using GDDR5X on a 384-bit bus.
Q: Can the Tesla P100 drive a display?
A: No, the Tesla P100 has no display outputs, while the Quadro P6000 has 1x DVI and 4x DisplayPort 1.4a outputs.
Q: What are the FP16 performance figures for each card?
A: The Tesla P100 delivers 19.05 TFLOPS with a 2:1 FP16 ratio, while the Quadro P6000 delivers 197.4 GFLOPS with a 1:64 ratio.
Q: Which card has more memory, and what type?
A: The Quadro P6000 has 24 GB of GDDR5X memory, while the Tesla P100 has 16 GB of HBM2 memory.
Q: How do the two cards rank in overall GPU percentile?
A: The Tesla P100 is in the 92nd percentile of all GPUs, while the Quadro P6000 is in the 90th percentile.
Specification Differences
The following table lists only the fields where the two cards differ, based on the provided data.
| Specification | NVIDIA Tesla P100 PCIe 16 GB | NVIDIA Quadro P6000 |
|---|---|---|
| Chip | GP100 | GP102 |
| Generation | Tesla Pascal (Pxx) | Quadro Pascal (Px000) |
| Transistors | 15,300 million | 11,800 million |
| Die Size | 610 mm² | 471 mm² |
| Base Clock | 1190 MHz | 1506 MHz |
| Boost Clock | 1329 MHz | 1645 MHz |
| Memory Clock | 715 MHz / 1430 Mbps effective | 1127 MHz / 9 Gbps effective |
| Memory Size | 16 GB | 24 GB |
| Memory Type | HBM2 | GDDR5X |
| Memory Bus Width | 4096 bit | 384 bit |
| Memory Bandwidth | 732.2 GB/s | 432.8 GB/s |
| Shading Units | 3584 | 3840 |
| TMUs | 224 | 240 |
| Pixel Rate | 127.6 GPixel/s | 157.9 GPixel/s |
| Texture Rate | 297.7 GTexel/s | 394.8 GTexel/s |
| FP32 Performance | 9.526 TFLOPS | 12.63 TFLOPS |
| FP16 Performance | 19.05 TFLOPS (2:1) | 197.4 GFLOPS (1:64) |
| Vulkan API | 1.3 | 1.4 |
| Display Outputs | No outputs | 1x DVI, 4x DisplayPort 1.4a |
| Height | N/A | 111 mm / 4.4 inches |
| Release Date | 2016-06-19 | 2016-09-30 |
| Predecessor | Tesla Maxwell | Quadro Maxwell |
| Successor | Tesla Volta | Quadro Volta |
| Launch MSRP | 5,699 USD | 5,999 USD |
| Geekbench OpenCL Score | 79,605 | 66,382 |
| Average Benchmark Score | 79,605 | 69,986 |
| Percentile vs All GPUs | 92 | 90 |