NVIDIA P106-090 vs NVIDIA Quadro M2000 Comparison

NVIDIA Quadro M2000 and NVIDIA P106-090 are both end-of-life graphics cards from NVIDIA, but they target entirely different workloads. The M2000 is a professional workstation card built on the Maxwell architecture, while the P106-090 is a mining-focused GPU based on Pascal. Benchmark data shows the P106-090 is decisively faster in compute tasks, but the M2000 offers display outputs and a broader feature set for actual desktop use. The head-to-head results are one-sided, with the P106-090 winning both recorded tests by significant margins.

Head-to-Head Benchmarks

The benchmark comparison between these two cards is straightforward, as the P106-090 dominates both recorded tests. In Geekbench OpenCL, the P106-090 scores 21,304 points against the Quadro M2000’s 14,588 points, a delta of -31.5% from the M2000’s perspective. This means the P106-090 is roughly 46% faster in raw compute throughput, a substantial gap that reflects the architectural leap from Maxwell to Pascal. The Vulkan test tells a similar story: the P106-090 posts 18,596 points versus the M2000’s 14,475 points, a 22.2% advantage for the P106-090. These are not close contests; the P106-090 wins both head-to-head benchmarks, giving it a 2-0 record in wins.

Looking at average benchmark scores, the picture becomes more nuanced. The M2000 has an average benchmark score of 14,532 across its two recorded tests, while the P106-090 averages 13,470 across three tests, which includes a much lower 3DMark Steel Nomad DX12 score of 509 that drags its average down. This discrepancy highlights that the P106-090’s strength is in compute-oriented workloads like OpenCL and Vulkan, not necessarily in modern DirectX 12 gaming tests. The M2000’s percentile ranking is 56 versus the P106-090’s 54, suggesting the M2000 sits slightly higher relative to all GPUs in the database, despite losing the direct comparisons.

When examining nearest rivals, the M2000’s closest competitor is the GeForce GTX 965M with an average score of 14,404, a mere 0.9% difference, while the P106-090’s nearest rival is the GeForce GTX 570 at 13,515, just 0.3% off. This places both cards in similar performance tiers overall, but the P106-090’s raw compute advantage in specific APIs is undeniable. The data shows a clear trade-off: the P106-090 wins on raw speed in OpenCL and Vulkan, but the M2000 has a more consistent profile across its tested workloads.

Architecture Differences

The fundamental architectural split between these two cards explains their performance disparity. The Quadro M2000 uses the GM206 chip based on Maxwell 2.0 architecture, manufactured on a 28 nm process at TSMC. The P106-090 uses the GP106 chip based on Pascal architecture, built on a newer 16 nm process, also at TSMC. This process shrink is significant: the M2000 packs 2,940 million transistors into a 228 mm² die, yielding a density of 12.9 million transistors per square millimeter. The P106-090 crams 4,400 million transistors into a smaller 200 mm² die, achieving 22.0 million transistors per square millimeter. The newer node allows the P106-090 to fit 50% more transistors into a smaller physical area, directly enabling its higher clock speeds and performance.

Clock speeds differ dramatically. The M2000 runs at a base clock of 796 MHz with a boost of 1163 MHz, while the P106-090 starts at 1354 MHz and boosts to 1531 MHz. That is a 70% higher base clock and a 32% higher boost clock. Memory clocks also favor the P106-090: its memory runs at 2002 MHz (8 Gbps effective) versus the M2000’s 1653 MHz (6.6 Gbps effective). Even the memory bus differs, with the M2000 using a 128-bit interface and the P106-090 using a wider 192-bit bus. The result is memory bandwidth of 105.8 GB/s for the M2000 versus 192.2 GB/s for the P106-090, an 82% advantage for the newer card.

Both cards share the same shading unit count (768) and TMU count (48), but the P106-090 has 48 ROPs versus the M2000’s 32. This translates to higher pixel and texture rates: the P106-090 achieves 73.49 GPixel/s and 73.49 GTexel/s, while the M2000 manages 37.22 GPixel/s and 55.82 GTexel/s. FP32 compute is also higher on the P106-090 at 2.352 TFLOPS versus 1.786 TFLOPS. The P106-090 even lists FP16 performance at 36.74 GFLOPS with a 1:64 ratio, while the M2000 has no FP16 data. Both cards support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, so API compatibility is identical, but the underlying hardware favors the Pascal part.

Where Each One Wins

The P106-090 wins decisively in raw compute benchmarks. Its Geekbench OpenCL score of 21,304 is 46% higher than the M2000’s 14,588, and its Vulkan score of 18,596 beats the M2000’s 14,475 by 28%. For any workload that leverages OpenCL or Vulkan compute—such as scientific simulations, rendering tasks, or machine learning inference—the P106-090 is the clear choice. Its higher memory bandwidth (192.2 GB/s) and faster clocks make it better suited for memory-intensive compute tasks.

The M2000’s wins are not in raw speed but in practical usability. It has four DisplayPort 1.2 outputs, making it a functional workstation card for multi-monitor setups, while the P106-090 has no display outputs at all—it is purely a compute or mining card. The M2000 is also a single-slot card with no power connectors, drawing its 75 W entirely from the PCIe slot, whereas the P106-090 is a dual-slot card requiring a single 6-pin power connector. The M2000 is shorter at 201 mm versus the P106-090’s 250 mm, making it easier to fit in compact systems. For professional visualization, CAD, or any task requiring video output, the M2000 is the only viable option of the two.

The P106-090’s bus interface is PCIe 1.0 x1, a severe limitation that could bottleneck data transfer in systems with limited PCIe lanes, while the M2000 uses PCIe 3.0 x16. This makes the M2000 preferable for tasks that involve heavy host-to-device communication, despite its lower compute throughput. The M2000 also has a higher percentile ranking (56 vs 54), indicating it sits better relative to all GPUs in the database when considering its full benchmark profile.

FAQ

Q: Which card is faster in OpenCL compute?

A: The P106-090 is significantly faster, scoring 21,304 in Geekbench OpenCL versus the M2000’s 14,588, a 31.5% delta favoring the P106-090.

Q: Can the P106-090 be used for display output?

A: No, the P106-090 has no display outputs, while the M2000 offers four DisplayPort 1.2 connections for multi-monitor setups.

Q: How do their power requirements compare?

A: Both have a 75 W TDP and a suggested PSU of 250 W, but the M2000 uses no power connectors and is single-slot, while the P106-090 is dual-slot and requires one 6-pin connector.

Q: Which card has more memory bandwidth?

A: The P106-090 has 192.2 GB/s of bandwidth from its 3 GB GDDR5 memory on a 192-bit bus, compared to the M2000’s 105.8 GB/s from 4 GB GDDR5 on a 128-bit bus.

Q: Are these cards similar in overall performance ranking?

A: Yes, the M2000 sits at the 56th percentile of all GPUs, while the P106-090 is at the 54th percentile, despite the P106-090 winning direct head-to-head tests.

Q: Do they support the same graphics APIs?

A: Yes, both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, so API-level compatibility is identical.

Specification Differences

The key specification differences between the two cards are stark. The M2000 uses a 28 nm process with 2,940 million transistors on a 228 mm² die, while the P106-090 uses 16 nm with 4,400 million transistors on a 200 mm² die. Base clocks are 796 MHz for the M2000 versus 1354 MHz for the P106-090, with boost clocks of 1163 MHz and 1531 MHz respectively. Memory configurations diverge: the M2000 has 4 GB GDDR5 on a 128-bit bus with 105.8 GB/s bandwidth, while the P106-090 has 3 GB GDDR5 on a 192-bit bus with 192.2 GB/s bandwidth.

Pixel rate is 37.22 GPixel/s for the M2000 versus 73.49 GPixel/s for the P106-090, and texture rate is 55.82 GTexel/s versus 73.49 GTexel/s. FP32 compute is 1.786 TFLOPS versus 2.352 TFLOPS. The M2000 has 32 ROPs, while the P106-090 has 48 ROPs, though both have 768 shading units and 48 TMUs. Physical dimensions differ: the M2000 is 201 mm long and 111 mm tall, while the P106-090 is 250 mm long. The M2000 is single-slot with no power connectors, while the P106-090 is dual-slot with one 6-pin connector. The bus interface is PCIe 3.0 x16 for the M2000 versus PCIe 1.0 x1 for the P106-090. Display outputs are 4x DisplayPort 1.2 on the M2000, and none on the P106-090. Release dates also differ, with the M2000 launching in April 2016 and the P106-090 in July 2017.

The Verdict

The data points to a clear split based on use case. For pure compute performance, the P106-090 is the superior card, with 46% higher OpenCL scores, 28% higher Vulkan scores, higher clock speeds, and nearly double the memory bandwidth. Its 2.352 TFLOPS of FP32 performance and 48 ROPs make it a more capable compute engine. However, the P106-090’s lack of display outputs, PCIe 1.0 x1 interface, and dual-slot design with a 6-pin power connector make it unsuitable for general desktop use or professional workstation tasks.

The Quadro M2000, despite losing both head-to-head benchmarks, is the more versatile card. It offers four DisplayPort outputs, a single-slot design with no power connectors, a compact 201 mm length, and a full PCIe 3.0 x16 interface. Its 4 GB of memory, while slower, is more than the P106-090’s 3 GB. The M2000 also holds a slightly higher percentile ranking (56 vs 54), suggesting it performs better relative to its peers across a broader range of workloads.

For a professional user needing a reliable workstation card with multi-monitor support, the M2000 is the only choice between these two. For a miner or compute-focused builder who does not need video output and can work around the PCIe 1.0 x1 limitation, the P106-090 offers significantly more raw performance. The verdict is simple: the P106-090 wins on speed, but the M2000 wins on usability. Choose based on whether you need to see the output or just crunch the numbers.