NVIDIA GeForce GTX 660 vs NVIDIA Quadro K5100M Comparison

Head-to-Head Benchmarks

The benchmark data shows a clear overall winner in the direct comparisons between the NVIDIA Quadro K5100M and the NVIDIA GeForce GTX 660, with the Quadro taking both recorded tests. The most dramatic separation occurs in the Geekbench Metal workload, where the Quadro K5100M scores 8315 against the GTX 660's 4305. That is a 93.1% advantage for the Quadro, a near doubling of performance in this particular API test. The Metal result is striking because it far exceeds what the underlying specification gap would suggest, indicating that the Quadro's architecture and driver stack handle this workload with considerably more efficiency.

The second direct comparison, Geekbench OpenCL, is much tighter. The Quadro K5100M posts 11771, while the GeForce GTX 660 manages 11347. The delta shrinks to just 3.7% in favor of the Quadro. This is a far more competitive result, and it suggests that in compute-oriented OpenCL tasks, the two cards are nearly peers despite their different positioning. The GTX 660's higher clock speeds help close the gap that the Quadro opens through its larger chip and higher shading unit count.

Looking at the wider database context, the average benchmark scores reinforce the Quadro's edge. The K5100M averages 10043 across all recorded tests, placing it in the 48th percentile of all GPUs. The GTX 660 averages 9022, landing in the 45th percentile. The 11.3% difference in average score is smaller than the Metal delta but larger than the OpenCL delta, reflecting the mixed nature of the two cards' performance profiles.

The nearest rival data places both cards in interesting company. The Quadro K5100M sits within 0.3% of the AMD Radeon R9 M375, which averages 10070, and is 0.3% ahead of the AMD Radeon Pro 5300M at 10013. It also leads the NVIDIA GeForce GTX 870M by 0.8% and the NVIDIA Quadro 6000 by 2%. These are marginal spreads, indicating that the K5100M occupies a tightly contested performance tier where small architectural differences determine placement.

The GeForce GTX 660, by contrast, finds itself bracketed by very different hardware. It trails the NVIDIA TITAN V CEO Edition by 0.2% and the NVIDIA GeForce GTX 560 by 0.4%, while leading the AMD Radeon 550X by 1.2% and the AMD Radeon Pro WX 5100 by 1.8%. The GTX 660's average score of 9022 places it in a cluster where the deltas are all within 2%, meaning the card is essentially performance-equivalent to several very different GPUs from different generations and market segments.

The wins tally is straightforward: the Quadro K5100M wins 2 head-to-head benchmarks, while the GeForce GTX 660 wins none. However, the magnitude of those wins matters. The Metal test is a landslide, but the OpenCL test is close enough that real-world workloads could easily flip the result depending on driver maturity and application optimization. The data does not support calling the GTX 660 a weak card; it supports calling the Quadro a specialist that excels in specific environments.

The Verdict

The data points to a simple conclusion for most buyers: the NVIDIA Quadro K5100M is the stronger card in overall benchmark performance, but the choice depends heavily on the intended workload. The Quadro's 93.1% lead in Geekbench Metal is the single most decisive data point in this comparison. Anyone running Metal-based applications, particularly in an Apple ecosystem context, should treat the K5100M as the only serious option between the two. The OpenCL result, at just 3.7% in favor of the Quadro, suggests that pure compute workloads see only a modest benefit from choosing the Quadro over the GTX 660.

The GTX 660, with its higher base clock of 980 MHz and boost of 1032 MHz, compensates for its smaller configuration through raw frequency. Its average benchmark score of 9022, while lower, is not embarrassingly far behind the Quadro's 10043. The GTX 660 also carries a launch MSRP of 229 USD, a fact worth noting for historical context, though the card is now end-of-life. The Quadro K5100M, also end-of-life, launched without a recorded MSRP in the database.

For users who prioritize the specific API performance that Metal tests capture, the Quadro K5100M is the unambiguous choice. For users whose workloads are OpenCL-heavy, the 3.7% delta is small enough that other factors, such as the GTX 660's dual-slot desktop form factor and standard PCIe 3.0 x16 interface, might make it the more practical option. The Quadro's MXM Module slot width and portable-device-dependent display outputs indicate it was designed for mobile workstations, not desktop builds. The GTX 660, with its 241 mm length, 2x DVI, HDMI 1.4a, and DisplayPort 1.2 outputs, is a conventional desktop card.

The percentile rankings tell a similar story. The Quadro sits in the 48th percentile of all GPUs, while the GTX 660 sits in the 45th. That three-percentile gap is modest in the grand scheme of the database, but it consistently favors the Quadro across average scores and both direct tests. The verdict is therefore: the Quadro K5100M wins on performance, while the GTX 660 wins on integration ease for standard desktop systems.

Architecture Differences

Both GPUs are built on NVIDIA's Kepler architecture and manufactured by TSMC on a 28 nm process node, so the fundamental architectural generation is identical. The differences lie in the specific chip implementations. The Quadro K5100M uses the GK104 chip, while the GeForce GTX 660 uses the GK106. This is a significant distinction, as GK104 is a larger, more fully featured die. The Quadro packs 3,540 million transistors on a 294 mm² die, yielding a transistor density of 12.0 million per mm². The GTX 660's GK106 contains 2,540 million transistors on a 221 mm² die, with a density of 11.5 million per mm². The Quadro's chip is both physically larger and denser, reflecting its positioning as a higher-end part.

The execution resources differ accordingly. The Quadro K5100M features 1536 shading units, 128 texture mapping units, and 32 ROPs. The GTX 660 is configured with 960 shading units, 80 TMUs, and 24 ROPs. That gives the Quadro a 60% advantage in shading units, a 60% advantage in TMUs, and a 33% advantage in ROPs. These are substantial resource gaps that explain the Quadro's performance lead in most scenarios.

Neither card includes ray tracing cores or tensor cores, which is expected given their Kepler-generation design. Both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, so the API feature sets are identical. The architecture is the same Kepler family, but the GK104 in the Quadro is a more robust implementation of that architecture than the GK106 in the GTX 660.

The production status for both is end-of-life. The Quadro K5100M's generation is listed as "Quadro Kepler-M (Kx100M)" and its predecessor is Quadro Fermi-M, with Quadro Maxwell-M as its successor. The GTX 660 belongs to the GeForce 600 generation, succeeding GeForce 500 and preceding GeForce 700. These lineage details show that both cards occupied similar temporal positions in NVIDIA's lineup, but served different market segments: professional mobile workstations versus mainstream desktop gaming.

Specification Differences

The clock speeds diverge notably. The GTX 660 runs at a 980 MHz base clock and boosts to 1032 MHz, while the Quadro K5100M is locked at 771 MHz for both base and boost. The GTX 660's 27% higher base clock is a major reason it remains competitive in OpenCL despite its smaller chip. The memory clocks also differ: the Quadro runs at 900 MHz with 3.6 Gbps effective data rate, while the GTX 660 runs at 1502 MHz with 6 Gbps effective. The GTX 660's memory operates at a substantially higher frequency.

Memory capacity and bus width favor the Quadro. The K5100M has 8 GB of GDDR5 on a 256 bit bus, delivering 115.2 GB/s of bandwidth. The GTX 660 has 2 GB of GDDR5 on a 192 bit bus, delivering 144.2 GB/s. Despite having one quarter the capacity and a narrower bus, the GTX 660 achieves higher bandwidth thanks to its faster memory clock. The Quadro's much larger 8 GB frame buffer is clearly aimed at professional workloads like large datasets and multi-display configurations, while the GTX 660's smaller but faster memory suits gaming.

Compute and fill rates follow the resource counts. The Quadro K5100M produces 24.67 GPixel/s and 98.69 GTexel/s, with FP32 performance of 2.369 TFLOPS. The GTX 660 produces 20.64 GPixel/s and 82.56 GTexel/s, with FP32 performance of 1.981 TFLOPS. The Quadro leads in every throughput metric, though the pixel and texture rate gaps are smaller than the shading unit gap, reflecting the GTX 660's high clocks.

Power and form factor differences are stark. The Quadro K5100M draws 100 W TDP and uses an MXM Module slot width with no power connectors, confirming its mobile workstation design. The GTX 660 draws 140 W TDP, uses a dual-slot cooler with a single 6-pin power connector, and suggests a 300 W power supply. The GTX 660's bus interface is PCIe 3.0 x16, while the Quadro uses MXM-B (3.0). Display outputs also differ: the Quadro is "portable device dependent," while the GTX 660 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.

The release dates place the GTX 660 earlier, on September 5, 2012, while the Quadro K5100M arrived on July 22, 2013. The GTX 660 measures 241 mm in length. The Quadro's dimensions are not recorded in the database.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA Quadro K5100M averages 10043 across all recorded benchmarks, while the NVIDIA GeForce GTX 660 averages 9022.

Q: How large is the performance gap in Geekbench Metal?

A: The Quadro K5100M scores 8315 in Geekbench Metal, which is 93.1% higher than the GTX 660's score of 4305.

Q: Is the GTX 660 competitive in any benchmark?

A: In Geekbench OpenCL, the GTX 660 scores 11347, just 3.7% behind the Quadro K5100M's 11771.

Q: Do both GPUs support the same APIs?

A: Yes, both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

Q: What are the memory capacities of each card?

A: The Quadro K5100M has 8 GB of GDDR5, while the GeForce GTX 660 has 2 GB of GDDR5.

Q: What is the GTX 660's launch MSRP?

A: The GeForce GTX 660 had a launch MSRP of 229 USD.