NVIDIA GeForce 920MX vs NVIDIA Quadro K2000D Comparison

Head-to-Head Benchmarks

The recorded data contains a single direct comparison between these two GPUs, and it is a narrow victory for the mobile part. In the Geekbench OpenCL test, the NVIDIA GeForce 920MX scores 4068 points, while the NVIDIA Quadro K2000D trails with 3919 points. That works out to a 3.7% deficit for the Quadro, a margin that is meaningful but not decisive. The OpenCL workload is compute-oriented, and the 920MX’s advantage here suggests that its Maxwell architecture extracts more efficiency from its smaller configuration than the older Kepler design can muster with its larger one.

Looking at the broader database context, neither card is a heavyweight. The Quadro K2000D sits at the 23rd percentile among all GPUs, while the 920MX sits at the 21st percentile. Their nearest rivals tell a similar story of mediocrity. The Quadro K2000D is effectively tied with the NVIDIA Quadro 2000D (3930 points, a 0.3% difference) and the NVIDIA Quadro 2000 (3898 points, a 0.5% difference in the other direction). It also trades blows with the NVIDIA GeForce GT 745M (3953 points, 0.9% behind) and the AMD Radeon R5 Graphics (3883 points, 0.9% ahead). The 920MX, meanwhile, is 1.8% behind the NVIDIA GeForce GT 545 (3594 points), 1.9% behind the NVIDIA RTX 5000 Mobile Ada Generation (3596 points), and 2.4% behind the NVIDIA GeForce GT 735M (3616 points). Its only clear win among listed rivals is against the NVIDIA Quadro 2000M (3434 points), where it leads by 2.7%.

One curious detail emerges from these rival sets: the 920MX’s nearest rivals are mostly older or higher-end parts, yet its average benchmark score across all recorded tests is only 3528 points, far below its OpenCL score of 4068. That gap comes from the inclusion of a Geekbench Vulkan result (2987 points), which drags the average down. The Quadro K2000D has no such second benchmark recorded, so its average (3919) equals its OpenCL score. This means the two cards’ average scores (3919 vs 3528) are not directly comparable, but the head-to-head OpenCL result is, and there the 920MX wins by a slim but real margin.

Architecture Differences

The architectural gap between these two is substantial, despite both being 28 nm TSMC parts. The Quadro K2000D uses the GK107 chip, a Kepler design with 1,270 million transistors on a 118 mm² die. That yields a transistor density of 10.8 million per mm². The 920MX uses the GM108S chip, a Maxwell design with 1,020 million transistors on a much smaller 77 mm² die, giving it a higher density of 13.2 million per mm². Maxwell’s architectural refinements allow it to pack more transistors per area, and its efficiency gains are visible in the power figures.

Speaking of power, the difference is stark. The Quadro K2000D has a TDP of 51 W, while the 920MX draws just 16 W. That is a 35 W gap, and it explains why the 920MX requires no power connectors and comes as an MXM module, while the Quadro is a single-slot card with no power connectors either but a suggested PSU of 250 W. The 920MX has no suggested PSU listed, which is typical for a mobile part. The Quadro’s higher power budget buys it more raw compute, but the 920MX’s lower power demand is what enables its placement in portable devices.

The memory subsystems diverge sharply. The Quadro K2000D uses 2 GB of GDDR5 on a 128-bit bus, delivering 64.00 GB/s of bandwidth. Its memory clock is 1000 MHz, which translates to 4 Gbps effective. The 920MX also has 2 GB, but it is DDR3 on a 64-bit bus, yielding only 14.40 GB/s of bandwidth at 900 MHz (1800 Mbps effective). That is a 4.4x bandwidth advantage for the Quadro, and it is the single largest specification gap between the two. For workloads that are memory-bandwidth sensitive, the Quadro should pull ahead despite its OpenCL deficit.

The compute units also differ. The Quadro has 384 shading units, 32 texture mapping units, and 16 raster output units. The 920MX has 256 shading units, 24 TMUs, and 8 ROPs. The Quadro’s larger counts translate to higher theoretical rates: 732.7 GFLOPS FP32, 30.53 GTexel/s, and 7.632 GPixel/s. The 920MX offers 508.4 GFLOPS FP32, 23.83 GTexel/s, and 7.944 GPixel/s. Interestingly, the 920MX’s pixel rate is slightly higher (7.944 vs 7.632), which suggests better ROP efficiency despite fewer ROPs. The texture rate and FP32 throughput are clearly in the Quadro’s favor, by 28% and 44% respectively.

The API support is nearly identical: both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan. The 920MX lists Vulkan 1.4, while the Quadro lists Vulkan 1.2.175. The bus interface differs: the Quadro uses PCIe 2.0 x16, the 920MX uses PCIe 3.0 x8. In practice, the 920MX’s newer bus standard partially compensates for its narrower lane count, but for a low-power mobile GPU, bandwidth to the CPU is rarely the bottleneck.

The Verdict

The data presents a clear but nuanced verdict. For raw compute in the recorded OpenCL benchmark, the 920MX wins by 3.7%. That is a real, measurable advantage, and it comes despite the 920MX having fewer shading units, fewer TMUs, and dramatically less memory bandwidth. The Maxwell architecture’s efficiency is the likely explanation: it achieves higher performance per watt and per transistor than Kepler.

However, the Quadro K2000D is not without merit. Its 44% higher FP32 throughput and 4.4x higher memory bandwidth are massive theoretical advantages that the OpenCL test does not capture. For tasks that stress memory bandwidth, such as large texture loads or certain compute kernels, the Quadro should outperform the 920MX by a wide margin. The 920MX’s 14.40 GB/s bandwidth is a severe limitation; it is less than a quarter of the Quadro’s 64.00 GB/s.

The percentile rankings favor the Quadro slightly (23rd vs 21st), but that is a marginal difference. The nearest rival sets show both cards clustered within a few percent of their peers, indicating that neither is an outlier in either direction. The 920MX’s Vulkan score of 2987 is notably lower than its OpenCL score, suggesting that its API performance varies more than the Quadro’s single recorded result.

For a user choosing between these, the decision hinges on context. In a desktop workstation with a 250 W PSU budget, the Quadro K2000D offers more headroom for bandwidth-hungry tasks. In a portable device, the 920MX’s 16 W TDP is the only viable option, and its compute advantage in the recorded test is a bonus. There is no universal winner; the data shows two different design philosophies aimed at different physical constraints.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The NVIDIA GeForce 920MX scores 4068 points, while the NVIDIA Quadro K2000D scores 3919 points, giving the 920MX a 3.7% lead.

Q: How do the memory bandwidth figures compare?

A: The Quadro K2000D has 64.00 GB/s of bandwidth from 2 GB of GDDR5 on a 128-bit bus, while the 920MX has 14.40 GB/s from 2 GB of DDR3 on a 64-bit bus.

Q: What are the power consumption levels?

A: The Quadro K2000D has a TDP of 51 W with a suggested PSU of 250 W, while the 920MX has a TDP of 16 W and no listed PSU requirement.

Q: Which GPU has more shading units?

A: The Quadro K2000D has 384 shading units, compared to 256 shading units on the 920MX, a 50% difference.

Q: What is the FP32 performance gap?

A: The Quadro K2000D delivers 732.7 GFLOPS, while the 920MX delivers 508.4 GFLOPS, making the Quadro 44% faster in theoretical FP32 throughput.

Q: Are both GPUs the same architecture?

A: No, the Quadro K2000D uses the Kepler architecture (GK107 chip), while the 920MX uses the Maxwell architecture (GM108S chip), though both are fabricated on a 28 nm process at TSMC.

Where Each One Wins

The 920MX wins in the only recorded head-to-head test, Geekbench OpenCL, with a 3.7% margin. It also wins decisively on power efficiency, drawing 16 W versus the Quadro’s 51 W. For any use case where thermal or power constraints are paramount, such as a thin-and-light laptop, the 920MX is the clear choice. Its Vulkan score of 2987, while lower than its OpenCL score, still represents a functional API path that the Quadro does not have a recorded result for.

The Quadro K2000D wins on theoretical compute throughput. Its 732.7 GFLOPS FP32 performance and 30.53 GTexel/s texture rate outclass the 920MX’s 508.4 GFLOPS and 23.83 GTexel/s. Its 64.00 GB/s memory bandwidth is in a different league entirely, and for workloads that cannot fit in 14.40 GB/s of bandwidth, the Quadro will pull ahead regardless of the OpenCL score. The Quadro also has a higher pixel rate in theory, though the 920MX’s measured pixel rate (7.944 GPixel/s) slightly exceeds it (7.632 GPixel/s), an oddity that suggests the Maxwell ROPs are more efficient per unit.

The Quadro’s desktop form factor (single-slot, 202 mm length, 111 mm height, 2x DVI and 1x mini-DisplayPort outputs) makes it suitable for fixed workstations with multi-monitor setups. The 920MX’s MXM module form factor and portable-device-dependent outputs make it unsuitable for any desktop application. The bus interface also favors the Quadro for bandwidth to the CPU: PCIe 2.0 x16 provides more lanes than the 920MX’s PCIe 3.0 x8, though the newer protocol on the latter partially offsets this.

Specification Differences

The two GPUs differ in almost every measurable specification except for memory size, manufacturer, and process node. Both have 2 GB of memory, are made by NVIDIA, and use a 28 nm TSMC process. Beyond that, the divergence is complete.

The chip designs are different: GK107 (Kepler) with 1,270 million transistors on a 118 mm² die versus GM108S (Maxwell) with 1,020 million transistors on a 77 mm² die. Transistor density favors the 920MX at 13.2M per mm² versus 10.8M per mm². The clock speeds are not directly comparable, as the Quadro has no base or boost clock listed, while the 920MX has a 965 MHz base and 993 MHz boost. Memory clocks differ: 1000 MHz (4 Gbps effective) for the Quadro versus 900 MHz (1800 Mbps effective) for the 920MX.

Memory bandwidth is the largest gap: 64.00 GB/s versus 14.40 GB/s. Bus width is 128-bit versus 64-bit. Shading units (384 vs 256), TMUs (32 vs 24), and ROPs (16 vs 8) all favor the Quadro. Pixel rate is nearly identical (7.632 vs 7.944 GPixel/s), but texture rate (30.53 vs 23.83 GTexel/s) and FP32 (732.7 vs 508.4 GFLOPS) favor the Quadro. TDP is 51 W versus 16 W. Form factors are single-slot versus MXM module. The Quadro has a suggested PSU of 250 W; the 920MX has none. Bus interfaces are PCIe 2.0 x16 versus PCIe 3.0 x8. Display outputs are 2x DVI and 1x mini-DisplayPort versus portable-device-dependent. Vulkan support differs slightly: 1.2.175 for the Quadro, 1.4 for the 920MX. The Quadro’s release date is 2013-02-28, while the 920MX’s is 2016-03-24. The Quadro has a launch MSRP of 599 USD; the 920MX has no listed MSRP.