NVIDIA GeForce 945M vs NVIDIA GeForce GT 1010 Comparison

Where Each One Wins

The benchmark data separates these two NVIDIA parts clearly. The NVIDIA GeForce 945M wins the only recorded head-to-head test, the Geekbench OpenCL benchmark, by a substantial 20.9% margin. That single result defines the entire comparison: the 945M is the faster GPU in raw compute workloads.

Looking at the broader database, the 945M sits at the 42nd percentile of all GPUs, while the GT 1010 sits at the 38th percentile. That four-point percentile gap is small in the grand scheme of the GPU landscape, but it translates into a meaningful performance difference between these two specific cards. The 945M's average benchmark score of 8099 places it in company with the NVIDIA GRID K2 (8080, just 0.2% behind), the AMD Radeon R9 M360 (8129, 0.4% ahead), the NVIDIA GeForce GTX 650 Ti Boost (8067, 0.4% behind), and the NVIDIA GeForce GTX 950M (8135, 0.4% ahead). It is effectively neck-and-neck with a cluster of mid-range parts from its era.

The GT 1010's average score of 6698 puts it in a lower tier. Its nearest rivals include the AMD Radeon R7 M370 (6764, 1% ahead), the AMD Radeon R7 M460 (6612, 1.3% behind), the AMD Radeon HD 7730M (6581, 1.8% behind), and the AMD FirePro M5100 (6830, 1.9% ahead). The GT 1010 is competitive within its own class, but that class is simply a step below where the 945M operates.

So in terms of where each one wins: the 945M wins the compute performance contest outright. The GT 1010 wins in different areas, namely power efficiency, physical size, and modern connectivity, but not in raw benchmark performance. If the task is purely about OpenCL compute output, the 945M is the clear choice. If the task involves fitting into a small chassis or minimizing power draw, the GT 1010 has arguments in its favor that have nothing to do with benchmark scores.

Architecture Differences

These two GPUs come from different architectural generations and different manufacturing approaches. The 945M is built on the Maxwell architecture, using the GM107 chip produced on a 28 nm process at TSMC. The GT 1010 is a Pascal part, using the GP108 chip fabricated on a 14 nm process at Samsung. The process shrink from 28 nm to 14 nm is significant and explains much of the efficiency difference between the two.

The transistor counts are surprisingly close. The 945M packs 1,870 million transistors on a 148 mm² die, while the GT 1010 has 1,800 million transistors on just 74 mm². That is nearly the same transistor budget in roughly half the silicon area, which is exactly what the move from 28 nm to 14 nm enables. The transistor density figures tell the story: the 945M has 12.6 million transistors per square millimeter, while the GT 1010 reaches 24.3 million per square millimeter.

The compute resource allocation differs dramatically. The 945M carries 640 shading units, 40 texture mapping units, and 16 raster output units. The GT 1010 has only 256 shading units, 16 TMUs, and 8 ROPs. The 945M has two and a half times the shader count and double the TMU and ROP counts. Maxwell's architecture leaned on raw parallel throughput, and the 945M's wider configuration reflects that design philosophy.

Both chips lack ray tracing cores and tensor cores, so neither offers dedicated hardware for those workloads. Both support DirectX 12, though the 945M is limited to the 11_0 feature level while the GT 1010 reaches the 12_1 feature level. OpenGL support is identical at 4.6, and Vulkan support is also identical at 1.4. The GT 1010 has a modern API feature-level advantage despite its lower raw performance.

The memory architectures are fundamentally different. The 945M uses 2 GB of DDR3 on a 128-bit bus, yielding 28.80 GB/s of bandwidth. The GT 1010 uses 2 GB of GDDR5 on a 64-bit bus, yet achieves 48.06 GB/s of bandwidth. The GDDR5 memory runs at 6 Gbps effective versus the 945M's 1800 Mbps effective, and the higher speed more than compensates for the narrower bus. The GT 1010 has 67% more memory bandwidth despite having half the bus width.

Head-to-Head Benchmarks

The database contains one head-to-head benchmark between these two GPUs: Geekbench OpenCL. The 945M scores 8099 against the GT 1010's 6698, a 20.9% advantage. This is not a small margin; it is a decisive win that reflects the 945M's much larger compute core configuration.

The 945M's advantage comes from its 640 shading units versus the GT 1010's 256. Even though the GT 1010 clocks much higher, with a 1228 MHz base and 1468 MHz boost versus the 945M's 928 MHz base and 1020 MHz boost, the 945M's sheer width of execution resources wins out. The GT 1010's boost clock is 43.9% higher than the 945M's boost clock, but that cannot overcome a 2.5x shader count disadvantage.

The FP32 throughput numbers illustrate this clearly. The 945M delivers 1,305.6 GFLOPS against the GT 1010's 751.6 GFLOPS, a 73.7% advantage. Texture rate follows suit: 40.80 GTexel/s versus 23.49 GTexel/s. Pixel rate is closer, with the 945M at 16.32 GPixel/s versus the GT 1010's 11.74 GPixel/s, but the 945M still leads by 39%.

The one area where the GT 1010 posts a clear technical win is memory bandwidth, 48.06 GB/s versus 28.80 GB/s. Yet that advantage does not translate into a benchmark win in the recorded test. The OpenCL workload appears to be more sensitive to compute throughput than to memory bandwidth, which favors the 945M's architecture.

In the context of their respective rival clusters, the 945M's 8099 score sits right in the middle of a tight group ranging from 8067 to 8135. The GT 1010's 6698 sits in a lower group ranging from 6581 to 6830. The gap between the two cards, 20.9%, is larger than the spread within either card's nearest rival cluster.

Specification Differences

The two cards differ across nearly every specification category. The 945M uses the GM107 chip on Maxwell architecture at 28 nm TSMC, while the GT 1010 uses GP108 on Pascal at 14 nm Samsung. Transistor counts are close at 1,870 million versus 1,800 million, but die size differs enormously at 148 mm² versus 74 mm².

Compute units: 640 shading units, 40 TMUs, and 16 ROPs for the 945M; 256 shading units, 16 TMUs, and 8 ROPs for the GT 1010. Clock speeds: 928 MHz base and 1020 MHz boost for the 945M; 1228 MHz base and 1468 MHz boost for the GT 1010. Memory: 2 GB DDR3 on a 128-bit bus at 900 MHz with 28.80 GB/s bandwidth for the 945M; 2 GB GDDR5 on a 64-bit bus at 1502 MHz with 48.06 GB/s bandwidth for the GT 1010.

Throughput rates: 16.32 GPixel/s and 40.80 GTexel/s for the 945M; 11.74 GPixel/s and 23.49 GTexel/s for the GT 1010. FP32 compute: 1,305.6 GFLOPS versus 751.6 GFLOPS. Power: 75 W TDP versus 30 W TDP. Physical format: MXM Module with no power connectors for the 945M; Single-slot at 147 mm length with no power connectors for the GT 1010.

The bus interface differs: MXM-B (3.0) for the 945M versus PCIe 3.0 x4 for the GT 1010. Display outputs: the 945M is portable-device dependent, while the GT 1010 offers 1x DVI and 1x mini-HDMI 2.0. The GT 1010 lists a suggested PSU of 200 W, while the 945M has no suggested PSU listed. DirectX support: 12 (11_0) for the 945M versus 12 (12_1) for the GT 1010. Release dates: October 2015 for the 945M versus January 2021 for the GT 1010. Both are end-of-life.

FAQ

Q: Which GPU is faster in the recorded benchmark?

A: The NVIDIA GeForce 945M scores 8099 in Geekbench OpenCL, which is 20.9% ahead of the GT 1010's 6698.

Q: How do their power requirements compare?

A: The 945M has a 75 W TDP, while the GT 1010 has a 30 W TDP. The GT 1010 also lists a suggested PSU of 200 W, while the 945M has no suggested PSU listed.

Q: Why does the GT 1010 have higher memory bandwidth despite a narrower bus?

A: The GT 1010 uses GDDR5 memory at 6 Gbps effective on a 64-bit bus, achieving 48.06 GB/s. The 945M uses DDR3 at 1800 Mbps effective on a 128-bit bus, achieving only 28.80 GB/s.

Q: Which card has more shading units?

A: The 945M has 640 shading units, while the GT 1010 has 256. The 945M also has 40 TMUs and 16 ROPs versus the GT 1010's 16 TMUs and 8 ROPs.

Q: What is the API feature-level difference?

A: The 945M supports DirectX 12 at the 11_0 feature level, while the GT 1010 supports DirectX 12 at the 12_1 feature level. Both support OpenGL 4.6 and Vulkan 1.4.

Q: How do their physical formats differ?

A: The 945M is an MXM Module with a portable-device-dependent display output, while the GT 1010 is a single-slot card measuring 147 mm with 1x DVI and 1x mini-HDMI 2.0 outputs.

The Verdict

The data points to a straightforward conclusion for compute performance: the 945M wins by a 20.9% margin in the only recorded head-to-head benchmark. Its 640 shading units and 1,305.6 GFLOPS of FP32 throughput simply outclass the GT 1010's 256 shading units and 751.6 GFLOPS. Anyone whose priority is OpenCL compute output should pick the 945M without hesitation.

The GT 1010 has its own strengths, but they are not compute strengths. It draws only 30 W versus 75 W, making it far more suitable for low-power systems. It is a single-slot card at 147 mm length, versus the 945M's MXM module form factor, which means it fits in standard desktop PCIe slots. It offers 67% more memory bandwidth at 48.06 GB/s, supports a higher DirectX 12 feature level, and was released in January 2021 versus October 2015, so it is a much younger product.

The 945M's position in the database, at the 42nd percentile with an 8099 average score, places it in a tight cluster with the GTX 950M and R9 M360. The GT 1010, at the 38th percentile with 6698, sits with the R7 M370 and FirePro M5100. That four-percentile gap is real but modest in the broader GPU landscape.

For a builder choosing between these two, the decision hinges on the use case. The 945M is the pick for maximum compute throughput in a mobile or MXM-compatible system. The GT 1010 is the pick for a compact, low-power desktop card with modern display outputs and a higher API feature level. The benchmark data does not show the GT 1010 winning any compute test, so raw performance users should not expect it to close the gap.