Intel UHD Graphics P750 vs NVIDIA GeForce 945M Comparison

# Head-to-Head Benchmarks

The database records a single head-to-head benchmark run between the NVIDIA GeForce 945M and the Intel UHD Graphics P750, using the Geekbench OpenCL workload. The result is decisive: the NVIDIA GeForce 945M scores 8,099 points, while the Intel UHD Graphics P750 scores 6,554 points. That is a 23.6% advantage for the NVIDIA part, a substantial margin for a processor comparison in this test category. The delta is large enough that the GeForce 945M finishes the workload with a clear lead, and the gap is not a narrow one that could be attributed to run-to-run variance.

Looking at the nearest rivals in the database, the 945M's score of 8,099 places it in a cluster with the NVIDIA GRID K2 (8,080, a 0.2% gap) and the AMD Radeon R9 M360 (8,129, with the 945M trailing by 0.4%). The NVIDIA GeForce GTX 950M sits at 8,067 (0.4% behind the 945M), while the NVIDIA GeForce GTX 950M is 0.4% ahead at 8,135. This means the 945M is not an outlier; it sits inside a tightly packed performance band where a few percentage points separate the participants. The 23.6% head-to-head margin over the Intel part is therefore not just a single-test artifact; it is consistent with the two parts' overall percentile positions in the database (42nd vs 38th percentile among all GPUs).

Where Each One Wins

Based on the recorded data, the NVIDIA GeForce 945M holds the single head-to-head benchmark victory. There is no recorded benchmark in the database where the Intel UHD Graphics P750 wins against the 945M. The Intel part has no wins in the head-to-head comparison fields.

Use case split: the NVIDIA GeForce 945M is for tasks where the OpenCL compute workload is the primary concern. It wins the only recorded benchmark by a significant 23.6%, which indicates a meaningful performance advantage in general-purpose compute. The Intel UHD Graphics P750, on the other hand, is the part that appears in systems where the motherboard display output is the only graphics path. Its lower score in the same benchmark means it trails in raw compute throughput, but the database does not show a single test where it pulls ahead. So, in a straight compute comparison, the choice is the NVIDIA part. If the use case is a low-power integrated part with no dedicated memory, the Intel part has a role to play, but not in raw compute performance.

FAQ

Q: Which GPU has the higher Geekbench OpenCL score?

The NVIDIA GeForce 945M scores 8,099, while the Intel UHD Graphics P750 scores 6,554. The NVIDIA part is 23.6% higher.

Q: Is the NVIDIA GeForce 945M faster in compute workloads?

Yes, in the recorded OpenCL benchmark, the NVIDIA GeForce 945M is faster. It wins the only head-to-head test with a 23.6% lead.

Q: How do these two compare in percentile ranking?

The NVIDIA 945M sits at the 42nd percentile of all GPUs in the database, while the Intel UHD P750 sits at the 38th percentile. Both are in the lower half of the database, but the NVIDIA part ranks higher.

Q: Which card is better for OpenCL compute?

The NVIDIA GeForce 945M, the OpenCL score is 8,099 vs 6,554, so it is the better part for compute workloads.

Q: Are there any benchmark tests where the Intel UHD P750 wins?

No, in the recorded head-to-head results, the NVIDIA GeForce 945M wins the only OpenCL test. The data shows no test where the Intel part wins.

Specification Differences

The two processors differ in almost every fundamental specification. The NVIDIA GeForce 945M is built on a 28 nm process at TSMC, while the Intel UHD Graphics P750 is built on Intel's 14 nm+++ process. The transistor counts tell a different story: the NVIDIA chip packs 1,870 million transistors on a 148 mm² die, giving a transistor density of 12.6 million per mm². The Intel part has no listed transistor count, die size, or density in the database. This reflects the different design goals: the NVIDIA part is a discrete, full-featured GPU, while the Intel is an integrated graphics processor (IGP).

Clock behavior is one of the largest divisions. The GeForce 945M runs with a base clock of 928 MHz, boost of 1,020 MHz, and a memory clock of 900 MHz (1,800 Mbps effective). The Intel P750 runs with a base clock of 350 MHz, a boost of 1,300 MHz, and its memory clock is listed as "System Shared".

Memory is a major separation. The NVIDIA includes 2 GB of dedicated DDR3 memory on a 128-bit bus, delivering 28.80 GB/s of bandwidth. The Intel part uses "System Shared" memory, with a bus width and bandwidth marked as "System Shared" and "System Dependent". The shading units also differ: 640 NVIDIA vs 256 Intel. Texture mapping units count 40 vs 64, and ROPs 16 vs 32. The pixel rate is 16.32 GPixel/s for NVIDIA vs 41.60 GPixel/s for Intel, and texture rate 40.80 GTexel/s vs 83.20 GTexel/s. On paper, the Intel part has higher raw pixel and texture throughput figures, but the database OpenCL benchmark tells a different story, with NVIDIA taking the win.

FP32 compute also favors the Intel in terms of the listed values: 1,305.6 GFLOPS for NVIDIA vs 665.6 GFLOPS for Intel. FP16 is listed for NVIDIA only (1,331.2 GFLOPS at 2:1 ratio). Thermal design power is also different: 75 W for the NVIDIA part versus 15 W for the Intel. The NVIDIA is an MXM Module with MXM-B (3.0) bus interface, and Intel is an IGP. In terms of APIs, NVIDIA supports DirectX 12 (11_0), OpenGL 4.6, Vulkan 1.4; Intel supports DirectX 12 (12_1), OpenGL 4.6, Vulkan 1.4.

Architecture Differences

The architectural split is the biggest gap here. The NVIDIA GeForce 945M uses the GM107 chip with a Maxwell architecture, a design that has a dedicated GPU core. The Intel P750 uses the Rocket Lake chip with "Generation 12.1" architecture, part of the "HD Graphics-W" family as Intel puts it. The NVIDIA is a discrete GPU with its own memory and power budget; the Intel is an integrated GPU that shares system memory and runs on the ring bus.

The NVIDIA part has 640 shading units, 40 texture mapping units, and 16 ROPs. The Intel part has 256 shading units, 64 TMUs, and 32 ROPs. NVIDIA does not provide RT cores or tensor cores in either part, so ray tracing and AI tensor work are not in the spec sheet.

The memory subsystem is also fundamentally different. The NVIDIA GeForce 945M uses 2 GB of dedicated DDR3, 128-bit memory bus, with 28.80 GB/s bandwidth. The Intel P750 uses "System Shared" memory, with bus width and bandwidth "System Dependent". This is critical for real-world use: the discrete memory gives the NVIDIA part a fixed resource, while the Intel part relies on the platform's memory configuration.

The process node is a clear split too: NVIDIA at 28 nm, Intel at 14 nm+++. The clock strategy also diverges: the NVIDIA part bases at 928 MHz and boosts to 1,020 MHz, while the Intel part bases at 350 MHz and boosts to 1,300 MHz. The Intel boosts much higher, but its base is much lower, which suggests a different power and thermal envelope.

The Verdict

From the recorded data, the NVIDIA GeForce 945M is the part to get if OpenCL compute performance is the primary criterion. It wins the head-to-head benchmark by 23.6%, has a higher database percentile (42nd vs 38th), and provides a discrete memory subsystem with 2 GB of dedicated DDR3 and 28.80 GB/s of bandwidth. The Intel UHD Graphics P750 is not competitive in this specific test: it loses the only head-to-head, has a lower percentile, and uses shared system memory.

The Intel part has its place in a system where a low-power integrated solution is needed: 15 W TDP, no extra power connectors, and motherboard-dependent display outputs. But strictly from the benchmark numbers, the NVIDIA part is the stronger compute part. The data suggests that if the workload is OpenCL compute, the NVIDIA part is the one to choose. The Intel part is the one to choose when the GPU must fit into an integrated, low-power slot, not for raw performance. The database's numbers are decisive: one win for NVIDIA, zero for Intel.