GPU Comparison

The data places the NVIDIA GeForce 920MX and Intel HD Graphics P4600 at nearly the same tier of overall performance, separated by only a few percentage points in average benchmark score. The 920MX averages 3528 across its tested workloads, while the P4600 averages 3389. Despite this close overall standing, the available benchmark results reveal a clear, if narrow, advantage for the discrete NVIDIA solution, particularly in compute-oriented tasks. The following analysis breaks down where each part holds an edge, who should choose which, and what the underlying specifications reveal about their respective capabilities.

Where Each One Wins

Based strictly on the available head-to-head benchmark data, the NVIDIA GeForce 920MX is the outright winner. It secures the only direct comparison in the pack, the Geekbench OpenCL test, with a score of 4068 against the Intel HD Graphics P4600’s 3389, a 20% margin. This is a substantial lead in a compute-heavy workload, suggesting the 920MX is better suited for tasks that leverage general-purpose GPU computation, such as OpenCL-accelerated applications.

The Intel HD Graphics P4600, by contrast, does not win any direct benchmark comparison in this dataset. Its only recorded score is that same Geekbench OpenCL result of 3389. However, its relevance is not zero. Its average benchmark score of 3389 places it at the 20th percentile of all GPUs, which is only one percentile point below the 920MX’s 21st percentile. This indicates that while the 920MX leads in raw compute, the P4600 is not dramatically far behind in overall standing. The P4600’s wins, if any, would have to come from scenarios not captured in this limited benchmark set, for instance, its integrated nature means it shares system memory and requires no separate power delivery, but the data here does not quantify any such advantage.

The 920MX also demonstrates a wider performance envelope. Its Geekbench Vulkan score of 2987, while not directly compared to the P4600 (which has no Vulkan result in this pack), shows that the NVIDIA part has been tested across multiple modern APIs. The P4600’s single benchmark result limits the scope of its demonstrated capability. In terms of pure wins from the data, the 920MX takes the head-to-head 1–0, and its higher average score reinforces that position.

The Verdict

The choice between these two parts depends on whether the priority is raw compute performance or system integration. From the data, the NVIDIA GeForce 920MX is the clear performance pick. It leads the only shared benchmark by 20%, and its average score of 3528 is 4.1% higher than the P4600’s 3389. For any user whose workload involves OpenCL compute or who wants the higher percentile ranking (21 vs. 20), the 920MX is the data-backed choice.

The Intel HD Graphics P4600 is the fallback for scenarios where a discrete GPU is not an option. Its specifications show it is an integrated graphics processor (IGP) with a 22 nm process node and a 84 W TDP, whereas the 920MX is a 16 W MXM module. The P4600 uses system-shared memory, making it simpler to integrate into a motherboard design, but the benchmark data does not show it winning any performance test. Therefore, the verdict from the numbers is straightforward: pick the 920MX for better compute performance, and only consider the P4600 if the platform requires an IGP with no add-in card. The 920MX also has a higher OpenGL support (4.6 vs. 4.3) and a more recent Vulkan version (1.4 vs. 1.0), which may matter for application compatibility.

Head-to-Head Benchmarks

The sole direct comparison in the FACT PACK is the Geekbench OpenCL test. The NVIDIA GeForce 920MX scores 4068, while the Intel HD Graphics P4600 scores 3389. The delta is 20% in favor of the 920MX. This is the largest and only quantified head-to-head margin, so it carries the full weight of the performance difference between these two parts.

Breaking down why the 920MX wins this test, the specifications offer clues. The 920MX has 256 shading units, 24 texture mapping units (TMUs), and 8 raster output units (ROPs). The P4600 has 160 shading units, 20 TMUs, and only 2 ROPs. The 920MX also has a much higher pixel rate at 7.944 GPixel/s versus 2.400 GPixel/s, and a higher FP32 throughput at 508.4 GFLOPS versus 384.0 GFLOPS. These raw compute advantages directly translate to the OpenCL score lead.

The texture rates are nearly identical, 23.83 GTexel/s for the 920MX and 24.00 GTexel/s for the P4600, which explains why the gap is not even larger. However, the 920MX’s dedicated 2 GB of DDR3 memory on a 64-bit bus with 14.40 GB/s bandwidth, versus the P4600’s system-shared memory with system-dependent bandwidth, likely contributes to the 20% compute win. The 920MX also boosts its clock from 965 MHz base to 993 MHz boost, while the P4600 runs at a 350 MHz base and 1200 MHz boost; the higher sustained clocks on the NVIDIA part help in sustained compute loads. In summary, the 920MX wins the only head-to-head by a decisive margin, driven by more shading units, higher pixel throughput, and dedicated memory.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce 920MX, with an average benchmark score of 3528, compared to the Intel HD Graphics P4600’s 3389.

Q: What is the performance difference in the Geekbench OpenCL test?

A: The 920MX scores 4068, and the P4600 scores 3389, giving the 920MX a 20% lead.

Q: Does the Intel HD Graphics P4600 win any benchmark in this comparison?

A: No. The P4600 has zero wins in the head-to-head data; the 920MX holds all 1 win.

Q: How do their percentile rankings compare?

A: The 920MX sits at the 21st percentile of all GPUs, while the P4600 is at the 20th percentile, a difference of one percentile point.

Q: Which GPU supports a newer version of Vulkan?

A: The NVIDIA GeForce 920MX supports Vulkan 1.4, while the Intel HD Graphics P4600 supports Vulkan 1.0.

Q: What is the memory configuration difference?

A: The 920MX has 2 GB of dedicated DDR3 memory on a 64-bit bus with 14.40 GB/s bandwidth. The P4600 uses system-shared memory, with bandwidth described as system-dependent.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The NVIDIA GeForce 920MX is built on the Maxwell architecture, using the GM108S chip manufactured on a 28 nm process at TSMC. It contains 1,020 million transistors on a 77 mm² die, yielding a transistor density of 13.2 million per square millimeter. The Intel HD Graphics P4600 is based on the Haswell GT2 chip, using Intel’s Generation 7.5 architecture on a 22 nm process at Intel’s own foundry. The P4600’s transistor count and die size are not listed, but the process node is smaller.

The core configurations diverge sharply. The 920MX fields 256 shading units, 24 TMUs, and 8 ROPs. The P4600 has 160 shading units, 20 TMUs, and just 2 ROPs. The ROP count difference is stark, 8 versus 2, which heavily impacts the pixel rate: 7.944 GPixel/s for the 920MX versus 2.400 GPixel/s for the P4600. The shading unit advantage (256 vs. 160) drives the FP32 throughput difference of 508.4 GFLOPS versus 384.0 GFLOPS.

Memory architecture is another major divergence. The 920MX is a discrete GPU with 2 GB of DDR3 on a 64-bit bus, providing 14.40 GB/s of dedicated bandwidth. The P4600 is an integrated part with no dedicated memory; it uses system-shared memory, and its bandwidth is system-dependent. This means the P4600’s performance scales with the host system’s memory configuration, while the 920MX has predictable, fixed bandwidth.

Feature support also differs. The 920MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4. The P4600 supports DirectX 12 (11_1), OpenGL 4.3, and Vulkan 1.0. The NVIDIA part has a higher OpenGL version and a much newer Vulkan revision. Both parts are marked as end-of-life, but the 920MX has a known predecessor (GeForce 800M) and successor (GeForce 10 Mobile), while the P4600 has neither listed.

Specification Differences

The following specifications differ between the NVIDIA GeForce 920MX and the Intel HD Graphics P4600:

• Process Node: The 920MX uses 28 nm, while the P4600 uses 22 nm.
• Transistors: The 920MX has 1,020 million; the P4600’s count is not listed.
• Die Size: The 920MX is 77 mm²; the P4600’s size is not listed.
• Transistor Density: The 920MX has 13.2M per mm²; the P4600’s density is not listed.
• Base Clock: 965 MHz for the 920MX versus 350 MHz for the P4600.
• Boost Clock: 993 MHz for the 920MX versus 1200 MHz for the P4600.
• Memory Size: 2 GB for the 920MX; system shared for the P4600.
• Memory Type: DDR3 for the 920MX; system shared for the P4600.
• Memory Bus Width: 64 bit for the 920MX; system shared for the P4600.
• Memory Bandwidth: 14.40 GB/s for the 920MX; system dependent for the P4600.
• Shading Units: 256 for the 920MX versus 160 for the P4600.
• TMUs: 24 for the 920MX versus 20 for the P4600.
• ROPs: 8 for the 920MX versus 2 for the P4600.
• Pixel Rate: 7.944 GPixel/s for the 920MX versus 2.400 GPixel/s for the P4600.
• Texture Rate: 23.83 GTexel/s for the 920MX versus 24.00 GTexel/s for the P4600.
• FP32: 508.4 GFLOPS for the 920MX versus 384.0 GFLOPS for the P4600.
• TDP: 16 W for the 920MX versus 84 W for the P4600.
• Slot Width: MXM Module for the 920MX; IGP for the P4600.
• Power Connectors: None for the 920MX; not listed for the P4600.
• Bus Interface: PCIe 3.0 x8 for the 920MX; Ring Bus for the P4600.
• Display Outputs: Portable Device Dependent for the 920MX; Motherboard Dependent for the P4600.
• DirectX Support: 12 (11_0) for the 920MX versus 12 (11_1) for the P4600.
• OpenGL Support: 4.6 for the 920MX versus 4.3 for the P4600.
• Vulkan Support: 1.4 for the 920MX versus 1.0 for the P4600.
• Release Date: The 920MX launched on 2016-03-24; the P4600 launched on 2013-05-31.
• Average Benchmark Score: 3528 for the 920MX versus 3389 for the P4600.
• Percentile vs. All GPUs: 21st for the 920MX versus 20th for the P4600.