Intel HD Graphics 630 vs NVIDIA GeForce 930M Comparison

The Verdict

The data presents a clear but narrow overall winner: the NVIDIA GeForce 930M takes both head-to-head benchmark victories, yet the margin in one test is razor-thin. In Geekbench OpenCL, the 930M scores 5046 against the Intel HD Graphics 630's 3587, a decisive 40.7% advantage. In Geekbench Vulkan, the gap shrinks dramatically: 3729 versus 3540, a 5.3% edge. The 930M's average benchmark score of 4388 places it at the 26th percentile of all GPUs, while the HD 630's 4075 average sits at the 24th percentile. The 930M is the choice for users prioritizing compute-heavy OpenCL workloads or any task where raw FP32 throughput matters. The HD 630, conversely, is better suited for systems where power efficiency is paramount, given its 15 W TDP against the 930M's 33 W, and for users who need DirectX 12_1 feature support rather than the 930M's 11_0 implementation. For general casual use, the Vulkan scores suggest near-parity, making the HD 630 the more sensible pick in a modern ultraportable context, provided the user does not depend on OpenCL acceleration.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The 930M is built on NVIDIA's Maxwell architecture, fabricated on a 28 nm process at TSMC, with the GM108S chip containing 1,020 million transistors on a 77 mm² die. The HD 630 is Intel's Generation 9.5 architecture, using the Kaby Lake GT2 chip on Intel's 14 nm++ process. This process advantage allows Intel to integrate the GPU onto the CPU die, whereas the 930M is a discrete part. The transistor density reflects this: the 930M achieves 13.2M transistors per mm², while Intel does not report a comparable figure for the HD 630 due to its integrated nature.

Compute resources differ significantly. The 930M has 384 shading units, 24 TMUs, and 8 ROPs. The HD 630 has 192 shading units, 24 TMUs, but only 3 ROPs. Despite half the shading units, the HD 630's texture rate is higher at 24.00 GTexel/s versus 13.18 GTexel/s for the 930M, a direct result of the Intel part's higher boost clock of 1000 MHz against the NVIDIA's 549 MHz base and boost. Pixel rate favors the 930M at 4.392 GPixel/s versus 3.000 GPixel/s, thanks to the 8 ROPs versus 3. FP32 throughput is close: 421.6 GFLOPS for NVIDIA and 384.0 GFLOPS for Intel. Notably, the HD 630 supports FP16 at 768.0 GFLOPS with a 2:1 ratio, a feature the 930M lacks entirely.

Memory architecture is another major divergence. The 930M uses 2 GB of dedicated DDR3 on a 64-bit bus, delivering 12.80 GB/s of bandwidth. The HD 630 relies on System Shared memory, with bandwidth described as System Dependent. This means the HD 630's memory performance is entirely contingent on the host system's RAM configuration, while the 930M has predictable, fixed bandwidth. The 930M interfaces via PCIe 3.0 x8, whereas the HD 630 uses a Ring Bus. API support also differs: the 930M offers DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4, while the HD 630 provides DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The 12_1 feature level on Intel is a meaningful advantage for newer game titles.

Head-to-Head Benchmarks

The Geekbench OpenCL test is the standout metric, showing a 40.7% lead for the 930M. The 930M's score of 5046 dwarfs the HD 630's 3587. This delta is substantial and likely stems from the combination of dedicated memory with fixed bandwidth, higher ROP count, and more shading units. In OpenCL workloads—often compute-heavy tasks like video encoding or physics simulation—the 930M is clearly the superior part. The data suggests that for any user running OpenCL-accelerated applications, the 930M provides a significantly smoother experience.

The Geekbench Vulkan test tells a different story. Here, the 930M scores 3729 against the HD 630's 3540, a mere 5.3% difference. This near-parity in Vulkan indicates that the HD 630's architectural efficiency, higher clock speed, and modern DirectX 12_1 support largely compensate for its lower shading unit count and inferior memory bandwidth in graphics-centric workloads. Vulkan's low-level nature may also favor the HD 630's integrated design, which avoids PCIe transfer overhead. The practical implication is that for gaming or graphics rendering using Vulkan, the two GPUs perform almost identically, making the HD 630's lower power draw a more compelling factor.

The 930M wins both head-to-head tests, but the average benchmark scores of 4388 and 4075 respectively (a 7.7% gap) reflect the OpenCL dominance being partially offset by the Vulkan closeness. The 930M's nearest rival, the GeForce GT 645M, scores 4411, a 0.5% delta against the 930M, placing the 930M in the same performance tier as a mid-range Kepler mobile part. The HD 630's nearest rival, the AMD Radeon RX 9060 XT 8 GB, scores 4093, a 0.4% delta against the HD 630, which is a curious comparison given the massive difference in market positioning, but the data shows the HD 630's average is within 1.5% of that AMD part.

Specification Differences

| Specification | NVIDIA GeForce 930M | Intel HD Graphics 630 |

|---|---|---|

| Architecture | Maxwell | Generation 9.5 |

| Process Node | 28 nm | 14 nm++ |

| Foundry | TSMC | Intel |

| Transistors | 1,020 million | Not reported |

| Die Size | 77 mm² | Not reported |

| Base Clock | 549 MHz | 350 MHz |

| Boost Clock | 549 MHz | 1000 MHz |

| Memory Size | 2 GB | System Shared |

| Memory Type | DDR3 | System Shared |

| Memory Bus Width | 64 bit | System Shared |

| Memory Bandwidth | 12.80 GB/s | System Dependent |

| Shading Units | 384 | 192 |

| TMUs | 24 | 24 |

| ROPs | 8 | 3 |

| Pixel Rate | 4.392 GPixel/s | 3.000 GPixel/s |

| Texture Rate | 13.18 GTexel/s | 24.00 GTexel/s |

| FP32 | 421.6 GFLOPS | 384.0 GFLOPS |

| FP16 | Not reported | 768.0 GFLOPS (2:1) |

| TDP | 33 W | 15 W |

| Bus Interface | PCIe 3.0 x8 | Ring Bus |

| DirectX | 12 (11_0) | 12 (12_1) |

| Vulkan | 1.4 | 1.3 |

| Release Date | 2015-03-12 | 2016-08-29 |

The clock speed difference is stark: the HD 630 boosts to 1000 MHz, nearly double the 930M's fixed 549 MHz. This explains the texture rate advantage despite fewer TMUs (24 each, but higher clocks win). The ROP deficit for Intel (3 versus 8) is severe, limiting pixel fill. FP16 support on Intel is notable, offering 768.0 GFLOPS, which can accelerate certain compute tasks if the software leverages half-precision. The 930M's FP32 output is 421.6 GFLOPS, only 9.8% higher than Intel's 384.0 GFLOPS, yet the OpenCL benchmark shows a 40.7% gap, indicating that memory bandwidth and driver optimization play a larger role than raw ALU count. The HD 630's TDP of 15 W is less than half the 930M's 33 W, a decisive factor for laptop battery life and thermals.

FAQ

Q: Which GPU wins in OpenCL performance?

A: The NVIDIA GeForce 930M wins decisively, scoring 5046 in Geekbench OpenCL against the Intel HD Graphics 630's 3587, a 40.7% advantage.

Q: How close are the two in Vulkan performance?

A: The 930M leads with a Geekbench Vulkan score of 3729 versus 3540 for the HD 630, a narrow 5.3% margin that suggests near-parity in Vulkan-based graphics workloads.

Q: Does the Intel HD Graphics 630 support DirectX 12_1?

A: Yes, the HD 630 supports DirectX 12 (12_1), while the NVIDIA GeForce 930M only supports DirectX 12 (11_0), giving Intel a feature-level advantage for newer game APIs.

Q: What is the memory bandwidth difference?

A: The 930M has 2 GB of dedicated DDR3 on a 64-bit bus with 12.80 GB/s bandwidth. The HD 630 uses System Shared memory with System Dependent bandwidth, meaning its performance relies on the host system's RAM.

Q: Which GPU has a higher pixel fill rate?

A: The 930M has a pixel rate of 4.392 GPixel/s, outperforming the HD 630's 3.000 GPixel/s, due to the NVIDIA part having 8 ROPs versus Intel's 3 ROPs.

Q: How do their average benchmark scores compare?

A: The 930M's average benchmark score is 4388, placing it at the 26th percentile of all GPUs. The HD 630's average is 4075, at the 24th percentile. The 930M is 7.7% higher on average.