Intel HD Graphics 630 vs NVIDIA GeForce GT 545 Comparison

Head-to-Head Benchmarks

The only directly comparable benchmark in the database is Geekbench OpenCL, and the results are remarkably tight. The NVIDIA GeForce GT 545 scores 3594, while the Intel HD Graphics 630 scores 3587. That is a delta of only 0.2% in favor of the NVIDIA card. In practical terms, these two GPUs are essentially identical in raw compute throughput for OpenCL workloads. The GT 545's victory here is so marginal that it falls within run-to-run variance for most testing methodologies.

Looking at the broader benchmark picture, the Intel HD Graphics 630 has three recorded scores: Geekbench Metal at 5099, Geekbench OpenCL at 3587, and Geekbench Vulkan at 3540. Its average benchmark score across all tests is 4075. The GT 545 has only one recorded benchmark, Geekbench OpenCL at 3594, giving it an average of 3594. The Intel part's average is 13.4% higher, but that is driven entirely by the Metal score, which the NVIDIA card cannot run due to its lack of Metal support.

The Intel HD Graphics 630 sits at the 24th percentile of all GPUs in the database, while the GT 545 sits at the 21st percentile. That 3-percentile gap reflects the Intel part's additional benchmark coverage rather than a fundamental performance advantage. The nearest rivals for the Intel part include the AMD Radeon RX 9060 XT 8 GB at 4093 (0.4% ahead), the NVIDIA Quadro K2100M at 4151 (1.8% ahead), and the NVIDIA GeForce GT 755M at 4033 (1% behind). The GT 545's nearest rivals include the NVIDIA RTX 5000 Mobile Ada Generation at 3596 (0.1% ahead), the AMD Radeon HD 6770 at 3649 (1.5% ahead), and the NVIDIA GeForce GTX 1050 at 3629 (1% ahead).

The takeaway from the head-to-head data is that neither card has a meaningful edge in the one test where they can be directly compared. The GT 545 wins, but by a margin that is statistically indistinguishable from a tie. If you are choosing solely on OpenCL performance, flip a coin.

Architecture Differences

The Intel HD Graphics 630 is built on Intel's Generation 9.5 architecture, specifically the Kaby Lake GT2 chip. It uses a 14 nm++ process node fabricated by Intel. The GT 545 uses NVIDIA's Fermi 2.0 architecture, based on the GF116 chip, manufactured on a 40 nm process at TSMC. The process node difference is substantial: 14 nm++ versus 40 nm means Intel's part has a major transistor density advantage, even though the database does not record a transistor count for the Intel chip. The GT 545 has 1,170 million transistors on a 238 mm² die, giving a transistor density of 4.9M per mm². The Intel part's density is not listed, but the node advantage implies significantly higher density.

The Intel HD Graphics 630 has 192 shading units, 24 texture mapping units, and only 3 raster operations pipelines. The GT 545 has 144 shading units, 24 TMUs, and 16 ROPs. The Intel part has more shaders and the same TMU count, but the GT 545 has over five times the ROP count. This shows up in the pixel rates: the GT 545 delivers 4.320 GPixel/s versus 3.000 GPixel/s for the Intel part, a 44% advantage for NVIDIA. The texture rates tell a different story: Intel's 24.00 GTexel/s beats NVIDIA's 17.28 GTexel/s by 39%. FP32 compute is close, with the GT 545 at 414.7 GFLOPS and the Intel part at 384.0 GFLOPS, a 7.4% edge for NVIDIA.

Memory is a major differentiator. The GT 545 has 1536 MB of dedicated DDR3 memory on a 192-bit bus, delivering 38.40 GB/s of bandwidth. The Intel HD Graphics 630 uses system shared memory, with system-dependent bandwidth and no dedicated VRAM. This means the GT 545 does not compete with the CPU for memory bandwidth, while the Intel part does. The GT 545's memory clock is 800 MHz, or 1600 Mbps effective. The Intel part's memory clock is listed as system shared, which is a fundamental architectural limitation for integrated graphics.

The GT 545 is a single-slot discrete card with no power connectors and a 70 W TDP, requiring a 250 W power supply. The Intel HD Graphics 630 is an IGP with a 15 W TDP, drawing power from the motherboard rather than a dedicated connector. The GT 545 connects via PCIe 2.0 x16, while the Intel part uses a ring bus interface. Display outputs differ as well: the GT 545 offers 1x DVI, 1x HDMI 1.3a, and 1x VGA, while the Intel part's outputs are motherboard dependent.

API support favors the Intel part. The HD Graphics 630 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The GT 545 supports DirectX 12 (11_0) and OpenGL 4.6, but has no Vulkan support. The GT 545 also lacks the FP16 capability that the Intel part has, with the HD Graphics 630 offering 768.0 GFLOPS FP16 at a 2:1 ratio versus its FP32 rate.

FAQ

Q: Which GPU is faster in OpenCL compute?

A: The NVIDIA GeForce GT 545 scores 3594 in Geekbench OpenCL, while the Intel HD Graphics 630 scores 3587. The GT 545 is 0.2% ahead, which is a negligible difference.

Q: Does the Intel HD Graphics 630 support Vulkan?

A: Yes, the Intel part supports Vulkan 1.3. The NVIDIA GeForce GT 545 has no Vulkan support listed in the database.

Q: How much dedicated memory does each GPU have?

A: The GT 545 has 1536 MB of DDR3 memory on a 192-bit bus with 38.40 GB/s bandwidth. The Intel HD Graphics 630 uses system shared memory with system-dependent bandwidth.

Q: What is the power consumption difference?

A: The Intel HD Graphics 630 has a 15 W TDP, while the GT 545 has a 70 W TDP and requires a 250 W power supply.

Q: Which GPU has more shading units?

A: The Intel HD Graphics 630 has 192 shading units, while the GT 545 has 144. However, the GT 545 has more ROPs, 16 versus 3.

Q: Are both GPUs still in production?

A: No, both are end-of-life. The Intel part was released in 2016, and the GT 545 was released in 2011.

Specification Differences

| Specification | Intel HD Graphics 630 | NVIDIA GeForce GT 545 |

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

| Architecture | Generation 9.5 | Fermi 2.0 |

| Process node | 14 nm++ | 40 nm |

| Transistors | Not listed | 1,170 million |

| Die size | Not listed | 238 mm² |

| Transistor density | Not listed | 4.9M / mm² |

| Shading units | 192 | 144 |

| ROPs | 3 | 16 |

| Pixel rate | 3.000 GPixel/s | 4.320 GPixel/s |

| Texture rate | 24.00 GTexel/s | 17.28 GTexel/s |

| FP32 | 384.0 GFLOPS | 414.7 GFLOPS |

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

| TDP | 15 W | 70 W |

| Memory size | System shared | 1536 MB |

| Memory type | System shared | DDR3 |

| Memory bus | System shared | 192 bit |

| Memory bandwidth | System dependent | 38.40 GB/s |

| Slot width | IGP | Single-slot |

| Power connectors | None | None |

| Suggested PSU | Not listed | 250 W |

| Bus interface | Ring Bus | PCIe 2.0 x16 |

| Display outputs | Motherboard dependent | 1x DVI, 1x HDMI 1.3a, 1x VGA |

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

| Vulkan support | 1.3 | Not listed |

| Launch MSRP | Not listed | 149 USD |

The Verdict

The data paints a clear picture for different use cases. In raw compute, the GT 545 holds a slight edge in OpenCL and FP32, but the margins are small. The GT 545's 414.7 GFLOPS FP32 is 7.4% higher than the Intel part's 384.0 GFLOPS, and its pixel rate of 4.320 GPixel/s is 44% higher. If you need rasterization throughput, the GT 545 is the stronger choice.

However, the Intel HD Graphics 630 has significant advantages in modern API support. It offers Vulkan 1.3, which the GT 545 completely lacks, and it supports DirectX 12 at the 12_1 feature level versus the GT 545's 11_0. The Intel part also has FP16 capability at 768.0 GFLOPS, which the GT 545 does not support at all. For workloads that leverage these modern APIs or half-precision math, the Intel part is clearly better.

The efficiency gap is massive. The Intel part draws 15 W versus 70 W for the GT 545. That is over 4.5 times the power draw for the NVIDIA card, with only a 0.2% OpenCL advantage. The GT 545 also requires a 250 W power supply, while the Intel part runs entirely off the motherboard.

The GT 545's dedicated 1536 MB memory with 38.40 GB/s bandwidth is a real advantage over system shared memory, particularly for texture-heavy workloads that exceed what the Intel part can hold in its shared pool. But this comes at the cost of a discrete card slot and additional power draw.

If the database's percentile rankings matter, the Intel part at the 24th percentile edges out the GT 545 at the 21st percentile. The Intel part's average benchmark score of 4075 is 13.4% higher than the GT 545's 3594, though this is inflated by the Metal score, which the NVIDIA card cannot run.

For a modern system with a Kaby Lake processor, the integrated HD Graphics 630 is the sensible choice. It offers competitive compute, better API support, and dramatically lower power consumption. The GT 545 is a product of an older era, and its only advantages are dedicated memory and higher pixel throughput.

Where Each One Wins

Intel HD Graphics 630 wins on:

• Modern API support: Vulkan 1.3 and DirectX 12 (12_1) versus no Vulkan and DirectX 12 (11_0) on the GT 545.
• FP16 compute: 768.0 GFLOPS versus no FP16 capability on the GT 545.
• Texture rate: 24.00 GTexel/s versus 17.28 GTexel/s, a 39% advantage.
• Shading units: 192 versus 144, a 33% advantage.
• Power efficiency: 15 W TDP versus 70 W, over 4.5 times lower.
• Process node: 14 nm++ versus 40 nm, which explains the efficiency gap.
• Average benchmark score: 4075 versus 3594, driven by Metal and Vulkan results.

NVIDIA GeForce GT 545 wins on:

• OpenCL compute: 3594 versus 3587, a 0.2% margin.
• FP32 compute: 414.7 GFLOPS versus 384.0 GFLOPS, a 7.4% margin.
• Pixel rate: 4.320 GPixel/s versus 3.000 GPixel/s, a 44% margin.
• ROPs: 16 versus 3, over five times more.
• Dedicated memory: 1536 MB DDR3 with 38.40 GB/s bandwidth versus system shared memory.
• Memory bus: 192 bit versus system shared.
• Form factor: single-slot discrete card versus IGP, allowing installation in systems without modern integrated graphics.

If your workload stresses pixel fill or relies on dedicated VRAM, the GT 545 has a genuine edge. If you need Vulkan, FP16, or efficient operation, the Intel HD Graphics 630 is the only sensible option. The GT 545's launch MSRP was 149 USD, but that number reflects a 2011 price point and does not account for its end-of-life status.