GPU Comparison

The NVIDIA GeForce GT 645M and Intel HD Graphics 630 are both end-of-life integrated or mobile-class graphics solutions, but they represent very different design philosophies and performance profiles. The GT 645M, a discrete-oriented chip from the Kepler era, and the HD 630, Intel's mainstream integrated GPU from the Kaby Lake generation, trade blows depending on the workload. Benchmark data shows the GT 645M holds a slight edge in average score, 4411 versus 4075, which places it at the 26th percentile versus the 24th percentile for the Intel part. However, the individual benchmark suite reveals a more nuanced story, with each GPU dominating in different scenarios.

Head-to-Head Benchmarks

The most decisive victory for the NVIDIA GeForce GT 645M comes in the Geekbench Vulkan test. Here, the GT 645M scores 4875, while the Intel HD Graphics 630 manages only 3540. This represents a massive 37.7% lead for the NVIDIA part, making it the single largest performance gap in the entire comparison. This result suggests that for applications leveraging the Vulkan API, the GT 645M's dedicated hardware and driver maturity provide a substantial advantage.

The GT 645M also takes the win in the Geekbench Metal benchmark, though by a smaller margin. It scores 5679 against the HD 630's 5099, a difference of 11.4%. This is notable because it shows the older NVIDIA GPU outperforming the newer Intel solution in Apple's graphics API, likely due to the GT 645M's higher raw shading power and dedicated memory architecture.

The Intel HD Graphics 630, however, secures a significant victory in the Geekbench OpenCL test. It scores 3587, which is 25.3% higher than the GT 645M's 2680. This is a substantial reversal of fortune. The HD 630's strong OpenCL showing indicates that for compute-oriented tasks using this API, the Intel iGPU is clearly the superior choice, possibly benefiting from its newer architecture and higher boost clock in compute-heavy workloads.

When comparing to their respective nearest rivals, the picture sharpens. The GT 645M's average score of 4411 puts it a mere 0.5% ahead of the NVIDIA GeForce 930M (4388) and 1.2% ahead of the Intel Iris Pro Graphics 5200 (4360). It also leads the NVIDIA GeForce RTX 4070 GDDR6 (4335) by 1.8%, though that comparison is unusual given the modern GPU's different market position. Conversely, it trails the AMD Radeon R7 M260 (4499) by 1.9%. The HD 630's average of 4075 places it 1% ahead of the NVIDIA GeForce GT 755M (4033) and 1.5% ahead of the AMD FirePro M4150 (4013), while sitting 0.4% behind the AMD Radeon RX 9060 XT 8 GB (4093) and 1.8% behind the NVIDIA Quadro K2100M (4151). These deltas are all small, underscoring that both GPUs are clustered tightly with their direct contemporaries.

Where Each One Wins

The NVIDIA GeForce GT 645M wins in two of the three head-to-head benchmark categories, making it the better choice for graphics API-centric workloads. Its 37.7% Vulkan advantage is a clear indicator that it will handle modern cross-platform game engines and Vulkan-based applications with significantly more ease than the Intel HD 630. The 11.4% Metal win also suggests it is a more capable GPU for macOS-based graphics tasks, despite its older architecture.

The Intel HD Graphics 630 wins decisively in OpenCL compute performance, with its 25.3% lead. This makes it the preferred option for tasks that are heavily parallel and utilize the OpenCL framework for general-purpose GPU computing. This could include video encoding, image filtering, or physics simulations that are specifically optimized for OpenCL. The data shows a clear split: the GT 645M is the better pure graphics performer, while the HD 630 holds a distinct edge in compute throughput as measured by OpenCL.

Architecture Differences

The two GPUs are built on fundamentally different architectures from different manufacturing eras. The NVIDIA GeForce GT 645M is based on the GK107 chip using the Kepler architecture, fabricated on a 28 nm process at TSMC. This chip contains 1,270 million transistors on a die size of 118 mm², resulting in a transistor density of 10.8M per mm². In contrast, the Intel HD Graphics 630 is built on the Kaby Lake GT2 chip using Intel's Generation 9.5 architecture, manufactured on a 14 nm++ process at Intel's own fabs.

The most striking architectural difference is in the compute core configuration. The GT 645M boasts 384 shading units, 32 texture mapping units (TMUs), and 16 raster operations pipelines (ROPs). The HD 630 has significantly fewer: 192 shading units, 24 TMUs, and only 3 ROPs. This disparity in core counts directly explains the GT 645M's higher theoretical fill rates, with a pixel rate of 6.240 GPixel/s and texture rate of 24.96 GTexel/s, compared to the HD 630's 3.000 GPixel/s and 24.00 GTexel/s. The ROP count difference is particularly stark, giving the NVIDIA part a massive advantage in pixel throughput.

Clock speeds tell a different story. The GT 645M has a base clock of 709 MHz and a boost clock of 780 MHz, while the HD 630 has a much lower base clock of 350 MHz but a significantly higher boost clock of 1000 MHz. This higher boost clock helps the Intel part close the gap in some workloads, but it cannot fully compensate for the NVIDIA GPU's superior core and ROP count. The GT 645M also has a dedicated 2 GB of DDR3 memory on a 128-bit bus, providing 28.80 GB/s of bandwidth, whereas the HD 630 relies entirely on system shared memory, with bandwidth rated as "System Dependent."

Specification Differences

The specification sheets highlight several key differences beyond the core architecture. The NVIDIA GeForce GT 645M offers a maximum FP32 performance of 599.0 GFLOPS, while the Intel HD Graphics 630 delivers 384.0 GFLOPS. However, the Intel part supports FP16 compute at 768.0 GFLOPS (2:1), a capability the GT 645M does not list. The GT 645M supports up to DirectX 12 (11_0), while the HD 630 supports DirectX 12 (12_1). Both support OpenGL 4.6, but the HD 630 supports Vulkan 1.3 compared to the GT 645M's Vulkan 1.2.175.

Power consumption and interface also diverge. The GT 645M has a TDP of 32 W, while the HD 630 is rated at just 15 W. The GT 645M uses a PCIe 3.0 x16 bus interface, whereas the HD 630 connects via a Ring Bus. The GT 645M's display outputs are "Portable Device Dependent," indicating it is typically soldered into laptops, while the HD 630's are "Motherboard Dependent," reflecting its integrated nature. The GT 645M has no power connectors, and the HD 630's power connector field is null, both indicating low-power designs. Their production status is identical: both are end-of-life.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GT 645M has a higher average benchmark score of 4411, compared to the Intel HD Graphics 630's 4075.

Q: In which benchmark does the Intel HD Graphics 630 beat the NVIDIA GeForce GT 645M?

A: The Intel HD Graphics 630 wins the Geekbench OpenCL test, scoring 3587 against the GT 645M's 2680, a 25.3% advantage.

Q: What is the largest performance gap in the head-to-head benchmarks?

A: The largest gap is in the Geekbench Vulkan test, where the NVIDIA GeForce GT 645M leads by 37.7%, scoring 4875 versus the HD 630's 3540.

Q: How do the shading unit counts compare?

A: The NVIDIA GeForce GT 645M has 384 shading units, while the Intel HD Graphics 630 has 192, exactly half as many.

Q: What is the difference in their process nodes?

A: The NVIDIA GeForce GT 645M is built on a 28 nm process, while the Intel HD Graphics 630 uses a more advanced 14 nm++ process.

Q: What is the TDP difference between the two?

A: The NVIDIA GeForce GT 645M has a TDP of 32 W, whereas the Intel HD Graphics 630 is more power-efficient with a TDP of 15 W.

The Verdict

Based strictly on the benchmark data, the NVIDIA GeForce GT 645M is the more capable graphics processor overall. It wins two of the three head-to-head tests, including a dominant 37.7% lead in Vulkan and a solid 11.4% lead in Metal. Its average benchmark score is higher, and it offers superior pixel and texture fill rates, along with dedicated video memory, which prevents it from contending with system RAM for bandwidth. The GT 645M is the pick for users who prioritize gaming or graphics-API-specific performance in legacy applications.

The Intel HD Graphics 630 is the better choice for compute-heavy workloads that rely on OpenCL, where its 25.3% performance advantage is substantial. Its lower 15 W TDP also makes it the more power-efficient solution, which is a critical consideration for thin-and-light laptops. Users who need OpenCL compute capability and can accept lower Vulkan and Metal performance should favor the Intel HD 630.

The data does not support a universal winner. The GT 645M offers better traditional graphics performance and API support for Vulkan and Metal, while the HD 630 counters with superior OpenCL compute and lower power draw. The choice hinges entirely on the intended workload. If the priority is gaming or graphics rendering, the GT 645M's higher scores and dedicated memory make it the clear pick. If the priority is compute tasks within the OpenCL framework or maximum battery efficiency, the HD 630 is the data-backed choice.