Intel HD Graphics 630 vs NVIDIA GeForce GT 740 Comparison
Intel HD Graphics 630
GeForce GT 740
PERFORMANCE BENCHMARKS
Analysis: Intel HD Graphics 630 vs NVIDIA GeForce GT 740
# Head-to-Head Benchmarks
The recorded benchmark data places the Intel HD Graphics 630 and the NVIDIA GeForce GT 740 in a close contest, but the wins are split decisively along test type. In the Geekbench Metal test, the Intel part finishes with a score of 5099 against 2363 for the NVIDIA card, a 115.8% advantage. That is the single largest margin in the entire comparison, and it is not a small one: the Intel integrated graphics solution is nearly 2.2 times faster than the discrete GeForce part in this test. The other two recorded tests, however, go the other way. In Geekbench OpenCL, the NVIDIA card scores 3847 versus 3587 for the Intel, a 6.8% lead. The Vulkan test shows an even larger NVIDIA advantage: 4083 versus 3540, which is 13.3% ahead. So the head-to-head tally is 1 win for Intel and 2 wins for NVIDIA, but the margin in Intel's favor is much larger than either single NVIDIA win.
This pattern suggests a workload-dependent relationship. The GeForce GT 740 is a small discrete solution, while the HD 630 is an integrated processor. The data shows the NVIDIA card performs better in compute-heavy, driver-managed APIs (OpenCL, Vulkan), while the Intel part excels in the Metal API, which tends to run closer to the GPU's native path. The average benchmark score for the Intel part is 4075, while the NVIDIA part averages 3431, a difference of about 15.8% in favor of Intel. However, the percentile ranking against all GPUs in the database tells a similar story from a different angle: the HD 630 sits at the 24th percentile, while the GT 740 sits at the 21st percentile. Both are low-end entries, but the Intel part is slightly higher among all recorded GPUs.
Architecture Differences
The two parts diverge in almost every core architectural parameter. The Intel HD Graphics 630 is built on the Generation 9.5 architecture, using Intel's 14 nm++ process node, with Kaby Lake GT2 silicon. It packs 24 texture mapping units and 3 ROPs, with a shading unit count of 192. The GPU has 24 TMUs. The pixel rate is listed at 3.000 GPixel/s, and texture rate is 24.00 GTexel/s. FP32 performance is 384.0 GFLOPS, while FP16 is 768.0 GFLOPS (at a 2:1 ratio). The power is rated at 15 W TDP. The bus interface is a ring bus. Display outputs are motherboard-dependent. The production status is end-of-life, with a release date of 2016-08-29.
The NVIDIA GeForce GT 740 is a fundamentally different design. It is a Kepler architecture chip, GK107, fabricated at TSMC on a 28 nm process. The die size is 118 mm² with 1,270 million transistors, giving a transistor density of 10.8M / mm². The memory configuration is 1024 MB of GDDR5 on a 128-bit bus, with an effective bandwidth of 80.19 GB/s. The shading units count is 384, and there are 32 TMUs and 16 ROPs. Pixel throughput is 7.944 GPixel/s and texture rate is 31.78 GTexel/s. FP32 is 762.6 GFLOPS. The board consumes 64 W and requires a 250 W PSU. The interface is PCIe 3.0 x16. Display outputs include 2x DVI and 1x mini-HDMI 1.4a. It has 2 GB of DDR3 memory.
The most significant differences are therefore process (14nm++ vs 28nm), memory (system shared vs dedicated 1GB GDDR5), and power (15W vs 64W). The GT 740 is a full slot-width card with a 6-pin power connector, while the HD 630 is an integrated processor with no dedicated memory of its own.
FAQ
Q: Which is faster, the Intel HD Graphics 630 or the NVIDIA GeForce GT 740?
The answer depends on the workload. In the Metal API test, the Intel part is 115.8% faster. In OpenCL and Vulkan, the NVIDIA part is faster by 6.8% and 13.3%, respectively. Overall, the average benchmark score favors the Intel: 4075 versus 3431.
Q: Does the GeForce GT 740 support modern APIs?
Yes. The data lists DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175 as supported APIs.
Q: What is the memory configuration of the GeForce GT 740?
It has 64 MB of GDDR5 on a 128-bit bus, with an effective memory clock of 1253 MHz and bandwidth of 80.19 GB/s.
Q: Does the HD 630 have dedicated memory?
No. The memory size is system-shared, just like many integrated GPUs. The total memory is "System Shared", and it is not a dedicated memory.
Q: Which GPU is more power efficient?
The 630 has a 15 W TDP, while the GT 740 is 64 W. The Intel part is significantly more power efficient under load.
The Verdict
From the recorded data alone, the Intel HD Graphics 630 is the preferred option for anyone running compute-heavy workloads. It takes the Metal benchmark by a very large margin, 115.8% over the GeForce GT 740. It also has a higher average benchmark score (4075 vs 3431), and a better percentile ranking among all GPUs (24th vs 21st). The Intel part is also far more power efficient, with a 15 W TDP versus 64 W for the NVIDIA part. However, the GT 740 wins in OpenCL and Vulkan workloads, by 6.8% and 13.3%, respectively. If the application is built around OpenCL or Vulkan compute, the NVIDIA card is the better choice. For general, driver-optimized or integrated workloads, the HD 630 is the better part. The data does not support the 740 for high-end gaming: only two of the three head-to-head tests favor it, and the margins are modest at best.
Specification Differences
| Field | Intel HD Graphics 630 | NVIDIA GeForce GT 740 |
|---|---|---|
| Chip | Kaby Lake GT2 | GK107 |
| Architecture | Generation 9.5 | Kepler |
| Process | 14 nm++ | 28 nm |
| Foundry | Intel | TSMC |
| Transistors | Not listed | 1,270 million |
| Die Size | Not listed | 118 mm² |
| Transistor Density | Not listed | 10.8M / mm² |
| Base Clock | 350 MHz | Not listed |
| Boost Clock | 1000 MHz | Not listed |
| Memory | System Shared | 1024 MB GDDR5 |
| Memory Bus | System Shared | 128 bit |
| Memory Bandwidth | System Dependent | 80.19 GB/s |
| Shading Units | 192 | 384 |
| TMUs | 24 | 32 |
| ROPs | 3 | 16 |
| Pixel Rate | 3.000 GPixel/s | 7.944 GPixel/s |
| Texture Rate | 24.00 GTexel/s | 31.97 GTex/s |
| FP32 | 384.0 GFLOPS | 762.6 GFLOPS |
| TDP | 15 W | 64 W |
| Slot Width | IGP | Single-slot |
| Power Connectors | Not needed | 1x 6-pin |
| Suggested PSU | Not listed | 250 W |
| Bus Interface | Ring Bus | PCIe 3.0 x16 |
| Display Outputs | Motherboard Dependent | 2x DVI, 1x mini-HDMI 1.4a |