Intel HD Graphics 630 vs NVIDIA GeForce 920MX Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The Intel HD Graphics 630 records an average benchmark score of 4075, while the NVIDIA GeForce 920MX averages 3528. The Intel part sits at the 24th percentile of all GPUs, compared to the 21st percentile for the NVIDIA part.

Q: How do the two compare in OpenCL performance?

A: The NVIDIA GeForce 920MX wins the OpenCL test with a score of 4068, which is 11.8% ahead of the Intel HD Graphics 630's 3587.

Q: Which GPU wins in Vulkan?

A: The Intel HD Graphics 630 wins the Vulkan test with a score of 3540, beating the GeForce 920MX's 2987 by 18.5%.

Q: What are the nearest rivals to the Intel HD Graphics 630?

A: The AMD Radeon RX 9060 XT 8 GB scores 4093, 0.4% higher than the Intel part; the NVIDIA GeForce GT 755M scores 4033, 1% lower than Intel; the AMD FirePro M4150 scores 4013, 1.5% lower; and the NVIDIA Quadro K2100M scores 4151, 1.8% higher.

Q: What is the nearest rival to the GeForce 920MX?

A: The NVIDIA GeForce GT 545 scores 3594, 1.8% higher than the 920MX; the NVIDIA RTX 5000 Mobile Ada Generation scores 3596, 1.9% higher; the NVIDIA GeForce GT 735M scores 3616, 2.4% higher; and the NVIDIA Quadro 2000M scores 3434, 2.7% lower.

Q: What API levels does each GPU support?

A: The Intel HD Graphics 630 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The NVIDIA GeForce 920MX supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4.

Architecture Differences

The two GPUs come from completely different design philosophies. The Intel HD Graphics 630 is built on the Generation 9.5 architecture, specifically the Kaby Lake GT2 chip, and is manufactured on Intel's 14 nm++ process at Intel's own foundry. The NVIDIA GeForce 920MX uses the Maxwell architecture with the GM108S chip, fabricated on TSMC's 28 nm process. The process gap is significant: Intel uses a 14 nm++ node while NVIDIA relies on a 28 nm node, which explains part of the efficiency and feature differences.

The GeForce 920MX is a discrete part with a known physical footprint. The database lists its transistor count at 1,020 million, a die size of 77 mm², and a transistor density of 13.2M per mm². The Intel HD Graphics 630 has no listed transistor count or die size, as it is an integrated graphics processor (IGP) that shares the die with the CPU. The NVIDIA chip is packaged as an MXM module, while the Intel part is an IGP on a Ring Bus interface.

The shading resources differ. The Intel HD Graphics 630 has 192 shading units, 24 texture mapping units, and 3 ROPs. The GeForce 920MX has 256 shading units, 24 TMUs, and 8 ROPs. The NVIDIA part has more shading units and more than double the ROP count, but the same number of TMUs. The Intel part has no listed transistors, while the NVIDIA part's 1,020 million transistors are fully documented.

Memory architecture is a major divergence. The Intel HD Graphics 630 uses system shared memory, with the bus width listed as System Shared and bandwidth as System Dependent. The GeForce 920MX has a dedicated 2 GB of DDR3 memory on a 64-bit bus, delivering 14.40 GB/s of bandwidth. The NVIDIA chip runs its memory at 900 MHz, 1800 Mbps effective. The clock speeds also differ: Intel runs at a 350 MHz base and 1000 MHz boost, while NVIDIA runs at 965 MHz base and 993 MHz boost. The NVIDIA part has a much higher base clock, but its boost is only slightly below the Intel boost.

The API feature sets are close but not identical. Both support OpenGL 4.6. The Intel part supports DirectX 12 (12_1) and Vulkan 1.3, while the NVIDIA part supports DirectX 12 (11_0) and Vulkan 1.4. The NVIDIA part has a higher Vulkan version, but the Intel part has a higher DirectX feature level.

The production status for both is end-of-life. The GeForce 920MX has a listed predecessor (GeForce 800M) and successor (GeForce 10 Mobile), while the Intel part has neither listed. The NVIDIA part has a release date of 2016-03-24, and the Intel part 2016-08-29.

Where Each One Wins

The benchmark results show a clean split between the two APIs. The NVIDIA GeForce 920MX is the stronger OpenCL performer. Its score of 4068 beats the Intel HD Graphics 630's 3587 by 11.8%. This suggests the discrete NVIDIA part maintains a lead in compute workloads that use OpenCL, likely due to its higher shading unit count and dedicated memory. The 920MX's 256 shading units and 8 ROPs give it a structural advantage in raw parallel throughput.

The Intel HD Graphics 630 wins decisively in Vulkan. Its score of 3540 beats the GeForce 920MX's 2987 by 18.5%. This is a notable reversal, given that the Intel part has fewer shading units and relies on system shared memory. The Intel part's Vulkan 1.3 driver implementation appears to extract more from the hardware in this API, or the Maxwell architecture's Vulkan path is less efficient. The delta is large enough that it likely reflects a real driver or hardware-level advantage rather than noise.

The average benchmark score also favors the Intel part. Intel averages 4075 across all recorded tests, while the NVIDIA part averages 3528. The Intel part records three benchmark scores (Geekbench Metal at 5099, OpenCL at 3587, and Vulkan at 3540), while the NVIDIA part only records two (OpenCL at 4068 and Vulkan at 2987). The Intel part's Metal score of 5099 is the highest single score among both GPUs, and it contributes to the higher average. The NVIDIA part has no Metal score listed.

In percentile terms, the Intel part sits at the 24th percentile of all GPUs, above the 920MX's 21st percentile. The Intel part's nearest rivals, including the AMD Radeon RX 9060 XT 8 GB at 4093 and the NVIDIA Quadro K2100M at 4151, cluster closely around its average score. The NVIDIA part's nearest rivals, including the GeForce GT 545 at 3594 and the GeForce GT 735M at 3616, also cluster tightly around its average. Neither GPU is far from its nearest competitors.

The use-case split is therefore straightforward: for OpenCL-heavy workloads, the GeForce 920MX is the better choice; for Vulkan-based workloads, the Intel HD Graphics 630 has the edge. The Intel part also has the advantage of being integrated, which means it consumes no additional slot width, while the NVIDIA part requires an MXM module.

Specification Differences

The two GPUs differ in nearly every specification category. The process node is a clear distinction: the Intel HD Graphics 630 is on 14 nm++, while the GeForce 920MX is on 28 nm. The foundries also differ, with Intel using its own foundry and NVIDIA using TSMC.

The GeForce 920MX has fully listed transistor data: 1,020 million transistors, a 77 mm² die size, and a density of 13.2M per mm². The Intel HD Graphics 630 has no listed transistor count, die size, or density.

Clock speeds differ. The Intel part has a 350 MHz base clock and 1000 MHz boost. The NVIDIA part has a 965 MHz base clock and 993 MHz boost. The NVIDIA part's base is much higher, but the boost is nearly identical.

Memory configuration is a major split. The Intel part uses System Shared memory with System Shared type, bus width, and System Dependent bandwidth. The NVIDIA part has 2 GB of DDR3, a 64-bit bus, and 14.40 GB/s bandwidth, with memory clocked at 900 MHz (1800 Mbps effective).

The compute units differ. The Intel part has 192 shading units, 24 TMUs, and 3 ROPs. The NVIDIA part has 256 shading units, 24 TMUs, and 8 ROPs. Pixel rate and texture rate follow: the Intel part has 3.000 GPixel/s and 24.00 GTexel/s, while the NVIDIA part has 7.944 GPixel/s and 23.83 GTexel/s. The NVIDIA part has a much higher pixel rate, while the Intel part has a slightly higher texture rate.

FP32 performance also differs. The Intel part delivers 384.0 GFLOPS, while the NVIDIA part delivers 508.4 GFLOPS. The Intel part also lists FP16 performance at 768.0 GFLOPS (2:1), while the NVIDIA part has no FP16 listing.

Power and physical configuration: the Intel part has a 15 W TDP and is an IGP, while the NVIDIA part has a 16 W TDP and is an MXM module. The NVIDIA part lists power connectors as None, while the Intel part has none listed. The bus interface differs: the Intel part uses Ring Bus, the NVIDIA part uses PCIe 3.0 x8.

Display outputs: the Intel part is Motherboard Dependent, the NVIDIA part is Portable Device Dependent. API support: the Intel part has DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3; the NVIDIA part has DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4.

The Intel part has no listed launch MSRP, and the NVIDIA part also has no listed launch MSRP. Both are end-of-life. The NVIDIA part has a predecessor and successor listed, the Intel part does not. The NVIDIA part was released earlier, on 2016-03-24, while the Intel part followed on 2016-08-29.

Head-to-Head Benchmarks

The head-to-head data contains two recorded tests, and each GPU wins one. The first test is Geekbench OpenCL. The NVIDIA GeForce 920MX scores 4068, and the Intel HD Graphics 630 scores 3587. The NVIDIA part wins by 11.8%. This is the larger absolute margin of the two tests in raw points: 481 points separate them in OpenCL. The NVIDIA part's advantage in shading units (256 vs 192) and its dedicated 2 GB of DDR3 memory likely contribute to this result. The 920MX also has a higher FP32 rating at 508.4 GFLOPS versus 384.0 GFLOPS, which aligns with the OpenCL outcome.

The second test is Geekbench Vulkan. The Intel HD Graphics 630 scores 3540, and the NVIDIA GeForce 920MX scores 2987. The Intel part wins by 18.5%. This is the larger percentage margin of the two tests, though the raw point gap is 553, which is also the larger raw gap. The Intel part wins despite having fewer shading units and relying on system shared memory. Its Vulkan 1.3 driver appears to be more mature or better optimized than the NVIDIA part's Vulkan path. The NVIDIA part supports Vulkan 1.4, which is a higher API version, but that does not translate into a higher score.

The Intel HD Graphics 630 also has a third benchmark in its file: Geekbench Metal at 5099. This test has no head-to-head counterpart, as the GeForce 920MX has no Metal score listed. The Intel part's Metal result is its highest score and is well above both its OpenCL and Vulkan scores. This contributes to the Intel part's average of 4075, which is 547 points higher than the NVIDIA part's 3528 average.

The win count in the head-to-head section is one win for each GPU. The Intel part's win in Vulkan is larger in percentage terms (18.5%) than the NVIDIA part's win in OpenCL (11.8%). The NVIDIA part's win is larger in absolute terms in OpenCL when compared to the Intel part's OpenCL score, but the Intel part's Vulkan win has the larger raw point margin overall.

The average benchmark scores reflect the broader picture. The Intel HD Graphics 630's average of 4075 places it just 0.4% behind the AMD Radeon RX 9060 XT 8 GB (4093) and 1.8% behind the NVIDIA Quadro K2100M (4151). The GeForce 920MX's average of 3528 places it 2.7% ahead of the NVIDIA Quadro 2000M (3434) but 2.4% behind the NVIDIA GeForce GT 735M (3616). The Intel part's nearest rivals are generally higher-scoring than the NVIDIA part's nearest rivals, which is consistent with the Intel part's higher percentile ranking.