Intel UHD Graphics P630 vs NVIDIA GeForce GTX 580M Comparison

Where Each One Wins

The recorded benchmark data splits cleanly between these two parts. The NVIDIA GeForce GTX 580M wins the only shared test, Geekbench OpenCL, with a score of 6389 against the Intel UHD Graphics P630’s 5111. That is a 25 percent advantage for the NVIDIA part. The Intel adapter has no head-to-head victory in the database, but it does have a second recorded result, a Geekbench Vulkan score of 5628, which the NVIDIA part cannot contest because no Vulkan measurement exists for it.

That Vulkan result is worth noting. Even though the Intel part loses the OpenCL comparison, its Vulkan score of 5628 is higher than its own OpenCL score of 5111, suggesting that under a different API the Intel graphics can shift its performance profile. For applications that favor Vulkan, the Intel UHD Graphics P630 may behave differently than the OpenCL numbers alone imply. Still, in terms of direct comparison, the NVIDIA GeForce GTX 580M is the clear winner in compute workloads measured through OpenCL.

The use-case split is therefore straightforward. For OpenCL-based compute tasks, the NVIDIA GeForce GTX 580M holds a decisive edge. For Vulkan-capable workloads, the Intel part has a recorded result and the NVIDIA part does not, so the Intel adapter is the only one of the two with documented Vulkan performance in the database. The NVIDIA part also lacks any Vulkan API listing in its specifications, while the Intel part lists Vulkan 1.3 support. That means the Intel UHD Graphics P630 is the only option here for users who specifically need Vulkan compatibility, based on the recorded data.

Architecture Differences

The two parts come from different manufacturers, different process nodes, and different design philosophies. The NVIDIA GeForce GTX 580M uses the GF114 chip, built on the Fermi 2.0 architecture, and belongs to the GeForce 500M generation. It is fabricated by TSMC on a 40 nm process. The chip contains 1,950 million transistors on a 332 mm² die, giving a transistor density of 5.9 million transistors per square millimeter. The Intel UHD Graphics P630 uses the Comet Lake GT2 chip, based on the Generation 9.5 architecture, and belongs to the HD Graphics-W (Comet Lake) generation. It is fabricated by Intel on a 14 nm+++ process. The Intel chip has no recorded transistor count, die size, or density in the database.

The compute resources differ substantially. The NVIDIA part has 384 shading units, 64 texture mapping units, and 32 raster output pipelines. The Intel part has 192 shading units, 24 texture mapping units, and only 3 raster output pipelines. That is half the shading units, 40 fewer texture units, and 29 fewer ROPs. The pixel rate reflects this: the NVIDIA part delivers 9.920 GPixel/s, while the Intel part delivers 3.600 GPixel/s. Texture rate also favors NVIDIA at 39.68 GTexel/s versus 28.80 GTexel/s. Floating point performance follows the same pattern. The NVIDIA part reaches 952.3 GFLOPS in FP32, while the Intel part reaches 460.8 GFLOPS. The Intel part does have a recorded FP16 figure of 921.6 GFLOPS (2:1), while the NVIDIA part has no FP16 listing.

Memory architecture is fundamentally different. The NVIDIA GeForce GTX 580M uses 2 GB of dedicated GDDR5 memory on a 256 bit bus, with 96.00 GB/s of bandwidth and a memory clock of 750 MHz (3 Gbps effective). The Intel UHD Graphics P630 uses system shared memory, with a system dependent bandwidth and a system shared bus width. The Intel part’s base clock is 350 MHz with a boost clock of 1200 MHz, while the NVIDIA part has no recorded base or boost clock, only its memory clock.

Feature support also differs. The NVIDIA part supports DirectX 12 (11_0) and OpenGL 4.6, but has no Vulkan listing. The Intel part supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The NVIDIA part uses an MXM-B (3.0) bus interface and is a 100 W MXM module. The Intel part uses a Ring Bus interface and is a 15 W IGP. The NVIDIA part has no power connectors, while the Intel part has no recorded power connector data. Display outputs are portable device dependent for NVIDIA and motherboard dependent for Intel.

FAQ

Q: Which GPU has the higher OpenCL score?

A: The NVIDIA GeForce GTX 580M scores 6389 in Geekbench OpenCL, which is 25 percent higher than the Intel UHD Graphics P630’s 5111 in the same test.

Q: Does the Intel UHD Graphics P630 support Vulkan?

A: Yes. The Intel part lists Vulkan 1.3 in its API specifications and has a recorded Geekbench Vulkan score of 5628. The NVIDIA GeForce GTX 580M has no Vulkan API listing and no Vulkan benchmark result in the database.

Q: How do the shading unit counts compare?

A: The NVIDIA GeForce GTX 580M has 384 shading units, while the Intel UHD Graphics P630 has 192 shading units. The NVIDIA part also has 64 texture mapping units versus 24, and 32 ROPs versus 3.

Q: What memory configurations do these GPUs use?

A: The NVIDIA GeForce GTX 580M uses 2 GB of GDDR5 memory on a 256 bit bus with 96.00 GB/s bandwidth. The Intel UHD Graphics P630 uses system shared memory with system dependent bandwidth.

Q: Which GPU has the higher FP32 performance?

A: The NVIDIA GeForce GTX 580M delivers 952.3 GFLOPS in FP32, which is more than double the Intel UHD Graphics P630’s 460.8 GFLOPS.

Q: What is the process node difference between the two?

A: The NVIDIA GeForce GTX 580M is built on a 40 nm TSMC process, while the Intel UHD Graphics P630 is built on Intel’s 14 nm+++ process.

Specification Differences

The two parts differ across nearly every recorded specification field. The NVIDIA GeForce GTX 580M uses the GF114 chip with Fermi 2.0 architecture, while the Intel UHD Graphics P630 uses the Comet Lake GT2 chip with Generation 9.5 architecture. The NVIDIA part is on a 40 nm process from TSMC with 1,950 million transistors and a 332 mm² die. The Intel part is on a 14 nm+++ process from Intel with no recorded transistor count or die size.

Clock behavior differs. The NVIDIA part has no recorded base or boost clock, but its memory runs at 750 MHz (3 Gbps effective). The Intel part has a 350 MHz base clock and a 1200 MHz boost clock, with system shared memory.

Memory specifications are entirely different. The NVIDIA part has 2 GB of GDDR5, a 256 bit bus, and 96.00 GB/s bandwidth. The Intel part has system shared memory, a system shared bus width, and system dependent bandwidth.

Compute unit counts differ sharply. The NVIDIA part has 384 shading units, 64 TMUs, and 32 ROPs. The Intel part has 192 shading units, 24 TMUs, and 3 ROPs. Pixel rate is 9.920 GPixel/s for NVIDIA versus 3.600 GPixel/s for Intel. Texture rate is 39.68 GTexel/s versus 28.80 GTexel/s. FP32 is 952.3 GFLOPS versus 460.8 GFLOPS. The Intel part has an FP16 figure of 921.6 GFLOPS (2:1), while the NVIDIA part has none.

Power and physical form differ. The NVIDIA part is a 100 W MXM module with no power connectors and an MXM-B (3.0) bus interface. The Intel part is a 15 W IGP with a Ring Bus interface and no recorded power connector data. Display outputs are portable device dependent for NVIDIA and motherboard dependent for Intel.

API support differs in DirectX and Vulkan. The NVIDIA part supports DirectX 12 (11_0) and OpenGL 4.6. The Intel part supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3.

Release timing differs. The NVIDIA GeForce GTX 580M was released on 2011-06-27, while the Intel UHD Graphics P630 was released on 2020-05-12. The NVIDIA part has a predecessor (GeForce 400M) and a successor (GeForce 600M), while the Intel part has neither listed. Both are end-of-life products. Neither has a recorded launch MSRP.

Head-to-Head Benchmarks

The database contains one direct head-to-head benchmark between these two parts: Geekbench OpenCL. In that test, the NVIDIA GeForce GTX 580M scores 6389, while the Intel UHD Graphics P630 scores 5111. The NVIDIA part wins by 25 percent. That is a substantial margin in a compute workload, and it aligns with the underlying hardware differences. The NVIDIA part has double the shading units, more than double the FP32 throughput, and a dedicated 96.00 GB/s memory interface, all of which contribute to its OpenCL advantage.

The Intel part’s only other recorded benchmark is Geekbench Vulkan, where it scores 5628. There is no NVIDIA Vulkan result to compare, so the Intel part stands alone in that API. Notably, the Intel part’s Vulkan score of 5628 exceeds its own OpenCL score of 5111 by roughly 10 percent, which suggests that the Intel architecture performs relatively better under Vulkan than under OpenCL. However, that does not change the head-to-head outcome, because the NVIDIA part has no Vulkan measurement.

The nearest rivals in the database provide context for each part’s standing. The NVIDIA GeForce GTX 580M’s average benchmark score is 6389, which places it in the 37th percentile of all GPUs. Its closest rival is the NVIDIA GeForce GTX 460 SE with an identical average score of 6389 and a delta of 0 percent. The NVIDIA RTX PRO 5000 72 GB Blackwell scores 6407, 0.3 percent higher. The AMD Radeon Pro WX 4100 scores 6330, 0.9 percent lower, and the AMD Radeon R7 M350 scores 6327, 1 percent lower. The GTX 580M therefore sits in a tight cluster where the four nearest rivals are all within roughly 1 percent of its score.

The Intel UHD Graphics P630 has an average benchmark score of 5370, placing it in the 31st percentile of all GPUs. Its nearest rival is the AMD Radeon R7 M445 with an average score of 5358, which is 0.2 percent lower. The AMD Radeon R7 M365X scores 5416, 0.8 percent higher. The NVIDIA GeForce 840M scores 5322, 0.9 percent lower, and the NVIDIA GeForce 930A scores 5317, 1 percent lower. The Intel part’s closest rivals are also tightly grouped, all within 1 percent of its average score.

The overall picture is clear. In the only shared benchmark, the NVIDIA GeForce GTX 580M beats the Intel UHD Graphics P630 by 25 percent in OpenCL. The NVIDIA part also holds advantages in pixel rate, texture rate, FP32 throughput, and dedicated memory bandwidth. The Intel part counters with Vulkan support and a higher recorded Vulkan score than its own OpenCL score, plus a much lower 15 W power draw compared to the NVIDIA part’s 100 W. For compute-heavy OpenCL tasks, the NVIDIA part is the stronger choice based on the recorded data. For Vulkan-based workloads, the Intel part is the only one of the two with documented performance.