Intel Iris Pro Graphics P580 vs NVIDIA GeForce GTX 670M Comparison

The Intel Iris Pro Graphics P580 and the NVIDIA GeForce GTX 670M represent two very different approaches to mobile graphics, separated by several years and distinct architectural philosophies. The data available shows a clear overall winner in raw compute benchmarks, but the story is more nuanced when considering the specific roles each part was designed to fill. The Iris Pro P580 is an integrated solution from Intel, built on a 14 nm+ process, while the GTX 670M is a discrete NVIDIA part from the Fermi 2.0 era, fabricated on a 40 nm process at TSMC. This head-to-head comparison relies on a single benchmark entry, but the underlying specifications provide substantial context for interpreting those results.

Head-to-Head Benchmarks

The only direct benchmark comparison available is in Geekbench OpenCL, and the results are decisive. The Intel Iris Pro Graphics P580 scores 9,082 points, while the NVIDIA GeForce GTX 670M scores 6,513 points. This gives the Intel part a 39.4% advantage in this specific test. The delta is substantial, indicating that the P580 delivers significantly higher raw compute throughput in this OpenCL workload. For context, the P580’s average benchmark score across all tests is 7,170, while the GTX 670M’s average is 6,513, meaning the OpenCL result is not an outlier for the Intel part but rather a reflection of its compute-oriented design.

The GTX 670M’s score places it in the 38th percentile of all GPUs, while the P580 sits slightly higher at the 39th percentile. These percentile values are close, but the actual benchmark delta is what matters most. The P580 wins the only head-to-head test, securing one win out of one possible comparison. The GTX 670M has no wins in this dataset. It is notably the P580 also has a Geekbench Vulkan score of 5,258, but no comparable Vulkan result exists for the GTX 670M, as the NVIDIA part lacks Vulkan support entirely. This absence of a Vulkan score for the GTX 670M is itself a significant data point, as it highlights a feature gap that goes beyond raw performance.

The 39.4% lead for the P580 in OpenCL is not a marginal victory. It suggests that the Intel architecture, despite being an integrated part with a 15 W TDP, is capable of outperforming a discrete mobile GPU with a 75 W TDP in compute-heavy tasks. The GTX 670M’s architecture, Fermi 2.0, was designed for a different era, and its 336 shading units and 803.7 GFLOPS FP32 throughput are simply lower than the P580’s 576 shading units and 1,152.0 GFLOPS FP32. These specification differences directly translate into the benchmark outcome.

Where Each One Wins

The data shows a single benchmark category, so the win distribution is straightforward: the Intel Iris Pro Graphics P580 wins the Geekbench OpenCL test, and the NVIDIA GeForce GTX 670M does not win any recorded benchmark. However, the broader specification sheet allows for a functional split. The P580’s compute advantage is clear, with its FP32 performance of 1,152.0 GFLOPS exceeding the GTX 670M’s 803.7 GFLOPS. The P580 also offers FP16 support at 2.304 TFLOPS (2:1), a feature entirely absent from the GTX 670M’s specifications, which list no FP16 capability. This makes the P580 better suited for workloads that leverage mixed-precision or half-precision arithmetic, such as certain machine learning inference or image processing tasks.

The GTX 670M, on the other hand, has its own strengths that are not captured in the benchmark data. It features 1,536 MB of dedicated GDDR5 memory on a 192-bit bus, providing 72.00 GB/s of bandwidth. The P580 uses system shared memory, with bandwidth listed as system dependent. In scenarios where dedicated memory bandwidth is critical, such as texture-heavy gaming at high resolutions, the GTX 670M’s fixed memory pool could offer more predictable performance. The GTX 670M also has a higher pixel rate, at 8.372 GPixel/s, though this is marginally lower than the P580’s 9.000 GPixel/s. The GTX 670M’s texture rate is 33.49 GTexel/s, which is less than half of the P580’s 72.00 GTexel/s, reinforcing the Intel part’s compute dominance.

For gaming, the GTX 670M’s discrete nature and 24 ROPs might provide better fill-rate-bound performance in certain legacy titles, but the benchmark data does not include gaming tests. The P580’s 9 ROPs are a limitation, but its higher texture rate and shading unit count suggest it excels in shader-bound workloads. In essence, the P580 wins in compute-heavy applications, while the GTX 670M’s dedicated memory could be argued to have an edge in memory-bandwidth-sensitive scenarios, though this is not empirically verified in the provided data.

Architecture Differences

The architectural divide between these two GPUs is stark. The Intel Iris Pro Graphics P580 is built on the Skylake GT4e chip, using Intel’s Generation 9.0 architecture, specifically the HD Graphics-W (Skylake) generation. It is manufactured on a 14 nm+ process at Intel’s foundry. The GTX 670M uses the GF114 chip, based on NVIDIA’s Fermi 2.0 architecture, and is produced on a 40 nm process at TSMC. The process node difference is significant: 14 nm+ versus 40 nm, which explains how Intel can pack 576 shading units, 72 TMUs, and 9 ROPs into a part with a 15 W TDP, while NVIDIA’s 336 shading units, 56 TMUs, and 24 ROPs require a 75 W TDP.

The GTX 670M does have specific transistor data: 1,950 million transistors on a 332 mm² die, with a transistor density of 5.9M / mm². The P580’s transistor count and die size are not listed, so no direct comparison can be made on those metrics. The GTX 670M’s memory configuration is fixed: 1,536 MB of GDDR5 on a 192-bit bus with 72.00 GB/s bandwidth. The P580 relies on system shared memory, with type and bus width also system shared, making its performance dependent on the host system’s memory subsystem.

Feature support also diverges. The P580 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The GTX 670M supports DirectX 12 (11_0) and OpenGL 4.6, but has no Vulkan support listed. This is a critical differentiator for modern applications that use Vulkan for lower overhead and better multi-threading. The P580’s bus interface is a Ring Bus, typical for integrated parts, while the GTX 670M uses an MXM-B (3.0) interface, designed for modular laptop graphics. The GTX 670M also has a predecessor (GeForce 500M) and successor (GeForce 700M), whereas the P580 lists none, indicating it may be a one-off or part of a broader integrated lineup without a clear product line succession.

The GTX 670M’s memory clock is listed at 750 MHz, with 3 Gbps effective, while the P580’s memory clock is system shared, and its base clock is 350 MHz with a boost of 1000 MHz. The GTX 670M does not list base or boost clocks for the core, only memory. This lack of core clock data for the GTX 670M makes direct clock-for-clock comparisons impossible, but the shading unit count and FP32 GFLOPS figures already tell the performance story.

The Verdict

The data is unambiguous in compute performance: the Intel Iris Pro Graphics P580 is the superior part in the Geekbench OpenCL benchmark, leading by 39.4%. Its higher shading unit count (576 vs. 336), higher texture rate (72.00 GTexel/s vs. 33.49 GTexel/s), and higher FP32 throughput (1,152.0 GFLOPS vs. 803.7 GFLOPS) all contribute to this outcome. The P580 also offers Vulkan support, which the GTX 670M lacks, making it more future-proof for modern APIs. The P580 achieves this while consuming only 15 W, compared to the GTX 670M’s 75 W, a remarkable efficiency advantage given the performance lead.

However, the GTX 670M is not without its merits. Its dedicated 1,536 MB of GDDR5 memory with 72.00 GB/s bandwidth provides a fixed memory resource that does not compete with the system RAM for bandwidth. For users running older games or applications that are memory-bandwidth-bound, the GTX 670M might still be viable, but the benchmark data does not support a performance win. The GTX 670M’s pixel rate of 8.372 GPixel/s is lower than the P580’s 9.000 GPixel/s, but its 24 ROPs versus the P580’s 9 ROPs could offer better fill-rate efficiency in specific rendering paths.

Strictly from the data, the Intel Iris Pro Graphics P580 is the recommended choice for anyone prioritizing compute performance, modern API support, and power efficiency. The GTX 670M’s only argument is its dedicated memory, which is not quantified in any benchmark win. Given that the P580 wins the only head-to-head test and offers a broader feature set, including Vulkan and FP16 support, it is the clear winner in this comparison. The GTX 670M, being end-of-life and from an older architecture, is best suited for legacy systems where its MXM form factor and dedicated memory are already in place.

FAQ

Q: Which GPU has a higher Geekbench OpenCL score?

A: The Intel Iris Pro Graphics P580 scores 9,082, while the NVIDIA GeForce GTX 670M scores 6,513, giving the P580 a 39.4% lead.

Q: Does the NVIDIA GeForce GTX 670M support Vulkan?

A: No, the GTX 670M’s API list includes DirectX 12 (11_0) and OpenGL 4.6, but no Vulkan support is listed. The P580 supports Vulkan 1.3.

Q: What is the difference in shading units between the two GPUs?

A: The Intel Iris Pro Graphics P580 has 576 shading units, while the NVIDIA GeForce GTX 670M has 336 shading units.

Q: How much dedicated memory does the GTX 670M have?

A: The GTX 670M has 1,536 MB of GDDR5 memory on a 192-bit bus, with a bandwidth of 72.00 GB/s. The P580 uses system shared memory.

Q: What is the TDP of each GPU?

A: The Intel Iris Pro Graphics P580 has a TDP of 15 W, while the NVIDIA GeForce GTX 670M has a TDP of 75 W.

Q: Which GPU has a higher FP32 performance?

A: The P580 delivers 1,152.0 GFLOPS, which is higher than the GTX 670M’s 803.7 GFLOPS.