Intel Arc Pro B60 vs NVIDIA GeForce 920M Comparison

The NVIDIA GeForce 920M and the Intel Arc Pro B60 represent two vastly different eras of GPU design, separated by a decade of architectural evolution. The data shows a generational chasm in raw capability, yet both cards occupy the same 20th percentile in the global performance distribution, which is a curious statistical artifact. This analysis will dissect their benchmark results, architectural philosophies, and practical use cases based strictly on the provided facts.

Head-to-Head Benchmarks

The benchmark data available for these two cards is almost entirely non-overlapping, making a direct apples-to-apples comparison difficult. The GeForce 920M has results for Geekbench OpenCL and Vulkan, scoring 3725 and 2849 respectively. The Arc Pro B60, on the other hand, has a suite of Passmark tests and a 3DMark Steel Nomad result. The only point of direct comparison comes from the `nearestRivals` data, which is revealing.

According to the rival data, the Intel Arc Pro B60's average benchmark score of 3182 places it 3.2% behind the NVIDIA GeForce 920M's average score of 3287. This is a counterintuitive result given the massive specification disparity. The 920M's average is buoyed by its Geekbench scores, which are likely compute-heavy workloads where its simpler architecture can still perform adequately. However, this single data point should not be interpreted as the 920M being the faster card overall. It is a statistical quirk of the averaging methodology across different test suites.

Looking at the broader rival landscape for each card provides more context. The GeForce 920M's closest competitor is the NVIDIA GeForce GT 730M, with the 920M trailing by a razor-thin 0.9%. It also sits 1.4% behind the Intel HD Graphics 530 and 3% behind the Intel HD Graphics P4600, but leads the NVIDIA GeForce GT 640 by 2.4%. These are all very low-end, integrated or entry-level discrete parts, confirming the 920M's position as a basic mobile GPU.

The Arc Pro B60's rival list is starkly different, even if the percentile is the same. It sits 0.6% ahead of the NVIDIA Quadro P1000, a professional workstation card. It trails the GeForce GT 640 by 0.9%, which is the same GPU the 920M beats. Most surprisingly, it is 3.5% ahead of the NVIDIA GeForce RTX 5080 SUPER, a flagship-class card. This suggests the `avgBenchmarkScore` for the Arc Pro B60 is being dragged down by its diverse set of tests, many of which (like Passmark DirectX 9 and 10) may not be representative of its true modern performance. The Arc Pro B60's raw performance in modern workloads is clearly in a different league, yet its average score is skewed by legacy test compatibility.

The Arc Pro B60's own benchmark suite shows a massive spread. Its Passmark G3D score is 14580, a figure that dwarfs the 920M's entire benchmark output. In contrast, its Passmark DirectX 10 score is a paltry 61, and DirectX 12 scores 76. This indicates that the card's drivers are heavily optimized for specific modern APIs and workloads, while legacy DirectX paths are severely underperforming. The 3DMark Steel Nomad DX12 score of 2646 is a modern, demanding test, and it represents a level of performance the 920M could not even attempt to render.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce 920M has a slightly higher average benchmark score of 3287 compared to the Intel Arc Pro B60's 3182. This is a difference of 3.2% in favor of the 920M, based on the `nearestRivals` data.

Q: Is the GeForce 920M truly faster than the Arc Pro B60?

A: No. The average score is a statistical artifact. The 920M's scores come from Geekbench tests, while the Arc Pro B60's average is pulled down by extremely low Passmark scores in legacy DirectX 9 (179) and DirectX 10 (61) tests. In modern workloads, the Arc Pro B60's Passmark G3D score of 14580 is several orders of magnitude higher than anything the 920M can produce.

Q: What is the performance delta between the Arc Pro B60 and the GeForce GT 640?

A: The Arc Pro B60 is 0.9% behind the NVIDIA GeForce GT 640 in average benchmark score. Interestingly, the GeForce 920M is 2.4% ahead of that same GT 640, further highlighting the inconsistencies in comparing across different test suites.

Q: How does the Arc Pro B60 compare to a modern flagship like the RTX 5080 SUPER?

A: According to the data, the Arc Pro B60's average benchmark score is 3.5% higher than that of the NVIDIA GeForce RTX 5080 SUPER. This is likely due to the RTX 5080 SUPER's score being an average across a different set of demanding tests, while the Arc Pro B60's average is diluted by its legacy Passmark results.

Q: What can the GeForce 920M's benchmark scores tell us about its target market?

A: The GeForce 920M's Geekbench OpenCL score of 3725 and Vulkan score of 2849 place it in direct competition with integrated graphics like the Intel HD Graphics 530. Its rival list is populated entirely by low-end discrete and integrated parts, indicating it was designed for basic multimedia and light, older gaming workloads.

Q: Does the Arc Pro B60 have any benchmark wins?

A: Yes, in its own test suite, the Arc Pro B60 achieves a Passmark G3D score of 14580 and a 3DMark Steel Nomad DX12 score of 2646, which are its strongest results. It also posts a Passmark GPU Compute score of 7029, indicating strong compute throughput.

Architecture Differences

The architectural gap between these two GPUs is immense, representing a decade of technological progress. The GeForce 920M is built on NVIDIA's Kepler 2.0 architecture using the GK208B chip, fabricated on a 28 nm process at TSMC. This is a legacy architecture with 1,020 million transistors packed into a small 87 mm² die, yielding a transistor density of 11.7M / mm². It lacks any dedicated ray tracing or tensor cores, relying on 384 shading units, 32 texture mapping units, and just 8 ROPs.

In stark contrast, the Intel Arc Pro B60 is based on the modern Xe2-HPG architecture, using the BMG-G21 chip on a advanced 5 nm TSMC process. This process allows Intel to pack a massive 19,600 million transistors into a 272 mm² die, achieving a density of 72.1M / mm². This is a fundamental shift in manufacturing capability, enabling 2,560 shading units, 160 TMUs, and 80 ROPs. Critically, the Arc Pro B60 features 20 dedicated ray tracing cores, a feature entirely absent from the 920M. The architecture also supports DirectX 12 Ultimate (12_2), while the 920M is limited to DirectX 12 (11_0), meaning it cannot leverage the latest rendering features like hardware-accelerated ray tracing or mesh shaders.

The memory subsystem also tells a story of architectural evolution. The 920M uses a 64-bit bus to connect to 2 GB of DDR3 memory, a configuration that was entry-level even in 2015. The Arc Pro B60 uses a 192-bit bus for its 24 GB of GDDR6 memory, providing a bandwidth of 456.0 GB/s versus the 920M's 14.40 GB/s. This is a 31.6x increase in memory bandwidth, which is critical for modern high-resolution textures and compute workloads. The API support also differs, with the Arc Pro B60 supporting Vulkan 1.4 compared to the 920M's Vulkan 1.2.175.

Specification Differences

The specification sheets for these two GPUs read like they are from different product categories. The clock speeds alone show a significant difference: the GeForce 920M is locked at a base and boost of 954 MHz, while the Arc Pro B60 has a base clock of 2000 MHz and a boost clock of 2400 MHz. This 145% increase in boost clock is a direct result of the more efficient 5 nm process node.

Memory specifications are the most dramatic differentiator. The 920M offers 2 GB of DDR3 on a 64-bit bus, resulting in 14.40 GB/s of bandwidth. The Arc Pro B60 offers 24 GB of GDDR6 on a 192-bit bus, delivering 456.0 GB/s. This is a 12x increase in memory capacity and a 31.6x increase in bandwidth. The power envelope is also worlds apart: the 920M is a 33 W IGP (integrated graphics processor) with no power connectors, while the Arc Pro B60 is a 200 W dual-slot card requiring a single 8-pin power connector and a suggested 550 W PSU.

Compute throughput is where the generational leap is most apparent. The 920M's FP32 performance is 732.7 GFLOPS, while the Arc Pro B60 achieves 12.29 TFLOPS, an improvement of over 16 times. The Arc Pro B60 also supports FP16 compute at 24.58 TFLOPS, a feature the 920M lacks entirely. Pixel and texture rates follow suit: the Arc Pro B60's 192.0 GPixel/s and 384.0 GTexel/s dwarf the 920M's 7.632 GPixel/s and 30.53 GTexel/s. The bus interface also changes from PCIe 3.0 x8 on the 920M to PCIe 5.0 x8 on the Arc Pro B60, doubling the potential bandwidth for data transfer to the CPU. Finally, the display outputs differ, with the 920M being "Portable Device Dependent" and the Arc Pro B60 offering 4x mini-DisplayPort 2.1.

Where Each One Wins

The GeForce 920M wins in the context of its original design goals: low-power, basic mobile acceleration. Its 33 W TDP and IGP form factor mean it can be integrated into thin laptops without discrete cooling solutions. Its benchmark scores, while low, are competitive with the integrated graphics of its era, such as the Intel HD Graphics 530 and HD Graphics P4600. For a user in 2015, it offered a modest step up from integrated graphics for light photo editing or older DirectX 11 games. It "wins" in any scenario where power efficiency and minimal physical footprint are the absolute priority, and where the software workload is from its generation.

The Intel Arc Pro B60 wins in every metric of raw performance and modern feature support. Its 24 GB of VRAM is essential for large AI models, complex 3D rendering, or video editing with high-resolution footage. The 20 ray tracing cores allow for hardware-accelerated ray tracing, a feature that is now standard in modern games and professional visualization. Its Passmark G3D score of 14580 and 3DMark Steel Nomad score of 2646 demonstrate that it is a competent performer in modern DirectX 12 titles and applications. Its compute capabilities, evidenced by the 7029 Passmark GPU Compute score, make it suitable for GPU-accelerated workloads that the 920M cannot handle at all. It wins on connectivity, offering four modern DisplayPort 2.1 outputs for multi-monitor professional setups.

The Verdict

The data paints a clear picture of two different products for two different times. The NVIDIA GeForce 920M is an end-of-life legacy part. It should be considered only for legacy systems where a drop-in, low-power replacement is needed, or for running software from its 2015 release era. Its performance is strictly entry-level, and any modern task will be a struggle.

The Intel Arc Pro B60, despite its low 20th percentile ranking, is a modern professional GPU. Its average benchmark score is misleadingly low due to poor performance in legacy DirectX 9 and 10 tests. Its true capabilities lie in its 24 GB of VRAM, ray tracing cores, and massive compute throughput. Anyone needing a workstation card for content creation, AI inference, or modern CAD visualization should look at the Arc Pro B60's high-end scores (like the 14580 G3D) rather than its average. The GeForce 920M is a relic; the Arc Pro B60 is a forward-looking tool that happens to share a percentile bracket with its ancient rival due to flawed averaging.