Tag: GFXBench

  • Apple M3 Ultra Beats Nvidia RTX 5070 Ti but Lags Behind RTX 5080

    Apple M3 Ultra Beats Nvidia RTX 5070 Ti but Lags Behind RTX 5080

    Key Takeaways

    1. The M3 Ultra features a 32-core CPU with 24 high-performance cores and an 80-core GPU, making it the fastest chip ever in a Mac.

    2. The M3 Ultra excels in GPU performance, achieving 374 FPS in the GFXBench 4K Aztec Ruins test, nearly doubling the M3 Max’s score.

    3. In the Cinebench 2024 GPU test, the M3 Ultra scored around 20,000 points, outperforming the AMD Radeon RX 7900 XTX by 25% and matching the Nvidia RTX 4070 Ti.

    4. While the M3 Ultra shows great potential, its single-threaded performance may not be as strong, and real-world use may vary across different tasks.

    5. The M3 Ultra’s efficiency and substantial VRAM capacity set it apart, with expectations for more detailed performance comparisons against competitors in the future.


    Apple has just introduced the M3 Ultra, which is, without a doubt, the fastest chip ever put into a Mac. It features an impressive 32-core CPU that includes 24 high-performance cores and an 80-core GPU. The M3 Ultra is expected to show stunning results in benchmarks, though its performance in single-threaded tasks might not be as remarkable. Recently, it appeared on Geekbench, but the results were a bit underwhelming. However, as more reviews are released, the M3 Ultra’s capabilities are becoming clearer.

    GPU Performance Takes Center Stage

    According to a review by Dave2D and data from GFXBench, the M3 Ultra-powered Mac Studio excels in GPU tasks, which aligns with expectations since the M3 Max, which is the base for the Ultra, was a significant leap in GPU tech. In the GFXBench 4K Aztec Ruins (High Tier) test, the M3 Max scored about 200 FPS, while the M3 Ultra nearly doubled that, achieving an impressive 374 FPS. This performance puts the M3 Ultra (Metal) roughly 9% below the GeForce RTX 5080 (DirectX 12) and around 20% ahead of the RTX 5070 Ti, based on GFXBench’s findings. When compared to the M4 Max found in higher-end MacBook Pros, priced at $3,564 on Amazon, the M3 Ultra is approximately 63.6% faster.

    Cinebench Results Show Promise

    In the Cinebench 2024 GPU test, the M3 Ultra scored about ~20,000 points, as reported by The Verge. This figure puts it a staggering 25% ahead of the AMD Radeon RX 7900 XTX, which scores around ~16,000 points on the same benchmark. Interestingly, the M3 Ultra performs similarly to the Nvidia RTX 4070 Ti in this test. However, it’s worth noting that a handful of benchmark tests aren’t necessarily indicative of real-world use, especially given the variety of APIs in play.

    Furthermore, it’s likely that the M3 Ultra will excel in many tasks due to its specialized engines, although it may struggle in areas like 3D rendering when pitted against top-tier RTX cards. When it comes to efficiency and VRAM, the M3 Ultra stands out, especially since no consumer GPU gets close to delivering half a terabyte of reasonably fast VRAM. As time goes on, more practical performance comparisons between the M3 Ultra and its x86 rivals are expected, providing clearer insights for potential buyers.

    Source:
    Link


  • Intel Arc 140T Graphics Offers 60% Performance Boost Over Lunar Lake

    Intel Arc 140T Graphics Offers 60% Performance Boost Over Lunar Lake

    A new Intel Arc graphics chip has surfaced on GFXBench, demonstrating notable performance improvements compared to its Lunar Lake version. This GPU, known as the Arc 140T, delivered frame rates that exceed those of the current Arc 140V integrated graphics by over 60 percent.

    Benchmark Results

    According to benchmark results shared by Michael (@miktdt) on X, the Arc 140T achieved a score of 11,056 frames at 178.3 FPS in GFXBench’s OpenGL test. In contrast, the Arc 140V models with 16GB and 8GB of memory only managed 110.3 FPS (6,839 frames) and 106.7 FPS (6,613 frames), respectively.

    Performance Differences

    Despite both the 140T and 140V featuring the same eight Xe Cores, there is a clear performance gap—likely stemming from differences in design or implementation. Based on the specifications and naming conventions, it appears that the 140T is more geared towards integrated graphics rather than functioning as a standalone GPU.

    Future Prospects

    It’s possible that the 140T is designed for Intel’s forthcoming Arrow Lake-H/HX mobile chips. This aligns with Intel’s approach: Arrow Lake is centered on boosting performance, while Lunar Lake emphasizes energy efficiency. The variation in performance between the two could be linked to the types of transistors used, with the 140T probably utilizing higher-performance transistors compared to the power-efficiency-oriented ones in the 140V from Lunar Lake.

    If everything proceeds as expected, the Arc 140T could debut in Arrow Lake mobile chips around early 2025. As of now, Intel has not officially verified these specifications or performance figures.