Skip to content
TheChatArchiveThe Chat Archive
Geminigemini-2.0-flash-thinking-exp-01-21· GoogleHugging Face LMArena Human Preference 140K

how far do I have to travel back in time for my iphone 15 to be considered a proper…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-mini-2024-07-18’s answer.
  1. User

    how far do I have to travel back in time for my iphone 15 to be considered a proper supercomputer?

  2. Gemini Assistant

    Okay, let's break down the computational power of an iPhone 15 and compare it to historical supercomputers.

    1. iPhone 15 Performance:

      • The iPhone 15 series uses either the A16 Bionic (iPhone 15, 15 Plus) or the A17 Pro (iPhone 15 Pro, 15 Pro Max) chip.
      • The A17 Pro chip in the iPhone 15 Pro is significantly more powerful, particularly its GPU and Neural Engine. While exact, directly comparable FLOPS numbers (especially 64-bit floating-point, common for scientific supercomputing) are hard to nail down for mobile chips optimized for graphics and AI, estimates for the A17 Pro's peak theoretical performance in 32-bit floating-point (FP32) tasks often range from 3 to 4 TFLOPS (TeraFLOPS - trillions of floating-point operations per second). The neural engine can hit much higher numbers for integer operations relevant to AI (around 35 TOPS). Let's use the 3-4 TFLOPS range as our benchmark for some kind of comparable peak computational power.
      • The A16 Bionic in the base iPhone 15 is closer to 1 TFLOP peak FP32 performance.
    2. Historical Supercomputer Performance:

      • Supercomputers are typically measured in FLOPS (Floating-point Operations Per Second).
      • The GFLOPS era (billions of FLOPS) was primarily the late 1980s and early 1990s.
      • The TFLOPS era (trillions of FLOPS) began in the mid-to-late 1990s.
      • The Intel ASCI Red, completed in 1996, is often cited as the first computer to achieve sustained performance over 1 TFLOP (reaching 1.8 TFLOPS initially, later upgraded to 3.1 TFLOPS).
      • Other leading supercomputers from the late 1990s, like the Hitachi SR8000/MPP (1996), also pushed into the low single-digit TFLOPS range (e.g., 2.2 TFLOPS peak).

    The Comparison:

    • An iPhone 15 (with the A16 chip, ~1 TFLOP) has peak performance comparable to the very first supercomputers to break the 1 TFLOP barrier in 1996.
    • An iPhone 15 Pro (with the A17 Pro chip, ~3-4 TFLOPS) has peak performance comparable to or exceeding the top supercomputers of the late 1990s (think 1997-1999).

    Conclusion:

    You would have to travel back to the mid-to-late 1990s (roughly 25-30 years ago) for your iPhone 15 Pro to be considered a proper supercomputer in terms of raw peak computational power (FLOPS). Even the base iPhone 15 would be competitive with the first machines to achieve TFLOPS performance.

    Keep in mind this is primarily a comparison of peak FLOPS. Supercomputers of that era still had advantages in areas like massive memory, specialized interconnects for large-scale parallel processing, and capabilities for double-precision floating-point (FP64) tasks crucial for scientific simulations, which mobile chips are less optimized for. But in terms of sheer operations per second potential, the iPhone 15 is truly astonishing compared to the machines of that time.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.