Apple’s M5 Pro & M5 Max easily beat the RTX 5090 in PugetBench creator benchmarks

In a shocking turn of events, Apple’s latest M5 Pro and M5 Max chips have left the tech community abuzz by outperforming NVIDIA’s flagship RTX 4090 graphics card in PugetBench creator benchmarks. For years, NVIDIA’s GeForce series has reigned supreme in the world of graphics processing, but Apple’s new ARM-based processors have seemingly bridged the gap. As we dive into the details, it’s clear that Apple’s innovative approach to chip design has paid off, at least in this specific benchmark.

The Rise of Apple’s M-Series Chips

Apple’s transition to its in-house M-series chips began with the M1 chip in 2020, which marked a significant departure from Intel’s processors. The M1 chip’s impressive performance and efficiency set the stage for future iterations, including the M5 Pro and M5 Max. These new chips are designed to cater to the growing demands of creators, who require immense processing power for tasks like video editing, 3D modeling, and software development.

According to Puget Systems, a renowned developer of high-performance workstations, the M5 Pro and M5 Max chips have demonstrated unparalleled performance in their creator benchmarks. The PugetBench benchmark suite is widely regarded as a reliable measure of a system’s capabilities in real-world creative applications. By leveraging this benchmark, we can gain a deeper understanding of how Apple’s latest chips stack up against the competition.

The M5 Pro and M5 Max chips boast an impressive array of features, including a unified memory architecture, high-bandwidth memory, and a robust neural engine. These features enable the chips to handle demanding tasks with ease, making them an attractive option for creators who require a seamless workflow.

Head-to-Head Comparison with NVIDIA’s RTX 4090

NVIDIA’s GeForce RTX 4090 is widely considered one of the most powerful consumer-grade graphics cards available, with a strong focus on AI-enhanced graphics and compute performance. However, when pitted against Apple’s M5 Pro and M5 Max chips in PugetBench creator benchmarks, the results are striking. According to benchmark results published by Puget Systems, the M5 Pro and M5 Max chips outperform the RTX 4090 in several key metrics, including video editing, color grading, and visual effects.

One possible explanation for this performance disparity lies in the differing architectures of the two systems. While the RTX 4090 relies on a traditional GPU-centric approach, Apple’s M5 Pro and M5 Max chips utilize a more integrated design, combining CPU, GPU, and neural engine capabilities into a single system-on-a-chip (SoC). This integrated approach may provide a significant advantage in applications that rely heavily on inter-component communication and data transfer.

It’s essential to note that these results are specific to PugetBench creator benchmarks and may not reflect real-world performance in all scenarios. However, they do suggest that Apple’s innovative approach to chip design has yielded significant benefits for creators who rely on these applications.

Implications for the Future of Creator Workstations

The emergence of Apple’s M5 Pro and M5 Max chips as a force to be reckoned with in creator benchmarks raises important questions about the future of workstation design. As the line between CPUs and GPUs continues to blur, we may see a shift towards more integrated, SoC-based designs that can efficiently handle a wide range of tasks. This could have significant implications for the development of future workstations, which may prioritize flexibility, scalability, and power efficiency over traditional performance metrics.

Moreover, the success of Apple’s M-series chips may prompt other manufacturers to re-examine their own design approaches, potentially leading to a new wave of innovative products that challenge the status quo. As the creative industry continues to evolve, one thing is clear: the next generation of workstations will be shaped by the intersection of AI, graphics, and high-performance computing.

With the M5 Pro and M5 Max chips leading the charge, it’s exciting to consider what the future holds for creator workstations. Will Apple’s momentum continue, or will NVIDIA and other manufacturers respond with their own innovative solutions? One thing is certain – the next chapter in the evolution of creator workstations is about to unfold.

Why PugetBench Matters More Than Raw FPS

Most gamers obsess over frame-rates, but creators live and die by frame-times—those micro-stutters that turn a four-hour color-grade into an overnight ordeal. PugetBench doesn’t just spit out synthetic numbers; it opens the same 8K RED raw timeline in DaVinci Resolve, layers on noise-reduction, adds a stack of LUTs, then exports to ProRes 4444 while tracking every hiccup. In that crucible, the M5 Max’s 40-core GPU and 512-bit unified memory bus shine because they’re not shuttling textures across PCIe lanes; they’re already sitting next to the CPU on the same slab of silicon. Result? A 17 % faster export than the RTX 5090, but—more importantly—zero dropped frames during live scrubbing, something even the 5090 can’t claim once VRAM usage creeps past 20 GB.

There’s a hidden kicker: Apple’s media engines. Every M5 Pro/Max packs two ProRes encode blocks and four decode blocks that sip a mere 35 W. NVIDIA’s NVENC is potent, but it’s general-purpose; Puget’s Premiere Pro test hits 4x 8K ProRes 422 streams simultaneously, and while the 5090 chokes after two, the M5 Max keeps previewing buttery-smooth playback on battery power. For documentarians cutting in the field from a tent in Patagonia, that’s the difference between shipping a cut or shipping a generator.

The Thermal Secret: A Laptop That Thinks It’s a Workstation

Picture this: a 15-inch MacBook Air-thin chassis pulling 3 100 points in PugetBench’s Photoshop test, 9 % above the 5090 rig that weighs 14 kg and sounds like a Dyson on steroids. Apple’s trick isn’t brute force; it’s the 3 nm “Ultra- Fusion” interposer that stacks DRAM vertically, shortening trace lengths to millimetres. Less distance means less resistance, so the M5 Max can hold its 3.9 GHz boost for 38 minutes before throttling—long enough to finish a Blender classroom scene render that would make most Intel HX laptops cry thermal uncle in eight.

Yes, the 5090 still claws back a few seconds in raw Blender CUDA, but add a second fan or open a window in summer and the desktop’s clock drift widens the gap again. Meanwhile the M5 Max keeps rendering on an airline tray table at 30 000 ft without the flight attendant demanding you unplug “for safety.”

The Ecosystem Lock-In Nobody Talks About

Here’s where Apple pivots from benchmark braggart to chess master. PugetBench’s After Effects test uses the new Multi-Frame Rendering API; Adobe wrote a Metal path that splits frames across the M5’s GPU, neural engine, and performance cores simultaneously. NVIDIA’s CUDA path is fast, but it’s still a single-queue pipeline. The result: a 22 % lead for the M5 Max in a 16-bit 4K composite with 150 layers of particle systems. More crucially, the same project opens identically on a Mac Studio M5 Ultra back at the office—no re-optimization, no driver roulette, no wondering if the Windows box will blue-screen after a forced update mid-client review.

That seamless hand-off is Apple’s real GPU: the Gnarly Productivity Umbrella. Editors can shoot B-roll on an iPhone 16 Pro, AirDrop the ProRes log to a MacBook, grade in HDR on the flight, then walk into post with the same cable-less machine already synced to the SAN. The RTX 5090 might win a raw teraflop beauty pageant, but it can’t offer that narrative cohesion—no Matter port, no shared clipboard, no Final Cut project that opens in