AMD vs. Apple vs. Intel: Top Integrated GPUs Compared

Comparing Current Integrated Graphics from AMD, Apple, and Intel

Processors in modern notebooks and desktops typically include a built-in (or integrated) GPU—some fully capable of ray tracing, accelerating advanced video codecs like AV1, and driving multiple high-resolution displays. While these integrated GPUs (iGPUs) obviously have fewer resources than large gaming GPUs—fewer shader cores, raster pipelines, and a narrower memory interface—they still serve most casual workloads.

We took a close look at the fastest desktop iGPUs from AMD, Apple, and Intel:

• AMD: Radeon 780M in the Ryzen 7 8700G
• Apple: The iGPU in the Apple M4 Pro (in the Mac mini)
• Intel: The Intel Graphics in the Core Ultra 245K

Our primary focus is on desktop chips to see if an iGPU is sufficient for a compact PC’s needs or if a dedicated graphics card is mandatory. While some mobile CPUs have even more powerful integrated graphics, this analysis covers the top-tier iGPUs in desktop processors from these three vendors.

iGPU Architecture Basics

For each company, the iGPU architecture is fundamentally similar:

1. Shader cores handle both 3D effects in games and GPGPU tasks (like AI or image processing). They’re accompanied by fixed-function units for ray tracing, texture filtering, and rasterisation. iGPUs typically have 256 to 5120 of these cores, far fewer than full-blown discrete GPUs (e.g., thousands or tens of thousands in high-end graphics cards).
2. Media engine for hardware-based video decode and encode, covering popular codecs like H.264, H.265, VP9, and increasingly AV1. iGPUs help offload these workloads from CPU cores. However, the available features vary per brand and generation.
3. Display engine for output to up to four monitors at once, usually including high-bandwidth outputs like HDMI 2.1 or DisplayPort 2.0. 8K or high refresh rates (120 Hz) in 4K are possible, though actual board-level connectors depend on the motherboard or system manufacturer.

Because iGPUs lack dedicated video memory, they share the system’s main memory, leading to potential bandwidth bottlenecks. Gaming performance heavily depends on memory throughput, so systems with LPDDR5X or higher clocked DDR5 can see better results.

Performance Overview

Synthetic Benchmarks (3DMark)

We tested the three desktop iGPUs in 3DMark’s DirectX 12 scene “Steel Nomad” (where possible), comparing them to a GeForce RTX 3050 (~€200). They offered roughly half the framerate of the RTX 3050 in Steel Nomad, except for the Apple M4 Pro’s GPU, which in a “Light” variant topped the RTX 3050 by about 50%. In ray-tracing benchmarks, AMD’s Radeon 780M was 3.6× faster than Intel’s Graphics, but nowhere near the discrete GPU. Even the RTX 3050 itself is generally too slow for comfortable ray tracing in modern games.

Real Gaming Tests

In actual 3D titles:

• Casual/E-sports Games: League of Legends or The Sims 4 generally run at high details with triple-digit framerates on iGPUs.
• Moderately Demanding Shooters: For instance, CounterStrike 2.0 at 1080p can stay near 60 fps if you slightly lower quality settings on Intel or AMD iGPUs—visually not a big compromise.
• Heavier Titles: In more demanding games (Cyberpunk, for example), you must drastically cut settings or resolution to approach playable performance—and even so, the result might be below 30 fps.

GPGPU and Video Encode

In addition to graphics, modern iGPUs accelerate:

1. Video Encoding/Decoding
   • AV1, H.265, H.264, and VP9 are usually covered.
   • Using these hardware blocks (Quick Sync for Intel, VCE for AMD, VideoToolbox for Apple, etc.) drastically reduces CPU usage compared to software encode.
2. Compute Tasks
   • Tasks such as image denoising in Adobe Lightroom or some AI-based upscaling can use iGPUs, albeit slower than discrete GPUs.
   • Apple’s M4 Pro and an entry-level discrete GPU like the RTX 3050 can handle a 36 MP RAW image in under 30 s, while AMD/Intel iGPUs need several minutes.

Memory Limitations

Because iGPUs share system memory with the CPU:

• Memory Bandwidth is a core bottleneck. DDR5-5200 in dual channel yields ~83 GB/s on AMD’s Ryzen 7 8700G. That’s a fraction of a discrete GPU’s bandwidth (e.g., 1.8 TB/s for an RTX 5090 with GDDR7).
• Apple leverages wide LPDDR5X memory on-package for higher throughput, enabling far more powerful iGPUs in M4 Pro/Max variants.
• Mobile Chips often use soldered LPDDR5X at 8533 MT/s, outpacing typical DDR5 modules for desktops (JEDEC 5600 MT/s).

Practical Conclusion

1. Light Gaming / E-Sports
   Many popular titles run fluidly on modern iGPUs from AMD, Apple, and Intel at 1080p, especially if you dial down effects. E-sports (Dota 2, League of Legends, Overwatch, CS:GO) typically hit 60+ fps.
2. Casual 3D
   The performance is enough for older or simpler 3D games. Ray tracing is beyond iGPUs’ scope, though they have the hardware support in theory—it’s generally too slow for AAA titles.
3. Video Tasks / Low-Power Systems
   Thanks to hardware-accelerated codecs, iGPUs can handle 4K playback, multi-display setups, and moderate compute tasks.
4. Apple M4 Pro outpaces AMD and Intel iGPUs, thanks to many more shader cores and wide memory channels. If you need a portable or compact system with strong built-in graphics, Apple’s M-series is unmatched among iGPUs.
5. Ryzen AI Max 300 (Strix Halo) could challenge Apple’s approach, melding powerful integrated graphics with a wide memory interface for Windows laptops—expected starting March.

Bottom line: Today’s integrated GPUs have improved significantly—fast enough for many games, even rendering older 3D titles at comfortable framerates. For advanced ray tracing or heavy AAA gaming, a discrete GPU remains essential. But for most everyday tasks, e-sports, or casual gaming, the best iGPUs from AMD, Apple, and Intel are up to the challenge. And as integrated memory bandwidth expands, the performance gap between iGPUs and entry-level discrete GPUs may narrow further.