Console specs get thrown around a lot in arguments about whether a PS5 can “keep up” with PC gaming hardware, but the actual answer to what GPU does the PS5 have is more specific than most of those debates assume. Sony didn’t just drop in an off-the-shelf graphics card.
- The PS5 uses a custom AMD GPU built on the RDNA 2 architecture, not a standard retail Radeon card.
- It has 36 compute units running at up to 2.23 GHz, delivering a peak of roughly 10.3 teraflops.
- It’s paired with 16GB of GDDR6 memory shared between the GPU and CPU, running at 448 GB/s bandwidth.
- This custom chip supports hardware-accelerated ray tracing, a first for a PlayStation console.
A Custom AMD Chip, Not a Shelf Part
The PS5’s GPU is built by AMD specifically for Sony, combined on the same system-on-chip as the console’s custom CPU. It’s based on AMD’s RDNA 2 architecture, the same graphics architecture family that powered AMD’s Radeon RX 6000 series desktop GPUs. AMD’s own overview of that architecture is available on its RDNA 2 technology page.
Because it’s a custom design, the PS5 GPU isn’t a direct match for any single retail graphics card. Sony worked with AMD to tune the configuration specifically around the console’s power, thermal, and cost targets, rather than reusing an existing desktop GPU design unchanged.
The Core GPU Specs
| Spec | PS5 |
|---|---|
| Architecture | Custom AMD RDNA 2 |
| Compute units | 36 CUs |
| Clock speed | Up to 2.23 GHz (variable frequency) |
| Peak performance | Approximately 10.28 to 10.3 teraflops |
| Memory | 16GB GDDR6, shared with CPU |
| Memory bandwidth | 448 GB/s |
| Ray tracing | Hardware-accelerated, supported |
That teraflop figure gets cited constantly in PC vs console comparisons, but it’s worth treating with some caution, a point covered further below.
Variable Frequency: The PS5’s Unusual Approach
Unlike most GPUs, which run at a fixed clock speed, the PS5’s GPU uses what Sony calls a variable frequency design, also referred to as SmartShift. The clock speed can adjust based on workload, power draw, and thermal conditions, rather than sitting at one locked number at all times.
- This lets the console balance power between the CPU and GPU dynamically depending on what a game needs most at a given moment.
- The 2.23 GHz figure represents the GPU’s typical ceiling under normal conditions, not a value it’s locked to at all times.
- This design was somewhat controversial at launch, since it makes direct power comparisons to fixed-clock PC GPUs less straightforward than a simple teraflop number suggests.
Fact Check: teraflops are a useful rough indicator but don’t translate directly into real-world gaming performance, since actual output also depends on architecture efficiency, memory bandwidth, and how well a game is optimized for the specific hardware. A console and a PC GPU with similar teraflop numbers won’t necessarily perform identically in a given game.
How the PS5’s GPU Roughly Compares to PC Graphics Cards
Because the PS5 uses a custom chip, there’s no perfect one-to-one match with a retail GPU, but its RDNA 2 architecture and compute unit count put it in a performance range loosely comparable to certain mid-range PC GPUs from that same generation. Independent coverage of the console’s architecture generally lands it somewhere in the territory of AMD’s own RDNA 2-based Radeon RX 6000 series cards, adjusted for the console’s specific configuration and optimizations. Full technical background on the console overall is also available on Wikipedia’s PlayStation 5 page.
It’s worth remembering that console game performance also benefits from fixed hardware optimization that PC games can’t always rely on, since developers tune a PS5 title specifically for this one exact hardware configuration rather than a wide range of possible PC setups.
Ray Tracing on the PS5
The RDNA 2 architecture brought hardware-accelerated ray tracing support to AMD GPUs for the first time, and that carried over into the PS5’s custom chip. This was a notable jump from the PS4 generation, which had no dedicated ray tracing hardware at all.
- Ray tracing support allows for more realistic lighting, shadows, and reflections in supported games.
- Not every PS5 game uses ray tracing, and those that do often offer a performance mode (prioritizing frame rate) versus a fidelity mode (prioritizing visual features like ray tracing) as separate options.
- Ray tracing implementation and quality still varies significantly by how a specific game’s developers choose to use the hardware.
Shared Memory: A Key Architectural Choice
The PS5’s 16GB of GDDR6 memory is shared between the CPU and GPU, rather than having separate dedicated pools like a typical gaming PC’s system RAM and VRAM. This unified approach is similar in concept to how Apple Silicon handles memory, though built for a very different purpose and platform. Sharing memory this way simplifies data transfer between the CPU and GPU, since both are drawing from the same fast pool rather than copying data back and forth between separate memory types.
PS5 vs PS5 Pro: Did the GPU Change?
Sony’s PS5 Pro revision increased GPU compute power significantly over the original PS5, while still building on the same general AMD RDNA lineage, with additional AI-based upscaling capability added for improved image reconstruction at higher resolutions. The base PS5 and PS5 Digital Edition share the same core GPU configuration described above, with the Digital Edition differing only by the removal of a disc drive, not the internal GPU specs.
Why Sony Doesn’t Call This a “Radeon RX” Chip
Since the PS5 GPU is a custom design built in partnership with AMD specifically for Sony’s console, it doesn’t carry a standard Radeon RX product name the way a retail graphics card does, such as the AMD Radeon RX 9070 or the AMD Radeon RX 7600. It shares the underlying RDNA 2 architecture family with those cards’ predecessors, but it’s a distinct chip tuned specifically for the PS5’s console hardware and software stack.
Why the Custom Approach Makes Sense for a Console
It might seem simpler for Sony to just drop a retail AMD GPU into the console, but there are real reasons console makers almost always go the custom silicon route instead.
- Combining the CPU and GPU onto a single chip (a system-on-chip, similar in principle to how Apple designs its own chips) reduces manufacturing cost and physical space compared to separate components.
- Custom power and thermal tuning lets Sony hit a specific price point and console form factor that a repurposed PC GPU wouldn’t fit as cleanly.
- Features like the variable frequency design and the specific memory configuration are tailored to how PS5 games are actually built and optimized, rather than general-purpose PC workloads.
This is also standard practice across the industry. Microsoft’s Xbox Series X and Series S consoles similarly use custom AMD silicon rather than off-the-shelf Radeon cards, following the same logic Sony applied to the PS5.
Frequently Asked Questions
1. Is the PS5 GPU the same as an AMD Radeon RX 6700?
Not exactly. It shares the same RDNA 2 architecture family and lands in a broadly similar performance territory to some RDNA 2 desktop cards, but it’s a custom chip with its own unique configuration, not a rebranded retail GPU.
2. How many teraflops does the PS5 have?
The PS5’s GPU delivers approximately 10.28 to 10.3 teraflops of peak performance, depending on the specific source and measurement method cited.
3. Does the PS5 support ray tracing?
Yes, hardware-accelerated ray tracing is supported through its RDNA 2-based GPU, though usage and quality vary by individual game.
4. Is the PS5 Pro’s GPU different from the standard PS5?
Yes, the PS5 Pro features a significantly more powerful GPU configuration with added AI upscaling support, while the base PS5 and PS5 Digital Edition share the same original GPU specs.
Console hardware specifications come from manufacturer and industry reporting and may be refined or clarified over a console’s lifecycle. Check Sony and AMD’s official statements for the most current technical details.
