A few days back, Microsoft jumped the gun and decided to unveil the detailed hardware specifications of its next-gen Xbox Series X console. This forced Sony to do the same in the next couple of days. However, there was a stark difference between the hardware reveal of the two consoles. Sony’s Mark Cerny relied on up-to and variable a lot more than Phil Spencer and the upcoming Xbox Series X.
Yes, this is with respect to the variable boost clock on the Sony PS5, and its peak 10.3 TFLOPs of rated FP32 compute performance. Many are calling it a stroke of genius and Cerny himself claimed that the performance would scale linearly with performance. However, that’s not quite the case.
Before we get to that, let’s have a look at how AMD’s existing Navi cards scale when overclocked. For starters, they’re not very conducive to overclocking in general. Second, they use a more sophisticated boost algorithm than GCN and Vega (similar to NVIDIA’s GPU Boost 3.0). The GPU doesn’t quite hit the maximum advertised boost clocks and in most scenarios, you’ll be stuck a notch lower due to power or thermal restrictions, a range AMD calls Game Boost Clock.
Thirdly, and something that you should really pay attention to is how much performance you can gain with Navi upon overclocking. Bumping the clocks up to 2.1 GHz on the RX 5700–an 18 percent boost over stock, generally yields just 5-10 percent higher performance. This is due to various reasons including a sensitive boosting behavior. Let’s have a look at the specs of the two consoles side-by-side now:
|Xbox Series X||Sony PS5|
|CPU||8 Core Zen 2||8 Core Zen 2|
|CPU Clock speed||3.8 GHz (3.66 with SMT)||3.5 GHz|
|RDNA 2 GPU||52 CUs / 3,328 Shaders||36 CUs / 2,304 Shader|
|GPU Clock||1.825 GHz (Constant)||2.23 GHz (Peak)|
|Performance||12.1 TFLOPs FP32||Up to 10.3 TFLOPs|
|Sustained GPU Clock Speed||1.825 GHz (Constant)||2.23 GHz (Peak)|
|Memory||16GB w/mixed 320/192-bit bus||16 GB GDDR6 256-bit bus|
|Memory Bandwidth||10GB 560 GB/s for 10GB/ 336 GB/s for rem. 6GB||448 GB/s|
|Internal Storage||1 TB Custom NVME SSD||825 GB Custom SSD|
|I/O Throughput||2.4 GB/s (Raw) / 4.8GB/s (Compressed)||5.5 GB/s (Raw)|
Both consoles feature an octa-core Zen 2 CPUs, but the Xbox variant has a faster clock, even with SMT enabled. The PS5, on the other hand, runs 300MHz slower at 3.5GHz and there’s no mention of SMT. However, that’s not the main problem here, the GPU’s boosting behavior is.
The Xbox One X features a 52 Compute Unit RDNA 2 GPU running at a sustained boost clock of 1.825GHz. The PS5, on the other hand, will be powered by a 36 CU GPU with a variable boost clock of up to 2.23GHz. The keyword here is variable, and the up-to 10.3 TFLOPs in the PS5’s rated performance figure.
I believe that the 2.23GHz clock here is the same as the advertised boost clock on the Navi based RX 5700 series GPUs. While the graphics processor can reach these clocks, it won’t in most scenarios. Only when you’re running a fairly less intensive game, something that draws lesser power (than average) will the GPU boost to the peak boost clock of 2.23GHz and produce the resulting 10.3TFLOPs of performance.
In most cases, the in-game GPU clock will be lower (similar to the Game Boost of the Navi 10 PC GPUs), somewhere around the 2GHz mark and the actual performance rating will be between 8-9 TFLOPs. That’s nearly 4 TFLOPs lower than the Xbox Series X.
Keep in mind that I’m not criticizing Sony’s choice of hardware. The PS5 may very well run on par with the XSX or more likely, it’ll be slightly slower. The problem here is the way they marketed it. It’s misleading and paints a more colorful picture of the hardware powering the console. That’s not fine with us.