Throughout Digital Foundry’s current Cyberpunk 2077 Swap 2 vs Steam Deck face-off, we used inside benchmarks hidden throughout the recreation to check CPU and GPU efficiency with some fascinating outcomes – however these take a look at sequences truly run on any model of the sport on the market. In reality, we would warrant that they’d even current on the Stadia model if it have been nonetheless about. Initially, the plan was to dump our PS4 and PS4 Professional knowledge on DF Clips – however we had a re-think in gentle of an interesting interview by Wccftech with Virtuos satellite tv for pc studio Black Shamrock.
It is uncommon for a developer to candidly discuss system efficiency, not to mention examine one console instantly towards one other, however feedback from Black Shamrock’s Eoin O’Grady have been intriguing, kicking off together with his ideas on graphics efficiency: “GPU-wise, the Swap 2 performs barely beneath the Collection S; this distinction is extra noticeable in handheld mode. Nevertheless, the Collection S doesn’t help applied sciences like DLSS, which the Swap 2 does. This makes the GPU capabilities of the 2 consoles comparable total.”
Judging graphics efficiency is comparatively easy – we have been doing this for some time now, nevertheless it was O’Grady’s feedback on CPU capabilities that me extra: “CPU-wise, there’s a clearer distinction between the 2 consoles. The Swap 2 is nearer to the PlayStation 4 on this respect, having a CPU only a bit extra highly effective than the PS4’s. Since most video games are usually extra GPU-bound than CPU-bound when properly optimised, the impression of this distinction largely will depend on the particular recreation and its goal body charge. Any recreation delivery at 60fps on the Collection S ought to simply port to the Swap 2. Likewise, a 30fps Collection S recreation that’s GPU-bound must also port properly. Video games with complicated physics, animations, or different CPU-intensive parts may incur further challenges in reaching 30 or 60fps or require additional optimisation throughout porting.”
On the one hand, the ARM Cortex A78C utilized in Swap 2 is clearly fairly potent in comparison with Swap 1, however then again with ballpark clocks of simply 1GHz to 1.1GHz relying on the mode, uncooked efficiency shall be skinny on the bottom. Cyberpunk 2077 is without doubt one of the most “benchable” video games we now have, so I used to be curious to see if that title’s outputs matched Black Shamrock’s estimations.
Properly, this is only one recreation, keep in mind. A primary era recreation, no much less, and there are key variations in the way in which that one recreation has been tailored for the varied programs, every of which have their very own key variations like distinctive graphics APIs. We’re shifting into “a little bit of enjoyable” territory right here as a result of clearly every system may have totally different set-ups, totally different upscalers and extra on the software program aspect, whereas the variations on a {hardware} degree between in the present day’s Swap 2, PS4/Professional and Xbox Collection S shall be huge.
I kicked off by testing CD Projekt RED‘s official benchmark – which just like the hidden efficiency checks, may be run on any console. At 30fps, all programs produce nice outcomes – only a few dropped frames on PlayStation 4 and PS4 Professional. In the meantime, Collection S and Swap 2 have larger frame-rate efficiency modes, which run virtually completely at 40fps and 60fps respectively. I additionally examined our oft-used Kabuki streaming run on all programs – a quick race by way of town, which taxes graphics and CPU (intimately wealthy areas, at the least). It seems that efficiency is once more fairly good, that means the main focus must shift to analysing dynamic decision scaling. My colleague, Tom Morgan, labored onerous to seek out “countable” pictures frequent to all programs – not straightforward with TAA, DLSS and FSR2 at play, relying on the system.
You may see the blow-by-blow leads to the video however in abstract, there is a respectable diploma of variability right here, however the Swap 2’s docked GPU efficiency up towards PlayStation 4 – the bottom PS4 – must be highlighted. On all of our checks, the Swap 2 delivers larger pixel counts and it is doing so with DLSS anti-aliasing and upscaling vs a inexpensive TAA resolution on the outdated Sony console. Throughout all of our take a look at outcomes, Swap 2 is delivering a median 32 % enchancment in decision. Up towards PS4 Professional although, Swap 2 is both on par decision clever or significantly decrease, a gulf in outcomes that makes a median pointless. In the end although, I Swap 2 falls between PS4 and Professional although ideally I would like extra knowledge right here – and a few thought of how a lot GPU load DLSS is accounting for.
In opposition to Xbox Collection S, decision variations fluctuate from nothing to upwards of 70 % in favour of the Collection S – the bigger numbers might be all the way down to the truth that we regularly hit the max 1080p on Swap 2, so there may very well be overhead past that we simply aren’t seeing. Wanting on the efficiency mode differentials, there’s a large gulf: a median 73 % decision benefit for Collection S which is concentrating on 60fps vs Swap 2’s 40fps.
So primarily based on Cyberpunk 2077 at the least then, it does really feel like a stretch to say that Swap 2 GPU efficiency is up there with Collection S, making me wonder if I would misunderstood one thing within the Virtuos interview, so I contacted them and that is the response I acquired:
“To make clear, when Eoin talked about that 60fps video games on the Collection S ought to ‘port simply’ to the Swap 2, he was referring to the relative ease of adapting video games to the platform somewhat than guaranteeing similar efficiency at 60fps. Precise body charges will naturally rely upon the particular recreation, its optimisation, and its developer’s objectives. As he acknowledged through the interview, “Video games with complicated physics, animations, or different CPU-intensive parts may incur further challenges in reaching 30 or 60fps or require additional optimisation throughout porting.”
Transferring onto the CPU aspect of the testing, it is a lot tougher to measure processor efficiency than it’s graphics as efficiency is restricted both by graphics or by the frame-rate cap imposed by the developer. Nevertheless, Cyberpunk 2077 is barely totally different in that traversing at velocity by way of town does counsel CPU bottlenecks – and we all know that they are CPU in nature as you see the identical areas improve CPU load within the PC model of the sport.
In Cyberpunk 2077, 30fps streaming checks clearly present Swap 2 a great chunk forward of PlayStation 4 however not fairly as quick as PS4 Professional – the place its Jaguar CPU cores featured a circa 30 % improve to clock speeds. Swap 2 could also be under-represented although to a sure diploma because it’s clearly streaming in additional element, the outdated Sony consoles bottlenecked by their 5400rpm onerous drives.
It is no shock to see Collection S exhibiting zero points in anyway dealing with Cyberpunk 2077 on the 30fps restrict – the Zen 2 cores in current-gen consoles are over twice as quick because the Jaguars in last-gen, so Virtuos’s feedback are confirmed out. In efficiency mode with a capped 40fps, Swap 2 finds its limits, with apparent frame-rate drops in heavier areas, whereas Collection S powers by way of at 60fps, with simply occasional drops. It is a class aside.
So, what have we realized? I believe Virtuos/Black Shamrock’s feedback concerning the CPU efficiency from Swap 2 are born out – ballpark last-gen efficiency is smart in a useful resource constrained gadget operating at low clock speeds. The GPU aspect of issues sees Swap 2 carry out very properly for a system within the 20W ballpark, pulling forward of PS4 and even moving into PS4 Professional territory in some instances – all of the whereas operating the computationally heavy DLSS upscaler.
Nevertheless, in Cyberpunk 2077 at the least, Collection S pulls forward right here convincingly, pushing many, many extra pixels. DLSS is clearly enjoying a major position although. Cyberpunk seems spectacular however I believe that Avenue Fighter 6 is probably the most fascinating instance – it is delivering arguably higher picture high quality than Collection S utilizing simply 25 per cent of the uncooked pixel depend. On this sense, Swap 2 and Collection S are extremely comparable, however maybe much less so by way of uncooked pixel throughput. In the end, Collection S is pulling round 4 instances as a lot energy from the wall!
I believe the general message from the Virtuos interview is basically concerning the viability of current-gen ports – a 60fps Collection S recreation must be simply moveable to Swap 2 – although not essentially on the similar frame-rate. Past that, whereas graphics are inherently scalable, CPU-heavy present era titles might have extra work. All eyes are on the following difficult port set for Swap 2: we will not wait to take a look at Star Wars Outlaws.
