Pages Navigation Menu

SHOWFUN - Show & Fun & More!

Remedy may not be able to fix all issues with Quantum Break

Last week, we covered how Remedy’s Quantum Break is the latest DirectX 12 title to run into problems on the PC. Despite a strong debut for its console, its PC version struggles, particularly on Nvidia hardware. Tweets from the game’s PR representative seem to suggest that at least some of the game’s problems aren’t really something that can be fixed.

When asked whether visual artifacts, ghosting, soft textures, screen-tearing, and texture pop-in could be remedied with a patch, Thomas Puha, PR head at Remedy, stated that “Ghosting is just a result of the temporal reconstruction. Just the way we render things.” Texture pop-in also got called out as an engine-specific issue by Puha, who wrote that it “unfortunately is a quirk of our graphics engine.” Also, there’s not going to be a way to turn film grain off, since that’s another locked-on attribute.

Whether or not these are major issues for most people, however, is open to considerable debate. I don’t like texture pop-in all that much, but so long as it’s not constant and game-breakingly egregious, it doesn’t wreck games. The Mass Effect titles often suffered from a few seconds of texture pop-in when you entered an area, and it was never more than briefly distracting. The bigger questions are whether Remedy can patch up the performance issues and bypass the limitations that are holding the game back on UWP and DirectX 12.

Multiple publications, including Eurogamer and PC Gamer, have noted that Quantum Break doesn’t perform well on the GTX 970 as compared with AMD hardware. Overclock3D ran some benchmarks on the game comparing the GTX 960 against the R9 380 as well as the GTX 980 Ti against the AMD Fury X. Oddly, where Eurogamer reported max frame rates topping out at 5/6 of refresh rate, OC3D managed around 91% of refresh rate — but never a flat 60 FPS, and always with some 1% and 0.1% frame times at or below 30 FPS. Why a $600 GPU can’t manage a uniform 60 FPS at 1080p is a mystery at this point.

The GTX 960 and AMD R9 380 are generally tied at 1080p at the “Lowest” and “Low” detail levels, but a huge gap opens up at 1080p “Medium,” where the R9 380’s average frame rate is 27.5% faster than the GTX 960.

The Fury X and the GTX 980 Ti tie each at 1080p until Max detail, where it’s the Fury X’s frame rate that collapses to 39 FPS, while the GTX 980 Ti holds up at 50 FPS. Overclock 3D’s 1440p and 4K tests show the same trend: At 1080p, the Fury X ties the GTX 980 Ti until you hit “Max” detail. At 1440p, the two cards tie until “Ultra” detail. At 4K, the Fury X is starting to slip behind the GeForce GTX 980 Ti, even at “Medium” detail.

These results collectively suggest that Quantum Break is extremely RAM hungry on the PC. The GTX 960 is a 2GB card while the R9 380 has 4GB of RAM, which would explain why the AMD card can maintain an excellent frame rate against its NV counterpart. The AMD Fury X is a 4GB GPU against the GeForce GTX 980 Ti’s 6GB, which again, would explain why the Fury X begins fading at the highest detail levels.

As for why the R9 390 would outperform the GTX 970 by 50%, it’s possible that Quantum Break is barking its shins on the GTX 970’s 3.5GB+512MB memory configuration, but the video playback from Eurogamer’s side-by-side comparison doesn’t really show that happening.

Typically when a GPU runs out of memory, there’s a brief catastrophic frame rate drop down into the single-digits. Instead, QB seems to maintain a steady (albeit slow) frame rate. There have been a handful of titles that could run into trouble with the GTX 970’s memory buffer allocation, but we’ve never seen it happen at 1080p. I’m fairly certain that the performance differentials in the OC3D results are RAM-based, but there’s no firm explanation for the discrepancy between the GTX 970 and the R9 390.

Hopefully these teething problems will be resolved in a few more months, and while the issues are annoying, they aren’t without precedent. Early DX10 titles struggled to show much benefit when running in that mode and even top-end cards of the day often ran better in DX9 than DX10. Some of you likely remember the user-created patches for games like Crysis that tried to backport some of the game’s DX10 visuals in DX9. The one difference between those todays and today is that back then, early DX10 titles almost all shipped with a DX9 fallback. The DX12 implementation in UWP games doesn’t have a DX11 fallback — you’re stuck with whatever the engine hands you in this case.

Leave a Comment

Captcha image