These are one writer's hands-on results, not lab benchmarks. No frame-time graphs or standardised test route were published. Treat the numbers as indicative.
The short version
- Steam Deck: inconsistent performance, with no fix found yet.
- ROG Ally (Z1 Extreme): a black screen on launch, then 60+ fps at 1080p once the refresh rate is lowered.
- Intel handhelds (MSI Claw, OneXPlayer 3): comfortably above 60 fps, with a few quirks.
- Frame limiter: an external limiter such as RTSS is recommended on the Intel handhelds.
What the PC build offers
The settings menu is sparse. Options cover window mode (fullscreen, borderless, windowed), resolution (4K, 1440p, 1080p, 720p), V-sync, a frame cap from 30 fps to unlimited, anti-aliasing and brightness. The anti-aliasing choices are FXAA, SMAA and a third option that is currently bugged.
Two limitations matter for handheld owners:
- At 16:10 or 21:9, gameplay can be cropped, and the only fix the writer found is forcing one of the listed resolutions.
- There are no performance-oriented options such as upscaling or lighting and shadow controls. If a handheld struggles, there is little to turn down.
According to the report, the demo and full game expose the same options.
Steam Deck: looks good, plays unevenly
The writer's verdict on Steam Deck is blunt. The game is zoomed in on first launch, so the left and right edges are cropped. The reported fix is to force the Deck to run the game at 720p or 1080p, then adjust the resolution in-game. That fixes the framing, not the performance.
On performance, he went through a lot of configurations:
- He tested both Steam Decks, one on stable SteamOS and one on the Preview channel, plus 15 Proton versions. These included 4 GE builds and Proton Experimental.
- None of it helped. Performance stayed inconsistent, and the game seemed not to be using the GPU much.
- Forcing GPU clock speeds helped a little at the 1000 MHz setting, but not enough to make the game feel good.
His recommendation is to wait for patches, because he can't tolerate random hitching in a side-scroller, even at 30 or 40 fps. That is a personal judgment, and it shouldn't be read as "unplayable on every Deck". Other outlets' reports from the demo period point the same way. One community guide site (castlevania-guide.wiki) summarised Deck owners' demo complaints as frame rate dips, a resolution ceiling, and a frame limiter that does not do what its label implies. That site is a community aggregator, not an official source. As of its 5 October note, it said Steam had assigned no Deck compatibility rating.
The SteamOS-versus-Windows split is notable. The Proton and GPU-clock testing makes a simple "wrong Proton version" explanation less likely. It does not show where the bottleneck is. Only a patch from Konami and developer Evil Empire would settle that, and the writer says he will update his piece if one arrives.
ROG Ally: black screen first, then fast
On Windows handhelds, the writer reports the game is unplayable on ROG Ally out of the box, because it launches to a black screen.
The refresh-rate workaround
With the fix applied, it runs very well. The workaround came from a Steam Community thread:
- Open the Ally's quick access menu or Armoury Crate.
- Set the display refresh rate to 120 Hz or lower.
- Launch the game.
The writer then reports 65–110 fps in the areas he tested at 25 W and 60+ fps at 1080p on a Z1 Extreme, unplugged. He used SMAA and saw one two-second freeze during an area transition. Frame rate dipped into the high 50s only where the camera zooms out a lot while you still have control. He says FXAA holds up better there.
It's a PC-wide problem, and the fix is not universal
The black screen is not Ally-specific. A Steam demo thread started October 1 shows desktop users hitting the same problem. One user reported that dropping a 240 Hz monitor to 120 Hz or lower fixed it. Others had different results:
- One user said changing the refresh rate didn't work for them.
- Another said turning off AMD Enhanced Sync and turning on vertical refresh worked on their AMD card.
- Turning on V-sync helped another user.
- Quitting and relaunching worked for one.
- One user deleted the demo's
PLAYERPREFS.savfile fromAppData\LocalLow\KONAMI\Castlevania Belmont's Curse.
A separate Steam thread describes a framerate-cap trick. One user capped the game's frame rate to 120 through RTSS so it would boot, then raised the cap again. Another thread suggests setting a 60 fps maximum in the Nvidia app. These are anecdotes from the demo, not official guidance.
The common thread is a launch-time interaction between high refresh rates, sync settings and the frame cap. That is an inference from the pattern of reports, not a diagnosis from Konami.
If you try the file-deletion fix, back up the folder first. Konami says demo save data can carry over to the full game, so a careless delete could cost progress. The demo and full game also have to be on the same platform and account, the demo save must be kept, and the full game must be updated to the latest version for the transfer.
Intel handhelds: MSI Claw and OneXPlayer 3
The writer says Intel hardware had no trouble. He saw 60+ fps at 1080p, and:
- MSI Claw 7 and 8 AI+: both use the same chip and TDP options. He usually saw 90–120 fps at the default 30 W maximum, and more than 60 fps at 17 W.
- OneXPlayer 3: performs even better. At 17 W it slightly beat the Claw 7 AI+ at 30 W. At its 35 W maximum it held an almost locked 144 fps, matching the display. Uncapped, it reached 180 fps in rooms where the Claw topped out around 90–100.
The writer also compares power. He says 30 W on the Claw should be similar in practice to 25 W on the Ally, because this Intel generation counts RAM power in the chip's TDP. That is his own contextual comparison, not a controlled whole-device power measurement.
The TDP quirk
On the Claw, some rooms settled at a stable 90 fps. Lowering the TDP and raising it again lifted performance, usually to around 100 fps and once to 120 fps, which was his cap and the display maximum. Leaving and re-entering a room reset it. He couldn't tell whether MSI software or his own custom software was responsible. The OneXPlayer showed none of this.
Use RTSS, not the in-game limiter
The most transferable advice is about frame limiting. The writer says the game's internal limiter hurt frame times on the handhelds he tested. Capping with RTSS kept things much smoother across both. This is his handheld testing, not developer guidance. On a game this reliant on tight timing, an external cap is easy to try. If the game starts to feel off at a high refresh rate, test it.
Which version to buy on a Windows handheld
The writer prefers the Xbox version on ROG Ally, based on the demo. His reason is that it supports Xbox Play Anywhere, so he can play on an Xbox Series X and continue on the Ally. That is a platform preference, not a measured comparison. All his performance data comes from the Steam build, so nothing here shows the Xbox PC version performs differently.
What to do now
- Steam Deck owners: if you've pre-ordered, wait for a patch or hold off. If you try it, force 720p or 1080p at the system level to fix the cropping.
- ROG Ally owners: set the refresh rate to 120 Hz or lower before first launch. If that fails, try V-sync and a relaunch before touching any files.
- Claw and OneXPlayer owners: use RTSS to cap the frame rate. If a room sits at 90 fps, a TDP toggle might help, but expect it to reset.
- Anyone with a black screen: try the community fixes one at a time. They are anecdotal and don't work for everyone.
Bottom line
The report's main finding is that Windows handhelds and Intel-based devices handle the game well, while Steam Deck does not. It is one writer's testing before launch, and a day-one patch could change it. The black-screen workaround is reasonable and low-risk, but community-sourced, so don't count on it working for you.