Puget Bench on ARM: A Hardware Desk Guide to Snapdragon and NVIDIA RTX Spark™ Test Timeouts, Beta Builds and Host App Gaps
By Sarah Chen — MyTechDailyNews Hardware desk
Before You Publish an ARM Score, Separate the Silicon From the Software Stack
A benchmark figure printed next to a laptop model is only as trustworthy as the applications underneath it. That point has become impossible to dodge in the MyTechDailyNews Hardware lab: following the recent launch of the NVIDIA RTX Spark™, the questions that arrived about ARM compatibility all collapsed into one concern — what actually runs, and what silently distorts a result. The most useful thing a hardware reader can take from this report is that the measurement tool is no longer the limiting factor. The host applications, and the exact version you install, are.
What follows is a one-stop compatibility reference for Snapdragon and RTX Spark users, describing the state of play as of October 2026 from both sides of the equation: the benchmark side and the host application side. It covers the latest release and the relevant beta versions, because developers are still working toward ARM compatibility and feature parity with x86 systems. As those updates expand the working application-and-benchmark combinations, individual lines below may become outdated, so read them as a dated snapshot rather than a permanent verdict.
The Benchmark Layer Is Already ARM-Native
Begin with the settled part of the picture. The launcher and scripts have been fully ARM-native and compatible since Puget Bench for Creators 1.4.30, and the latest 1.4.50 version has no known ARM problems. When a run fails or behaves oddly on an ARM device, the burden of explanation therefore shifts to the host application. The tested host applications, including Photoshop and Premiere, do not always offer complete ARM support, so missing features or instability can affect benchmark results. In practical terms, an unsupported codec path or an untestable effect does not drag a score down — it removes the workload that would have generated the score in the first place, which is exactly the trap to avoid when comparing an ARM unit against an x86 desktop.
Timeout Settings: The First Adjustment for Lower- and Mid-Tier ARM Hardware
Performance tiering shows up here immediately. Many ARM devices sold today are considered lower- or mid-tier in performance compared with desktop systems, and a manual timeout adjustment may be needed. If a benchmark reports that the timeout was reached, change the setting as follows:
Select the Settings icon in the lower-left corner of the Puget Bench launcher.
Raise the benchmark timeout from the 300-second default to 600 seconds or higher.
Click “Close”; the changed settings save automatically.
Treat that dial as a hardware signal rather than a nuisance. A machine that needs the ceiling lifted is telling you that individual tests run slowly on it, and two ARM units compared with different allowances are not really being compared at all. Record the timeout value you used in your published methodology alongside the score.
Host Application Compatibility, One Suite at a Time
Adobe Photoshop: the clean baseline of the group
Adobe Photoshop has had an ARM-native version for many years and has generally worked well on Windows ARM devices. At present, neither the release version v27.11 nor the beta version v28.0-beta has known issues. For anyone building a test plan, that makes Photoshop the control run: the suite you use to confirm the platform is behaving before moving on to the harder titles.
| Host application | Puget Bench version | Snapdragon | RTX Spark |
|---|---|---|---|
| Adobe Photoshop v27.11 | Puget Bench for Photoshop v1.0.7 | ✅ No Issues | ✅ No Issues |
| Adobe Photoshop v28.0-beta | Puget Bench for Photoshop v1.0.7 | ✅ No Issues | ✅ No Issues |
Adobe Lightroom Classic: emulated, but still usable
Adobe Lightroom Classic does not currently have an ARM-native version and runs in emulated mode instead. That nuance matters more than the pass or fail mark. Results on ARM devices can reflect what end users would experience in Lightroom Classic, but the application is not ideal for demonstrating ARM-based system capability. Compatibility is still very good overall. Some ARM-based systems may time out on certain tests, which can be handled by increasing the benchmark timeout, and performance may improve later if an ARM-native version is released.
| Host application | Puget Bench version | Snapdragon | RTX Spark |
|---|---|---|---|
| Adobe Lightroom Classic v15.6 | Puget Bench for Lightroom Classic v1.1.1 | ⚠️ The Extended preset may require a longer benchmark timeout because some AI tests run relatively slowly | ✅ No Issues |
Adobe After Effects: the beta build is the only full path today
Adobe After Effects is available as an ARM-native build, and related integrations such as Cinema 4D are compatible as of v26.5. The catch sits in the release branch: a bug in the current v26.5 release affects compositions that use the Advanced 3D rendering engine, as exposed by the “3D Camera Explode” benchmark test. Because of that bug, the Extended benchmark preset does not currently work on any Windows-based device, including ARM. The problem is fixed in the After Effects beta, version 27.1.0.021, released 10/21/2026. Beta builds can be difficult for testing because they are works in progress and performance may change, but the After Effects beta is currently the best way to benchmark ARM devices. Some features that are unavailable in the ARM version of After Effects are not tested by Puget Bench; additional details are available on Adobe’s support page for ARM-based Windows devices.
| Host application | Puget Bench version | Snapdragon | RTX Spark |
|---|---|---|---|
| Adobe After Effects v26.5 | Puget Bench for After Effects v1.1.1 | ⚠️ Standard preset only because of a Windows bug | ⛔ Does not detect the RTX Spark GPU |
| Adobe After Effects v27.1.0.021-beta | Puget Bench for After Effects v1.2.0-beta | ✅ No Issues | ✅ No Issues |
Adobe Premiere: RAW codec coverage decides everything
The current Adobe Premiere release, v26.5, is ARM-native but lacks RAW codec support on ARM devices. The latest beta, v27.1-beta, improves this situation and supports every RAW codec tested in Puget Bench except Sony X-OCN. As a result, the current release cannot run Puget Bench for Premiere Pro, and the beta is required to run only the Standard benchmark preset. If your purchase decision hinges on camera-native RAW workflows, this is the suite where the ARM platform still shows the clearest gap. Some features that are unavailable in the ARM version of Premiere are not tested by Puget Bench, and Adobe’s support page for ARM-based Windows devices carries the additional details.
| Host application | Puget Bench version | Snapdragon | RTX Spark |
|---|---|---|---|
| Adobe Premiere v26.5 | Puget Bench for Premiere Pro v2.0.2 | ⛔ No support for RAW codecs on ARM | ⛔ Does not detect the RTX Spark GPU; ⛔ No support for RAW codecs on ARM |
| Adobe Premiere v27.1-beta | Puget Bench for Premiere Pro v2.1.0-beta | ⚠️ Standard preset only because Sony X-OCN is not supported | ⚠️ Standard preset only because Sony X-OCN is not supported |
Blackmagic DaVinci Resolve Studio: parity with a timeout caveat
DaVinci Resolve Studio has supported ARM for several years and generally maintains good feature parity with x86. The limitation here is workload difficulty rather than missing code. Mid- and low-performance platforms often struggle with the difficulty of the Extended tests, and unified memory architectures such as Snapdragon and RTX Spark may also be slow, but they can usually complete the Extended tests when the default benchmark timeout is increased.
| Host application | Puget Bench version | Snapdragon | RTX Spark |
|---|---|---|---|
| DaVinci Resolve Studio v21.1.1 | Puget Bench for DaVinci Resolve v2.1.1 | ⚠️ The benchmark timeout will likely need to be increased, especially for the Extended preset, because some tests run relatively slowly | ✅ No issues |
Consolidated Scorecard: Build to Install, by Suite
| Suite | Application version | Snapdragon | RTX Spark | Build to install |
|---|---|---|---|---|
| Puget Bench for Photoshop v1.0.7 | Adobe Photoshop v27.11 | ✅ No Issues | ✅ No Issues | Release or v28.0-beta |
| Puget Bench for Lightroom Classic v1.1.1 | Adobe Lightroom Classic v15.6 | ⚠️ Longer timeout for Extended | ✅ No Issues | Emulated release (no ARM-native version) |
| Puget Bench for After Effects v1.1.1 | Adobe After Effects v26.5 | ⚠️ Standard preset only | ⛔ No RTX Spark GPU detection | Upgrade to beta 27.1.0.021 for full coverage |
| Puget Bench for After Effects v1.2.0-beta | Adobe After Effects v27.1.0.021-beta | ✅ No Issues | ✅ No Issues | Beta required |
| Puget Bench for Premiere Pro v2.0.2 | Adobe Premiere v26.5 | ⛔ No ARM RAW codecs | ⛔ No GPU detection, no ARM RAW codecs | Release cannot run the suite |
| Puget Bench for Premiere Pro v2.1.0-beta | Adobe Premiere v27.1-beta | ⚠️ Standard preset only | ⚠️ Standard preset only | Beta required; Sony X-OCN still unsupported |
| Puget Bench for DaVinci Resolve v2.1.1 | DaVinci Resolve Studio v21.1.1 | ⚠️ Increase timeout | ✅ No issues | Release build |
Common Misconceptions About ARM Benchmark Results
“An incomplete or failed run means the benchmark is not ARM-ready.” It is: the launcher and scripts have been ARM-native since Puget Bench for Creators 1.4.30, and 1.4.50 has no known ARM problems.
“Extended preset numbers are available on every suite and every machine.” They are not — the After Effects Extended preset currently does not work on any Windows-based device, ARM included.
“Any recent Premiere build can run Puget Bench for Premiere Pro on ARM.” The v26.5 release cannot, because ARM lacks the RAW codec support the suite exercises.
“A Lightroom Classic number measures ARM hardware capability.” Lightroom Classic still runs in emulated mode, so the result reflects the end-user experience rather than native ARM potential.
“Beta software is automatically off-limits for published testing.” Here the After Effects beta is the only way to reach full benchmark coverage; the honest compromise is to disclose the build number.
“A timeout is a test failure.” It is a setting, and raising it from the 300-second default to 600 seconds or higher is the documented remedy.
Action Recommendations for Test Labs and Buyers
Puget Bench for Photoshop: no known issues — run it first as the platform control.
Puget Bench for After Effects: the latest After Effects beta is required for full benchmark support.
Puget Bench for Premiere Pro: the latest Adobe Premiere beta is required for RAW support and RTX Spark GPU support; use the Standard preset only.
Puget Bench for Lightroom Classic: budget a longer benchmark timeout for the Extended preset on Snapdragon.
Puget Bench for DaVinci Resolve: timeouts may need to be increased, especially on mid- and low-tier units.
Log the suite version, the application version and build number, the October 2026 snapshot date, and the timeout value in every published methodology.
Where the Numbers Come From
Every status mark above is tied to a named version pair and a named platform, and Puget Bench results can be compared against thousands of systems in the benchmark database. The benchmark side is confirmed to run on both Snapdragon and NVIDIA RTX Spark devices, while the remaining caveats are documented by the application vendors — Adobe’s support page for ARM-based Windows devices lists the features absent from the ARM builds of Premiere and After Effects. When you cite this report, treat “No Issues” as a statement about that version combination on that date, not as a general claim about the product.
The Takeaway for Hardware Readers
ARM has been available for several years, and Adobe and Blackmagic continue to work toward feature parity with x86 and better overall performance. Most of the friction a reviewer or purchaser will hit comes from missing support for specific features that the benchmarks exercise, not from the platform itself, and those limitations should improve over time. For anyone shortlisting Snapdragon or RTX Spark machines for creative work, the practical read is simple: Photoshop and DaVinci Resolve Studio tell the truth today, Lightroom Classic tells it slowly and in emulation, and Premiere and After Effects only become fully measurable once you move to the current beta builds — with the timeout ceiling raised to match.