Does RevoScan5 MetroX use full system resource?

Tutorials Revo Scan
1

In my experience, RevoScan5 does not use full system resources - CPU, GPU for scanning and post processing.

Can Revopoint enhance the software to use more resources on the system in near future ?

Attached an example which shows only few CPU and very low GPU computing during the post processing on RevoScan5

Low cpu gpu usage
Low cpu gpu usage1920Ă—988 145 KB

3 Likes
2

Hi, we are working on trying to GPU acceleration, please be patient, thank you.

4 Likes
3

Sounds great !
Could you hint me when I would expect to have experience the GPU computing?

4

We can’t provide a specific time yet. If we manage to make it happen, we’ll post a notice here right away. Please stay tuned.

3 Likes
5

@Revopoint-Jane Current performance of RevoScan MetroX is really terrible. Laser mode is almost not useable with high-end gaming laptop. GPU is not used at all, but in your specs high end GPU is requested (???). Feedback from R&D about solution plan and timeline is urgently needed. With current version laser mode is not usable at all. You even should proactively inform your customer about the lack of GPU support.
CR-Scan Raptor is working on same pc without any issues!

1 Like
6

Not sure why you think it’s unusable. I use laser modes just fine on both a gaming Windows laptop and a MacBook. What is the CPU in your laptop?

1 Like
7

i9-13900HX, 32GB RAM, RTX4070, SSD … not enough?

Many people are complaining about the same. It’s not just me.
When comparing laser scan experience from Raptor with MetroX on same PC it’s just like day and night.Revopoint need to come up with solution plan and timeline … asap!

8

Same specs as mine - although I have 4080 but that makes no impact here.
It’s not as smooth FPS as the Raptor but it’s far from unusable.

9

Have you tried to perform global marker registration with about 200 markers? Impossible with MetroX while no issue with Raptor.

In laser scan mode you need to move the hand extremely slow while no issue with Raptor. Even the laser scan without PC and smartphone only is working 100x better compared with MetroX and gaming laptop. For me in laser mode not usable at all. Full-field is working well.

1 Like
10

Nobody expects exact date. But given the performance problem many users are having with laser scan Revopoint need to come up with solution plan and inform customer proactively. At least some indication about month or quarter is needed. E.g. waiting half a year would be not acceptable. Please come back with some more details from R&D team.

11

Hi I have 16 core Ryzen + 64GB ram. Literally no graphics card except the one in the CPU. I get 45FPS in cross line, 12FPS in full field. I scanned a thing with 270 markers that I first scan into global marker library, than I used cross line. All of this works for me.

But @Revopoint-Jane, I have the same experience. All operations on point cloud and mesh use only 5% of my CPU (that is 1 core). Even things like batch processing, which could be done in parallel.

12

RevoScan5 Single-Thread CPU Usage Issue

RevoScan5 currently uses CPU resources very inefficiently: it runs almost entirely on a single thread while leaving other cores idle. For example, on a Xeon E5-2680 system (28 logical threads), the software fully loads only one core and leaves the rest underutilized. Users have reported that during scanning or post-processing, overall CPU and GPU usage stays very low (below ~17%, noting that “RevoScan5 does not use full system resources – CPU, GPU”. This single-threaded behavior causes that one core to overheat locally, while the remaining cores and GPU sit idle.

Actions Taken

  • CPU Affinity Adjustment: We experimented with setting the process affinity (using PowerShell scripts) to allow RevoScan5 to migrate its workload across cores. However, because RevoScan5 is internally single-threaded, this merely shifts the same workload from one core to another without any parallelism. In practice, it did not improve overall throughput or reduce thermal stress on the busy core.
  • GPU Usage Verification: We checked that RevoScan5 does not make significant use of the GPU and provides no user options for GPU acceleration. As noted by users, the application “does not use full system resources – GPU”, with the GPU remaining essentially idle during scanning and processing.

Benefits of Multithreading Support

  • Higher Throughput: A multithreaded implementation would split computational tasks (such as point cloud registration or depth-map fusion) among multiple cores, greatly reducing total processing time. Instead of saturating one core, the workload would be spread out, allowing the program to complete scans and mesh generation much faster by leveraging many CPU threads in parallel.
  • Improved Thermal Efficiency: Modern CPUs use dynamic overclocking (Turbo Boost) that is more effective when multiple cores are idle. Intel explicitly notes that the achievable turbo frequency depends on the workload and available thermal headroom, and that Turbo Boost raises performance in both single-threaded and multithreaded workloads. In practice, this means that if several cores share the load, the cooler idle cores allow the heated cores to boost to higher clock speeds for short intervals before throttling. Conversely, if only one core is active, it heats up quickly and throttles at a lower frequency. For example, Intel Turbo Boost Max Technology 3.0 increases the frequency of the fastest cores when only one or a few cores are active (benefiting single-thread bursts). Overall, enabling multithreading would let RevoScan5 run faster and more consistently: the workload would be distributed, and the CPU could exploit dynamic frequency scaling without compromising thermal stability.

In summary, adding multithreading support to RevoScan5 would significantly boost performance and leverage modern CPU features. Distributing the computation across multiple cores would not only accelerate the software but also improve thermal headroom, as unburdened cores would remain cool while busy cores can turbocharge, resulting in faster processing without overheating.

For the development team’s reference, the scanner in use is a Revopoint MINI 2, and the system is equipped with an NVIDIA GTX 1660 Ti OC graphics card. This ensures that both the capture hardware and the available graphics processing power exceed the minimum requirements, so any performance limitations observed do not appear to stem from hardware bottlenecks, but rather from how the software is currently managing system resources.

image
image832Ă—531 32.2 KB