INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: JRiver on Surface pro 11 ARM64 Qualcomm Snapdragon X . Small test report  (Read 762 times)

abrise

  • Junior Woodchuck
  • **
  • Posts: 89

JRiver on Surface pro 11 ARM64 small report

2024 08 13

It seems that two months after the release of the Surface Pro , nobody has posted his experience with JRiver
I have just received mine less than a week ago.
I have purchased the cheapest model with the snapdragon X and only 256GB ssd. I have immediately replace the ssd with a crucial 2TB for only 170€. It is also twice faster than the original SSD.
The new Surface is to replace my old Surface Pro 2017 which was recently upgraded with a new battery and although not supposed to be compatible, Windows 11 . This old Surface Pro was still giving me satisfaction. I am using it only  during traveling so 4 to 5 months a year and a number of international trips. The Surface Pro takes the lowest real estate on an economy class seat tray of all computers and I do need a computer and not an android tablet as android doesn't run JRiver media center .

JRMark

For the battery I have put best performance settings for the JRMark , for all the other tests recommended settings will be used.
Synthesis
Higher is better
      Surface pro 11   Surface pro 2017   Intel I5 13500
Math            3157           1444                           4107
Image       2795             3786                           8203
Database    5597           4510                           7943
Total            3849           3246                           6751
This is quite good results , we don't feel at all the X64 to ARM emulation here . As you can see below in the details the only bad result is in image creation / destruction. I have no idea of the implication of this.

Details

Surface pro 11 2024
Running 'Math' benchmark...
    Single-threaded integer math... 2,423 seconds
    Single-threaded floating point math... 2,496 seconds
    Multi-threaded integer math... 0,534 seconds
    Multi-threaded mixed math... 0,567 seconds
Score: 3157

Running 'Image' benchmark...
    Image creation / destruction... 5,656 seconds
    Flood filling... 0,317 seconds
    Direct copying... 0,249 seconds
    Small renders... 0,578 seconds
    Bilinear rendering... 0,241 seconds
    Bicubic rendering... 0,830 seconds
Score: 2795

Running 'Database' benchmark...
    Create database... 0,124 seconds
    Populate database... 0,970 seconds
    Save database... 0,084 seconds
    Reload database... 0,039 seconds
    Search database... 1,264 seconds
    Sort database... 0,633 seconds
    Group database... 0,727 seconds
Score: 5597

JRMark (version 32.0.58 64 bit): 3849


Database and general operation

The importation of 12600 files took 90 seconds. Database operations on the files are instantaneous.

Theater view takes less toll on the battery than on the old Surface pro 2017 and Obsidian touchscreen is finger friendly as usual.

Audio
It plays audio seamlessly of course whatever it is through wifi DLNA or locally. For external DACs one only has to use built-in microsoft audio USB drivers, all modern DACs are USB 2 audio compliant . Some people will miss original ASIO drivers, it is not my case, WASAPI exclusive is as good.

WDM driver to WASAPi loopback
As expected the WDM and  ASIO drivers don't work, we have thus to come back to the archaic WASAPI loopback function which works perfectly. I am presently listening to Qobuz music with a Sony bluetooth headset and passing by the JRiver convolution DSP thanks to Wasapi loopback. It is operating perfectly. Not any glitch.

Televison
I have been testing remote wiewing of television with 2 different JRiver servers. One locally in the same house and another one 500 km away. In both cases it is operating perfectly. Conection to the server is instantaneous. Display of channels are fast. I use of course theater view in that case. JRiver media center is still its best remote !

Video

To measure the load of the video on the battery I am using Hwmonitor . It is probably not very acourate nor very repeatable nor linear but it gives a good idea  of the improvements.
Test are made with the film Seven 1920x1080 , 2 hours long, mkv , bitrate 8374 , using JRVR and Hardware accelaration .

Seven Film of 2 Hours and  2 minutes      
          Surface pro 11 2024   Surface pro 2017
JRiver     9900mWh or 20%     18310mWh or 40%
VlC UWP   8200mWh or 16%   12520mWh or 25%
      

   - I don't understand why the results are so good although JRVR is using emulation !
   - VLC is only slightly better than JRiver for the surface pro 11 but I am ready to pay a few percent more of battery to use JRiver Video Renderer.




Conclusion
I am happy with this purchase.
I am afraid that these results are not going to motivate the JRiver team to develop an ARM version as it already works quite well. However they can motivate other people to buy this new device and use JRiver.
Logged

JimH

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 72205
  • Where did I put my teeth?

Thanks for the excellent report and for the good news.  There is a thread asking about the new Surface here.

https://yabb.jriver.com/interact/index.php/topic,139387.0.html
Logged

zybex

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 2489

Excellent review, thank you!
The image creation/destruction benchmark is indeed strange.
Logged

Awesome Donkey

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 7672
  • August and everything after...

There's a chance the driver situation may change with newer Windows 11 updates. Others have noted issues with drivers in general and apparently Microsoft is aware of it and hopefully will improve compatibility with the newer updates. It might be something to re-test from time-to-time, especially after monthly Windows updates.
Logged
I don't work for JRiver... I help keep the forums safe from Viagra and other sources of sketchy pharmaceuticals.

Windows 11 2024 Update (24H2) 64-bit + Ubuntu 24.04.1 LTS Noble Numbat 64-bit | Windows 11 2024 Update (24H2) 64-bit (Intel N305 Fanless NUC 16GB RAM/500GB M.2 NVMe SSD)
JRiver Media Center 33 (Windows + Linux) | iFi ZEN DAC 3 | JBL 306P MkII Studio Monitors | Audio-Technica ATH-M50x Headphones

davewave

  • Regular Member
  • Junior Woodchuck
  • **
  • Posts: 67

Great work.  Thank you very much.
Logged

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5232
  • "Linux Merit Badge" Recipient

Excellent review, thank you!
The image creation/destruction benchmark is indeed strange.

So, for what it's worth, on Linux systems that particular benchmark (image creation/destruction) is always much, much slower than it is on Windows, even when testing using the exact same hardware.  On the order of 10x slower.  I wonder if that particular benchmark is tuned for something native x86-Windows specific and the ARM emulation layer is tripping it the same way Linux does? 
Logged
Pages: [1]   Go Up