Slow 2D performance with RDP

[Edberg Yen]

Hi All,

My test environment:
Host: Windoes server 2016
Card: M10
Client: Win10 (Laptop)

If I close RDP right immediately after the beginning of executing PASSMASK 2D test, I will get 420 score.
But if the RDP connection continues when running PASSMASK 2D test, the score down to ~365 score.
Is 420 (w/o RDP connector) ok for M10?
Is the performance drop because of RDP or other reason?
Is M10 for 2D benchmark ok or I should not use M10 for 2D application?

[Tim (PassMark)]

Closing the remote desktop connection will cause the 2D tests to render on the locally connected monitor rather than across the remote connection, yes this will change the score as running across a remote connection will slow down the execution of the test. Closing the connection may also cause the display output to become invalid, causing the test to stop running or in some cases produce invalid results.

The remote connection may also be running at a different resolution and DPI setting to the local display, which will also change the test result slightly.

[Shankboy]

I have a brand new PC just built. I have run the benchmark a number of times - as the system was first operating and as i added software and some hardware. The 2d portion of the test has gone down in performance pretty substantially over these first 24 hours. I have a GEFORCE GTX 1660. The only real change made that i can think of that would affect the card is the addition of a second monitor after my first few runs of the performance test. Would this cause the Complex Vectors and/or the Windows Interface portions of the test to slow down? Are there steps I cna take to enhance this portion of the systems performance? Is there a different card i shoudl consider?

[Edberg Yen]

Hi Tim,

Thanks for quick reply.
Just to clarify my configuration. My host is a system with M10, and I would like to run passmask 2D test on this system. The system's OS is windows server 2016.
Because there is no vga output on on system, so I use RDP as a connection utility to control the system.
The passmask is running on my system, and my laptop is just a client and this client use RDP to remote control my system.
As my last post, there will be a 420 v.s. 360 score gap between w/o and w/ connection by using RDP.
I would like to know why there is a drop when using RDP? Is because the system need resource to compressing the screen image and transfer it to my laptop or other reasons?
Is this drop normal if I use remote connection utility, no matter RDP or other utility?
Is 420 for M10 reasonable? Or M10 is not a good card for my application?

[Tim (PassMark)]

>>Because there is no vga output on on system

If you close the remote connection in this case some of the tests may fail to run (at least in my testing this is what happened) and some may continue to run thinking there is still a valid output when there isn't, so the results would not be correct. We're looking at making changes to detect and handle this case so the test errors out if possible rather than running like this but it's a complicated state.

>> I would like to know why there is a drop when using RDP?
>>Is this drop normal if I use remote connection utility, no matter RDP or other utility?

There is a lot of overhead involved when using such a connection, you're effectively emulating a software monitor across a network connection.
Any connection using a remote connection rather than a physical monitor will likely be slower.

>>Is 420 for M10 reasonable? Or M10 is not a good card for my application?

We don't have any results in the database for this card so we can't say. The developers of the specific application you're using should be able to make a recommendation.

[Tim (PassMark)]

>> addition of a second monitor after my first few runs of the performance test. Would this cause the Complex Vectors and/or the Windows Interface portions of the test to slow down?

This shouldn't change the score, unless the DPI and resolution settings have changed as well. Easiest way to check this is remove the second monitor before testing.

[Edberg Yen]

Hi Tim,

Thanks for your reply.
M10's score 420 is list on PassMark VideoCard Benchmarks, the link is https://www.videocardbenchmark.net/GPU_mega_page.html, and searching M10 can get 420 score in 2D test as attached photo. And my test result is pretty near this score.