Display Calibration and Profiling with Switchable Graphics

[lhodaniel]

I have a basic question that has come to mind because I'm shopping for a new laptop. I've never had a pc with dual switchable graphics, Intel + dGPU. How does that work wrt to monitor profiles? Take as two examples the 15" rMBP and the MS Surface Book. Does the dGPU always take over with certain apps: PS, C1P, etc? Or do these units switch back and forth based on operations or work load within the apps? Does one have the ability to profile both the integrated and dGPU separately for each display, or have to pick one (dGPU)? I feel like I should know this, but I've never had to consider switchable graphics before.

Thanks,
Lloyd

[digitaldog]

I don't know the answer, certainly for anything MS. But what you might be able to do is open an image in Photoshop with lots of differing colors. Make a screen capture of each possible method and then subtract the two in Photoshop:
http://digitaldog.net/files/Apply_Image.pdf
Low tech to be sure, but it might show large differences which would answer your question.

[AlterEgo]

I have a basic question that has come to mind because I'm shopping for a new laptop. I've never had a pc with dual switchable graphics, Intel + dGPU. How does that work wrt to monitor profiles? Take as two examples the 15" rMBP and the MS Surface Book. Does the dGPU always take over with certain apps: PS, C1P, etc? Or do these units switch back and forth based on operations or work load within the apps? Does one have the ability to profile both the integrated and dGPU separately for each display, or have to pick one (dGPU)? I feel like I should know this, but I've never had to consider switchable graphics before.

Thanks,
Lloyd

with modern software from vendors like nVidia technically iGPU always handles the connection with notebook LCD (or external), dGPU when used is passing the data to iGPU to pass it further to display... as for the numbers crunching the software can use both iGPU and dGPU at the same time (P1 claims C1 does for exampe)... that has nothing to do with "display/monitor" profiles though - you create one profile - it works on a higher level (OS/software/drivers) than iGPU/dGPU interaction controlled by a relevant software from nVidia or Amd seamlessly under the hood for CMS

for example = https://en.wikipedia.org/wiki/Nvidia_Optimus -> http://www.nvidia.com/object/LO_optimus_whitepapers.html

[howardm]

On most Mac's w/ dual graphics, you can sort of control it in the Energy Pref's to allow/prevent the OS from
auto-switching between the 2 choices but I'm not sure you can 'lock' it to  a specific GPU unless basically battery = IGPU

[lhodaniel]

Now that I'm up and running on the MBP, I have the answer to this question. I dl'd a nice little utility called gfxCardStatus which indicates which graphics system is being used at any given time. It can also be used to force one or the other on demand. It seems that the MBP always uses dedicated graphics when there is an external display attached, at least in clamshell mode plugged into AC.

I'm really enjoying this machine. It seems faster than my 3 year old E3 PC desktop (now stored in the attic). The only thing that has been dog slow was rebuilding standard previews for my entire catalog in LR CC. I understand that is an issue between LR and El Capitan. FWIW...

Lloyd

[Steve Upton]

I have a basic question that has come to mind because I'm shopping for a new laptop. I've never had a pc with dual switchable graphics, Intel + dGPU. How does that work wrt to monitor profiles? Take as two examples the 15" rMBP and the MS Surface Book. Does the dGPU always take over with certain apps: PS, C1P, etc? Or do these units switch back and forth based on operations or work load within the apps? Does one have the ability to profile both the integrated and dGPU separately for each display, or have to pick one (dGPU)? I feel like I should know this, but I've never had to consider switchable graphics before.

I've not had the time to exhaustively test this but I've not found information or evidence to dispute it either:

I think the idea of the graphics card affecting the calibration of the display is a hold-over from the world of analog display systems (CRT and analog-driven LCD).

In that "age" the D/A converters in the graphics card, along with the cables and connectors had a particular "signature" that could be unique to the system. Variations between systems could cause one system to behave differently than another (connected to the same physical display). As a result, the prevailing advice was to calibrate each system separately - even if they shared the same display (like two CPUs sharing a display, many CPUs sharing a projector system, etc).

In the digitally-connected world of digital displays, we have not seen evidence that the calibration curves in the graphics card, the cables, or anything else in the display chain is affected by changing any component other than the display itself. In other words, for a single display, one calibration / profile could fit all.

I'm curious if anyone else has seen these results.

[Doug Gray]

I've not had the time to exhaustively test this but I've not found information or evidence to dispute it either:

I think the idea of the graphics card affecting the calibration of the display is a hold-over from the world of analog display systems (CRT and analog-driven LCD).

In that "age" the D/A converters in the graphics card, along with the cables and connectors had a particular "signature" that could be unique to the system. Variations between systems could cause one system to behave differently than another (connected to the same physical display). As a result, the prevailing advice was to calibrate each system separately - even if they shared the same display (like two CPUs sharing a display, many CPUs sharing a projector system, etc).

In the digitally-connected world of digital displays, we have not seen evidence that the calibration curves in the graphics card, the cables, or anything else in the display chain is affected by changing any component other than the display itself. In other words, for a single display, one calibration / profile could fit all.

I'm curious if anyone else has seen these results.

I remember those bad old days. Glad they are in the past.

Your supposition sounds right to me and makes good sense but I haven't investigated it.

The monitors I've been using for over 5 years have internal higher precision LUTs so the video card output LUTS, with their associated banding issues never get used. It's all digital.