Disable HDCP to make your computer slightly more responsive

There is zero evidence that HDCP introduces any input lag, or reduces performance.
 
I recently installed latmon to troubleshoot an issue with a pice riser cable problem.

Perhaps I'll give it a try.
 
does disabling HDCP still prevent playing back certain types of media or was that just a DVD / Blu-ray thing? I haven't even thought about HDCP in awhile... it would be both hilarious and incredibly frustrating if that has been contributing to Nvidia's infamous DPC latency issues this whole time.

But this is the first I've heard of HDCP being a culprit, and I'd like to see some data from folks here to corroborate the YT vid.
 
NattyKathy said:
does disabling HDCP still prevent playing back certain types of media or was that just a DVD / Blu-ray thing? I haven't even thought about HDCP in awhile... it would be both hilarious and incredibly frustrating if that has been contributing to Nvidia's infamous DPC latency issues this whole time.

But this is the first I've heard of HDCP being a culprit, and I'd like to see some data from folks here to corroborate the YT vid.
Click to expand...
All the various streaming service apps won't work without HDCP. Disney+, Amazon Prime Video, etc.
You also have to disable HDCP to be able to use a capture card. Games don't require HDCP. But if its on, the capture card won't be able to grab the video signal.
 
chameleoneel said:
All the various streaming service apps won't work without HDCP. Disney+, Amazon Prime Video, etc.
You also have to disable HDCP to be able to use a capture card. Games don't require HDCP. But if its on, the capture card won't be able to grab the video signal.
Click to expand...
Yeah, I have noticed that HDCP is automatically disabled on my outputs that are connected to capture cards. Didn't know about the streaming thing but that makes sense.
 
Wow... someone's Big Mad.

You're going to need to come at us with some data before I can bring myself to care exactly how erect you get at changing a value in the registry.
 
Mchart said:
You’re free to do that and show the results.
Click to expand...

XenIneX said:
Wow... someone's Big Mad.

You're going to need to come at us with some data before I can bring myself to care exactly how erect you get at changing a value in the registry.
Click to expand...
Here is my testing. It is very basic, started latmon and ran furmark three times. Do with these results what you will, feel free to do more in depth testing.
  1. Stock, no reg entry
  2. Reg entry enabled, set to 1
  3. Reg entry disabled, set to 0
One the left is the reg entry enabled, set to 1, and on the right is with the reg entry disabled, set to 0.
1647123991759.png

On the left is no reg entry and on the right is with entry disabled, set to 0.
1647123907666.png


You can see with either no reg entry or the reg entry disabled, set to 0, the latency appears to be higher compared to reg entry enabled, set to 1. This was a highly unscientific test, please feel free to knock yourselves out with your own testing, I've already wasted enough time on it.
 
Last edited:
Even if you measure the delta between measured highs on latency and not the averages, that still isn't something anyone is going to be able to perceive. Keep in mind 1 millisecond is 1000 microseconds. So yes, you've demonstrated there may be a difference in that test, but we aren't even talking something that is even 1ms in difference. No human being is going to notice that.

If you've got some other test that demonstrates why this might be useful, please do so. At this point, i'm not convinced it's worth worrying about.
 
Silentbob343 said:
Here is my testing. It is very basic, started latmon and ran furmark three times. Do with these results what you will, feel free to do more in depth testing.

[...]

You can see with either no reg entry or the reg entry disabled, set to 0, the latency is much higher compared to reg entry enabled, set to 1.
Click to expand...

Looks like a whole lot of nothing to me. Like, that reads as a single 3ms spike on process instantiation, and then business as usual. Which is what I'd expect -- OS verifies the status of the HDCP hardware, which does its thing basically for free because it's a dedicated hardware IP block.

In any case this micro-benchmarking of interrupts isn't even relevant. Show motion-to-photon variance and we'll have something to work with. Show stuttering on frame-time graphs and >1% framerates. Absent that, it's just a whole lot of "oh noes, the OS is doing something in the background, and I don't understand it!"
 
chameleoneel said:
All the various streaming service apps won't work without HDCP. Disney+, Amazon Prime Video, etc.
You also have to disable HDCP to be able to use a capture card. Games don't require HDCP. But if its on, the capture card won't be able to grab the video signal.
Click to expand...
Depends on "how" it's disabled. I have no problems capturing "whatever".
 
cjcox said:
Depends on "how" it's disabled. I have no problems capturing "whatever".
Click to expand...
The point is, streaming apps won't start, if they detect that HDCP isn't being respected. And capture cards can't capture a signal where HDCP is being respected.
 
Mchart said:
Even if you measure the delta between measured highs on latency and not the averages, that still isn't something anyone is going to be able to perceive. Keep in mind 1 millisecond is 1000 microseconds. So yes, you've demonstrated there may be a difference in that test, but we aren't even talking something that is even 1ms in difference. No human being is going to notice that.

If you've got some other test that demonstrates why this might be useful, please do so. At this point, i'm not convinced it's worth worrying about.
Click to expand...

Does it really need to be stated that latency is a CUMULATIVE phenomenon? 1ms here, 1ms there. It all adds up until it's higher than your refresh rate. At 144hz, you have to be under 8ms TOTAL for next frame response. It's even worse at 240hz. Just stop.

Objectively proven wrong but such an ego case psychopath that you can't just admit it. Laughable.

Hi I'm mchart. I LIKE latency. Please. Give me more daddy.
 
bigbluefe said:
Does it really need to be stated that latency is a CUMULATIVE phenomenon? 1ms here, 1ms there. It all adds up until it's higher than your refresh rate. At 144hz, you have to be under 8ms TOTAL for next frame response. It's even worse at 240hz. Just stop.

Objectively proven wrong but such an ego case psychopath that you can't just admit it. Laughable.

Hi I'm mchart. I LIKE latency. Please. Give me more daddy.
Click to expand...
You haven't demonstrated a cumulative latency phenomenon, nor shown us the delta between those measures with HDCP on/off.
 
Mchart said:
You haven't demonstrated a cumulative latency phenomenon, nor shown us the delta between those measures with HDCP on/off.
Click to expand...
If you have a 360hz display, the entire frame has to be processed in 2.8ms, and you're pretending that 1ms isn't a big deal. 1ms of extra latency is catastrophic, and it's only going to get worse as refresh rates rise. End game is 1000hz so you can finally eliminate motion blur/persistence, which is 0.1ms for the whole frame meaning that this adds 10x more latency than the entire frame should take.
 
bigbluefe said:
If you have a 360hz display, the entire frame has to be processed in 2.8ms, and you're pretending that 1ms isn't a big deal. 1ms of extra latency is catastrophic, and it's only going to get worse as refresh rates rise. End game is 1000hz so you can finally eliminate motion blur/persistence, which is 0.1ms for the whole frame meaning that this adds 10x more latency than the entire frame should take.
Click to expand...
latency is a big deal. But, I don't see how HDCP adds latency, with the exception of connection latency, which is annoying, but doesn't count for anything in games or whatever, unless you're switching monitors a lot in the middle of a deathmatch. Beyond a handful of gate delays, I don't see how you would implement a stream cipher on an up to 48 Gbps link and add significant latency, you'd need a rather speedy dual ported buffer chip, and what would it get you? xor isn't more efficient in bulk. Probably the data stream always goes through the xor anyway, if you're not running HDCP, set the cipher input to zeros; that way you don't need to have a data path switch.
 
toast0 said:
latency is a big deal. But, I don't see how HDCP adds latency, with the exception of connection latency, which is annoying, but doesn't count for anything in games or whatever, unless you're switching monitors a lot in the middle of a deathmatch. Beyond a handful of gate delays, I don't see how you would implement a stream cipher on an up to 48 Gbps link and add significant latency, you'd need a rather speedy dual ported buffer chip, and what would it get you? xor isn't more efficient in bulk. Probably the data stream always goes through the xor anyway, if you're not running HDCP, set the cipher input to zeros; that way you don't need to have a data path switch.
Click to expand...
It's cute that you assume that something like HDCP would be implemented in a sensical manner.
 
chameleoneel said:
The point is, streaming apps won't start, if they detect that HDCP isn't being respected. And capture cards can't capture a signal where HDCP is being respected.
Click to expand...
A quick trip to wal*mart can fix all of that.... just saying...
 
You must log in or register to reply here.
Back
Top