RTL Test Losing bytes on USB3.1

[air-scan]

You're dealing with 20 dropouts per minute; that's one every three seconds. That'd definitely be noticeable.

ya its noticeable. nothing in the bios that allows for legacy power saving. Thats an area of the computer a rarely ever touch. Last time I touched it was last Bios update a few months ago. I hope the DSD+ FL devs will update it to eradicate this situation in some sort of workaround or redesign. Probably wishful thinking. It's really hard to find a brand new windows based laptop with only USB 2.0 anymore. This thread is probably throwing some people off. Not many replies. It's one of them things that the backwards compatibility in USB 3.1 to USB 2.0 isn't always going to be stable.

I tried to get DSD+ FL unvoiced speech quality increased to use more CPU load. I cant get it past 3% CPU usage. I went up to -u30 and AMBE unvoiced to 60. It isnt enough to load the CPU.

I guess we wait and see if anyone else noticed this who is using a USB 3.1 only laptop or desktop.

 

[a417]

@slicerwizard, have you tried running it at high priority or realtime? Maybe windows won't be as draconian at powersaving if it's run at elevated run-levels?

 

[a417]

Folding@Home's client set at light utilization uses approx 50-60% CPU utilization on the i7 I'm sitting in front of, if you need to keep it awake you could always let it be (semi-) productive. Gone are the active days of Seti@Home, and distributed.net hasn't updated their clients in 5 years (still plugging blocks away, tho) so there are some options if you need a keep-alive.

 

[slicerwizard]

@slicerwizard, have you tried running it at high priority or realtime? Maybe windows won't be as draconian at powersaving if it's run at elevated run-levels?

FMP24 runs at high priority. Used Task Manager to bump it to realtime priority. Had no effect. I'd expect that to only affect the Windows scheduler. From what I've read, the power saving is done automatically by the processor when it sees little use. Windows doesn't appear to be reprogramming the relevant control registers on the fly.

 

[air-scan]

FMP24 runs at high priority. Used Task Manager to bump it to realtime priority. Had no effect. I'd expect that to only affect the Windows scheduler. From what I've read, the power saving is done automatically by the processor when it sees little use. Windows doesn't appear to be reprogramming the relevant control registers on the fly.

I think DSD+ isn't properly supported in Windows 10 20H2. I too tried realtime priority on both FMP24 and dsdplus.exe. No bueno. It decodes but not as it should. I think DSD+ needs better Windows 10 20H2 support on Intel core i7 10th gen. DSD+ just isn't giving the hardware enough work to keep alive enough to decode as smooth as it should be. Asking DSD+ team for overhaul of it can be looked at as a positive thing because I see potential for beefier better audio output especially for APCO P25 P1 and PII systems. It just needs to take more advantage of newer CPU's offerings and better support for newer power management.

Rumor has that DSD+ team uses Windows 8 for development. Windows 8 Extended Support ends on January 10, 2023. I don't see any changes soon if the rumor were to be fact. Just hearsay.

slicerwizard try SDRTrunk on the same machine with the only USB3.1. It's better. Java works the machine harder.

 

[slicerwizard]

Rumor has that DSD+ team uses Windows 8 for development. Windows 8 Extended Support ends on January 10, 2023. I don't see any changes soon if the rumor were to be fact. Just hearsay.

I've never seen that rumour. I guess I don't know the right people. DSD+/FMPx et al run on XP (was using it on an old XP Atom netbook today), so they must be developing on XP? I can start rumours too...

slicerwizard try SDRTrunk on the same machine with the only USB3.1. It's better. Java works the machine harder.

The latest FMP24 (2.71) with CPU loading enabled is working fine. No lost samples reported by FMP24's test mode and no audio glitches, so I'm good.

 

[air-scan]

I just updated to 2.303 and enabled cpu loading. It's repeating no data loss. FMP24 31% cpu usage. Not the efficiency I like but it's only a 'work-around'. Still lots of room for improvement.

 

[slicerwizard]

5% usage here. I dropped the clock speed to 800 MHz.

 

[Kazzaw]

Also did a test on my machine, was dropping data like crazy apparently. Enabled CPU loading and no more data drops. CPU usage has increased to around 25-30%. Running on a 10th gen Intel mobile CPU.

 

[kj6psg]

If you're trying to keep the CPU clock high on Windows without burning extra electrons to make the CPU busy, you can go into Control Panel, Power Options, Edit Plan Settings for your current power plan, Change Advanced Power Settings. Find Processor Power Management, Minimum Processor State and raise that to something high (I use 90% on my computer). This should limit or entirely prevent the CPU from throttling down to save power, depending on how high the minimum state is set. It definitely makes an improvement on real-time processing tasks. There is some BIOS dependence, so you can verify the changes work by checking your current CPU speed in Task Manager. The setting is available since Windows Vista, and it's much healthier for the computer than producing excess work/heat.

Additionally, if you're trying to burn extra electrons to make the CPU busy and thus run faster, you run into the real risk of bringing the CPU up to temperatures where the BIOS forces the CPU to slow down to prevent catastrophic overheating. These thermal throttling states can be worse than the minimum processor state, depending on CPU and BIOS; typically, it's the bare minimum CPU state needed to keep the system 'running'. Your Windows settings won't change what the BIOS does when it needs to thermal-throttle the system. It's not often a problem on desktops (have you cleaned your heatsinks and fans out recently?), but it's easy to produce on laptops, which have fewer and narrower outlets for waste heat. In the event where the CPU can produce more heat than the heatsink can evacuate, you can estimate Maximum Processor State's value to prevent the CPU from putting out too much heat using a stress-test program and some patience