Precision 7560 & 7760 Owners' Thread

You should try driving those monitors via USB-C as it allows the Nvidia GPU to shut off unless an app requires it. If you have mDP/HDMI set as the primary display you will likely have to reboot to see a power drop even if the displays were unplugged as Windows likes to startup a bunch of background processes that tie themselves to whatever GPU the primary display is set to. Because of this issue, I think the Precision is really superior to the Thinkpad / Zbook line because it has the ability to drive addition displays via USB-C connected to the Intel GPU.

 

Note that this only works if you have displays attached via USB-C and direct discrete output mode turned OFF in the BIOS.

 

I'd be happy to:


That said, overclocking is not available on the Xeons (not that I need it: the CPU maintains 3.8 - 4 GHz easily).

Not yet: I'll try this. I've operated under the assumption that the core and cache voltage offset should be exactly the same. Can't remember where I read this.

 

https://www.dell.com/en-us/shop/pfydresults/272815?categoryId=4325

I noticed that they seem to have "corrected" the price of the 32GB ECC memory modules that you can purchase separately at the Dell website. They are now $450 USD. In early July they were $220 USD. Still a bit cheaper than ordering them with the system, but the difference is much less drastic...

 

Solved the mystery of the missing Cuda option in the task manager GPU usage graph.

With this setting on, it is not there.

With the setting off, it is there.

Anyone know what the setting does, and whether or not it should be on? Obviously it tells me what it does, but just wondering if it's a good thing in reality vs MS's recommendation...

I'm using hybrid graphics if that's relevant.

Thanks

 

https://devblogs.microsoft.com/directx/hardware-accelerated-gpu-scheduling/

Basically, you'll get a modest performance boost in some applications if this is enabled. Very modest (<2%). Here's some benchmarks:
https://www.gamersnexus.net/guides/3599-windows-10-hardware-accelerated-gpu-scheduling-benchmarks
(Some show better with it enabled, some show better with it disabled, margin of error is small in either case.)

 

Thanks a lot. Why would it stop me being able to see the cuda usage in TM? Really annoying...

 

Is there anyone try to run stress test both CPU and GPU using Furmark ? I tested my laptop and while the A2000 GPU was capped at 80W, the CPU was limited to 35 W !!!
I hope the total power for CPU + GPU (especially CPU) should go higher because the same cooling system is being used for higher specs.

 

You could just run another CPU benchmark (Ex. Cinebench).
Was the CPU capped at 35W when you put a load on the CPU and GPU? 80W GPU power is at the limit of the 7560, so the CPU being stuck at 35W isn't unrealistic.
However, the GPU using all the available wattage while giving much less to the CPU in a combined workload makes sense since most cases are either 100% GPU or 100% CPU.

 

If run CPU stress alone, the max power is 95W for CPU. When stressing both CPU and GPU, CPU was capped at 35W. I supposed that with GPU runs at 100% (80W), CPU can run with higher power.