F@H and BOINC projects don't play nicely together, if they happen to be CPU WUs.
What works is, to have Boinc WUs running on the CPU and F@H on the GPU. Still, you would need to dedicate 1 thread to F@H for GPU support.

Some projects seem to get slowed down by hyper threading a bit and run much better if you only allow a total of 4 WUs running on your quad core processor.
In your case I would set Boinc to use 50% of the available logical cores (in your case that would be 4) with HT enabled, run a F@H GPU project and let that one be supported by a 5th logical core and see how that works out.
Even though the Nvidia GPU would keep 1 thread / logical core busy running at 100%, that thread is actually not really doing much work except for waiting for the GPU.

Clive:

Intel Core i7-3740QM should be fine. I also run a CPU from the Ivy Bridge (3rd) generation Core and an even older Pentium E5800 (Wolfdale) that was released in 2010, and both handled ARP work units fine while running 24/7.

Looking at the Intel ARK page for your CPU, it's a 4 core, 8 thread CPU.

How much RAM does your laptop have? ARP1 is very RAM-hungry, requiring 1-2 GB for each work unit. If there's not enough RAM, the machine will have to constantly swap to the HDD or SSD, causing both wear-and-tear and dramatically slowing down progress.

If you are running 8 ARP1 work units on that machine, and it doesn't have 8-16 GB of RAM, I'd cut back to only 4 at a time (needing 4-8 GB of RAM).

There are two ways to accomplish this, depending on your situation.

1. Limit the number of ARP1 work units that each machine can have at a time. This applies to all machines using that profile:

https://www.worldcommunitygrid.org/ms/device/...iguration.do?name=Default

(Go to "Custom Profile," then scroll down to "Africa Rainfall Project," and change it to 4 or whatever your preference is.)

2. Another way to limit the # of ARP1 work units in progress at a time on that machine is to create an app_config.xml file and place it in C:\ProgramData\BOINC\projects\www.worldcommunitygrid.org\ folder.

<app_config>
   <app>
      <name>ARP1</name>
      <max_concurrent>4</max_concurrent>
   </app>
</app_config>

The benefit of doing it this way is that it only applies to that machine. Another benefit is it allows that machine to download more ARP1 at a time but only allow 4 to run at once.

I have tried to run F@H and OPN WUs on the CPU, 2 WUs OPN, 4 threads for F@H, leaving 1 thread for the GPU and 1 thread idle.
The run times of the OPN WUs where longer than when running 7 OPN WUs parallel and F@H on the CPU.
Apart from longer run times, there where a couple of OPN WUs returned as INVALID or ERROR. That stopped the moment I stopped F@H CPU WUs.
Others have made similar observations, it's mentioned repeatedly on the F@H forums. I wouldn't run F@H and Boinc on the same CPU together.