Make sure you take the slider and move it to full. There is a 60% drop in CPU usage by keeping it at the default medium.
You can check CPU utilization using the task manager or some other process viewer. The NaCl client lists its process as either Chrome or NaCl depending on the OS you are using.
At 3 minutes/WU at 125 points/WU and 1440 minutes/day then you are getting roughly 60,000 PPD. (60x24/3)x125 = 60,000. With 40 cores bigadv or even SMP is likely to do better.
As a comparison, my 4 core Q9450 @ 2.66GHz does a WU in 10-11 minutes.
Do note that very recently PG removed the QRB from the NaCl client but upped the base points from 10 to 125 points. While for low-end CPU's and part timers that is all to the good but the high-end machines are switching from an exponential PPD curve to a linear one which drops their potential.
PG did have a legitimate reason though. The NaCl client was completing WU's too fast and people were using it to bypass the 80% successful return rate requirement for QRB. You can kill a lot of GPU WU's if someone can generate 150-500 QRB WU's in a day by running one NaCl client on a relatively modest CPU.