An improvement in work-unit completion time due to an SSD will be dependent on what you were using before. In my scenario the disk utilization and response times (running 12 CEP work-units) improved significantly for the first 2 minutes of the work-unit. The remainder was only a little better, however I was already running these work-units on a 5 drive array. This will of course result in some overall CEP improvement. How much is difficult to measure due to the WU variability.

In my case I moved the BOINC data directory to an 64 gb Intel X-25E SLC SSD to remove the load from my RAID array. This unit has a 2 petabyte endurance rating before manual over-provisioning. Running more than a couple CEP units on a consumer level SSD drive, in my opinion, does not make sense due to the amount of writes generated. One of my quad systems is exclusively CEP and generates 1 TB of writes per week. The workstation is running 12 concurrent CEP work-units and has 22 TB of writes on the drive so far. I think it only makes sense to use an SSD if you are running a large number of work-units. Otherwise I would just run these on a HD. Hard drive performance can always be improved by placing additional drives in RAID 0.

Not that this is in response to your original question, but since you have a fair amount of RAM, have you considered moving your /slots directory to a RAMdisk (tmpfs)? I run one on my 2c4t Ubuntu SSD rig (12GB RAM) and don't write more than a gig every few days. Also the SSD isn't accessed except to read the work units onto the RAMdisk and write them off.

The new Q-Chem update is in the queue but there are several higher priority tasks ahead of it. Also it is not expected to greatly reduce the disk I/O.

Thanks,
armstrdj