Anybody in here running Rosetta and/or BOINC Data Folder in an SSD?

My wear leveling is @ 9 after a whole year of usage (500GB Samsung EVO 840) and I think I've been a bit too conservative towards it's usage when I could be milking it for all its worth.

---

Yes, but I only started a few weeks ago. There was a large performance gain on ATLAS@home tasks when I installed boinc on my SSD. Apparently my HDD read/write rates weren't fast enough to fully utilize the processor.

Anybody in here running Rosetta and/or BOINC Data Folder in an SSD?

All 5 of my dedicated PCs have BOINC on an SSD. For most projects, there won't be any problem with lifetime such that you need to worry about the wear leveling. But they each have at least one project with a very high write rate. They are World Community Grid/CEP2 and ATLAS, and maybe Climate Prediction Network though I have not measured it for a while.

Each of them can write over 100 GB/day to your SSD for each work unit that is running, so with several cores running you could shorten the lifetime of your SSD significantly. So for those PCs, I either place the BOINC data folder on a ramdisk (either Primo Ramdisk by Romex Softare or the DATARam Ramdisk work OK for me), or else I use a write-caching program (PrimoCache by Romex Software). They cut down the writes to the SSD to virtually nothing, and so the SSDs will last forever. Also, I have found a reduced error rate on at least the Climate Prediction Net, and it probably helps the others too.

As for Rosetta, it is so easy that I don't think you need to worry about the SSD lifetime at all.

---

Yes, but I only started a few weeks ago. There was a large performance gain on ATLAS@home tasks when I installed boinc on my SSD. Apparently my HDD read/write rates weren't fast enough to fully utilize the processor.

Anyone worried about SSD endurance should read this: http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

Then do the math. Maybe stay away from TLC drives if you're still worried.

Worrying about SSD failures.....:)-

I have been a heavy PC user since about 1983. In that time I gave seen a grand total of ONE drive failure!!!

Stop worrying and keep crunching.

---

Anyone worried about SSD endurance should read this: http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

That is good news for endurance, but errors are another matter. I have found that I pick up errors on Climate Prediction Net unless I use a ramdisk, as I have posted there in the "Number crunching" section. Apparently the write-rate is so high that there is contention for writing to the SSD. Whether that would be true for spinning-platter drives I don't know, but they would probably slow down the work even if they did not pick up the errors (I have seen people post on that at any rate).

And just a couple of days ago I tried my Samsung 840 EVO (128 GB) on ATLAS, which also has a high write-rate and is notorious for errors. Even when using the Samsung Rapid Mode cache that comes with the Samsung Magician utility, I picked up an error in a couple of hours. But that cache is limited to 1 GB, and does not try to cache all the writes, only those past a certain que depth. Since then, I have gone back to using my PrimoCache to implement a write-cache of much larger size, which caches all the writes and eliminates the errors. There is no point in spending hours on a work unit just to have it error out unnecessarily.

You can use a cache or ramdisk for spinning platter drives too of course; I just happen to have SSDs.