Romex Software Forum

Adding: the thrashing I mention is almost certainly a result of the the minimal uncached reading the writing app still needs to do. It's very little, but once a disk starts to thrash, it's so slow that it takes a while to recover. If PrimoCache would temporarily back off writing when it recognizes e...

First, thanks for clarifying. There's still a problem: When the disk is thrashing (100% active in windows task manager, *very* slow writing - fluctuating around 1 Mbps) even though there is plenty of both L1 and L2 cache space available (the deferred blocks is at less than 10000 which is about 2.5GB...

I wonder. The following from Support...: "By current design, when L1 is full, L1 cache will flush certain amount of "dirty data" into disk, at the same time incoming new write-data will go to L2 cache." ... doesn't qualify as applying only with multiple Volumes. Also, in the othe...

I changed Average to Intelligent and 60s to 90s. It didn't make a difference. Otherwise everything is identical. Thanks.

The only benefit I can see is that if the write cache storage isn't needed due to low writing activity, reads will be made from the cache. Defer Write makes a huge difference when you have heavy simultaneous writing to multiple files. As far as I can tell, PrimoCache essentially sequentializes the w...

To install this over an existing 3.0.2, is it simply a matter of stopping the cache and then running the new installer? Any gotchas? (like needing to entirely remove the cache and re-create it in 3.0.9) Thanks.

I have one cache task with an L1 cache of 16GB and L2 Cache of 96GB configured for an 8TB drive. Both L1 and L2 are write only (100% write) 256k block size, running Win10 1709 with 32GB RAM. Most of the time the L1 cache is sufficient to keep up with the disk with Defer Write (Average, 60s), but I r...

Thanks. What is the drawback of average write mode compared to the others?

That's great - thanks for the reference. Following up on your point above: "In current versions, when L1 write cache is almost full, "urgent writes" begins to count, though L2 write cache might still have lots of free space. We will improve this." Since in my case the cache is 10...

Ah - I didn't realize TRIM was also issued to spinning disks.

What about the discrepancy between requested writes and total writes (consistently at around 95%) ? Thanks.