fancycache uses 2x, 3x memory for 2nd & 3rd drive.

[logon]

This is in windows xp x64. I have three drives, 4GB of memory, 2.8gb available, I set cache of disk 0 at a 256mb, windows task manager shows memory usage is 280mb, OK, some overhead beyond cached size, set disk 1 cache to 256mb, shows usage is 581mb for that drive alone, set disk 2 cache to 256mb, shows usage is 894mb for that drive alone. If I turn on cache for just one drive at a time, disk 0 always uses a little more than set cache size, disk 1 always uses more than double the set cache size, disk 2 always uses more than triple the set cache size.

[Support]

Hi,

We don't find the issue you mentioned. Just a question: Which index did you look for in the Windows task manager in order to know the memeory usage?

[logon]

Performance tab, physical memory, available, before and after turning on cache. That and fancycache won't let you allocate a cache size under available, saying "The L1 cache size is larger than the available phsical memory." Trying to make a 512mb cache with 809380 bytes available, it makes a 476mb l1 cache, leaving 33xxx bytes available. More testing seems to show the larger the cache setting, the less "overage" it uses. Able to create a 2gb cache on disk 2 out of 3gb of available memory, with 256mb left, a 1gb cache out of 3gb on disk 2 leaves 1.3gb available.

[Support]

Trying to make a 512mb cache with 809380 bytes available, it makes a 476mb l1 cache, leaving 33xxx bytes available

The program automatically detects the current available physical memory to prevent some users from mistakenly inputting an invalid value of the cache size. Also it always trys to remain some physical memory for the operating system. The purpose is to prevent system slowdown because of insufficient memory.

More testing seems to show the larger the cache setting, the less "overage" it uses. Able to create a 2gb cache on disk 2 out of 3gb of available memory, with 256mb left, a 1gb cache out of 3gb on disk 2 leaves 1.3gb available.

I'm afraid that there might be something you misunderstood.
If possible, can you follow the below steps to trace the issue?
1) stop all the caching activated by FC to release the memory used. You shall click "Stop Caching" for each FC enabled volume/disk, otherwise the caching is not disabled whether you exit the FC's control program or reboot the computer.
2) The message "OS Managed Memory -Free/total" at the bottom of the program shows the available memory which value is same as that from the windows task manager. You can know the memory used by tracing the values before and after "Start Caching".
3) Now you can input the cache size, and then click "Start Caching" button to activate the caching. The actual memory used for caching is showed by "L1 Cache Size" in the "Status" panel. (this value doesn't count the overhead memory used.)

[logon]

With no cache set for any drive, going by what's reported by fancycache "OS memory managed - free/total(mb)," 3010/4094. Starting with no cache set for any drive, after setting a 2gb cache for disk 0, fc shows 898/4094. Turn it off, after setting a 2gb cache for just disk 1 shows 621/4094. Turn it off, after setting a 2gb cache for just disk 2 shows 313/4094. Doing the same(starting with nothing set/setting one drive at a time) but with a 256MB cache setting instead of 2gb, for disk 0 leaves 2730, for disk 1, 2443, for disk 2, 2136. 3010-2730=280mb, a little more than the 256mb set. 3010-2443= 567mb, over 2x 256mb, 3010-2136=874, over 3x 256mb. I don't know who would bother to cache disk 1 or 2 with fancycache when the cost is 567mb & 874mb of memory for a 256mb cache.

For 512mb, left over is 2469, 2181, 1874.
For 1gb, left over is 1947, 1661, 1351.

[Support]

Hi,

Can you check the disk size and block size (FC setting) of these 3 disks?
The overhead memory usage is related to these parameters if you didn't enable Defer-Write and L2 cache. Generally larger disk size or smaller block size, more memory used.

[logon]

4k block size, capacity 20gb, 298gb, 598gb for disk 0, 1, 2, defer write enabled, l2 cache disabled. This should be the cause, the solution, increasing block size, isn't much of a solution though, as it just hides/transfers the wasted space to the allocated cache, with most disk activity not being 64/128/256/512k blocks. A feature that would help would be to add compression like Linux compcache. Those mostly empty large blocks with zeros in them would compress rather well, even simple RLE would make quick work of it, or stuffing those large blocks with smaller r/w like reiser or ntfs is supposed to do with some, or using 4k blocks and compressing another way.

[Support]

We have ways to reduce the overhead memory used, but it degrades the performance. So we adopt the alternative solution: increasing the block size for the disk with large capacity.

[logon]

Referring to degradation to overall or memory? Compression adds overhead for memory access but faster than accessing disk, faster overall. http://code.google.com/p/compcache/wiki ... DiskVsRamz http://wiki.compcache.googlecode.com/hg ... ile_1g.png Still, for those who would want to pay the space cost for the highest memory access possible, compression as an option solves that issue. If most accesses are less than half the size of your block setting, your cache size is effectively less than half the set size, as if I had set 4k blocks at half the cache size setting for my disk 1 drive. Actually I imagine it's worse for larger cache settings.