2018-12-01 08:39:47
heh, pc hardware is currently bad enough for amazon to be impacted by it
https://www.anandtech.com/print/13657/amazon-launches-pc-hardware-compatibility-checker Amazon Launches Online Tool to Check Compatibility of PC Hardware
Rotonen
2018-12-01 14:20:52
i wonder if whole drive encryption affects mining, or if it only would once the processor is maxed out.
https://stackoverflow.com/questions/2762844/how-does-bitlocker-affect-performance
@Fable I'm an http://ASP.NET / C# developer. I use VS2010 all the time. I am thinking of enabling BitLocker on my laptop to protect the contents, but I am concerned about performance degradation. Developers who ...
cryptovape
2018-12-01 14:32:10
@cryptovape it absolutely effects mining for snowblossom, if it's using the drive at all.
Clueless
2018-12-01 14:33:43
• encrypted data has overhead, takes longer to fetch
• encrypted data might be distributed differently in storage, takes longer to fetch
• encrypted data has to be decrypted
• data has to be encrypted to be written again
etc.
Clueless
2018-12-01 14:33:51
so if you're mining on an encrypted drive, bad.
Clueless
2018-12-01 14:35:56
good info thanks. Probably why I was only getting 17 kH/s on my 2014 nvme. maybe Fable is also affected.
cryptovape
2018-12-01 17:16:10
I'm not sure about that.
Fireduck
2018-12-01 17:16:42
I think file system encryption uses a scheme where the key for a particular block is derived based on the offset of the block so the same number reads the same size of reads should be required.
Fireduck
2018-12-01 17:16:58
So minus the CPU cost I think you should have the same performance on an encrypted file system.
Fireduck
2018-12-01 17:21:18
Like this, let X be the master key for the drive
Fireduck
2018-12-01 17:21:31
the key for block N is hash(X,N)
Fireduck
2018-12-01 17:22:16
and the IV can be hash(master_IV,N)
Fireduck
2018-12-01 17:22:26
easy as eating pancakes
Fireduck
2018-12-01 17:22:58
there are probably some layers of indirection to allow to change the master dirve key/password
Fireduck
2018-12-01 17:23:00
but same idea
Fireduck
2018-12-01 20:09:22
:+1:
cryptovape
2018-12-01 20:35:45
I still expect it to add overhead and will assume that until proven otherwise. :P
Clueless
2018-12-01 23:11:54
@Clueless processors actually implement square millimetres worth of stuff to make that as transparent as possible
Rotonen
2018-12-01 23:15:00
@Rotonen every time I've benchmarked NVME (in general), I could swear I've seen diminished speeds.
Clueless
2018-12-01 23:15:07
I could be wrong though
Clueless
2018-12-01 23:30:53
throughput, yes, seek latencies. results vary
Rotonen