Michael, This is out of my head so forgive me if I am off on a factual point or two - but you will get the gist of it.
In 32 bit addressing each memory "address" takes up 4 bytes (4 x 8 bits) but a 64 bit address space is wider (as in 8 x 8 bits). So for example, if you store an integer - say 14 - on a 32 bit system, it will take 4 bytes to do it (00000000 00000000 00000000 00001110). Of course you might think "hmmmm... it could actually store that in a single byte - but being that draconian with storage would only slow the computer down. It needs lots of "address spaces" of uniform size. Think of a post office with dozens of "P.O. Boxes" - little cubbies where mail is placed. It would not make sense for them to resize each box for the volume of mail. They just accept the fact that some boxes have 1 letter and others may have 12 or 13. In a 64bit address space however the address spaces are wider (by another 32 bits). So that 14 is going to look like 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00001110 internally. The advantage of 64bit is NOT in speed of addressability. They are generally faster because they are new gen systems with faster FSB, shorter electron paths, better L2/3 caches etc. The advantage is in total addressability. The 32 bit system is limited to 2 to the 32nd power (4.2 billion add "addresses) whereas 64 bit systems have 2 to the 64th power addresses. The "width" means exponentially more addresses available as "PO boxes" for bits of things that need storing. When I say exponential I mean something like 16 exabytes of possible addressable space. So the good news is that a 64 bit system can utilize an unlimited amount of memory (unlimited in any practical sense at this point in our computer evolution). The bad news is, it takes more room to store equivalent data. -Mark Mark Kruger - CFG CF Webtools www.cfwebtools.com www.coldfusionmuse.com O: 402.932.3318 E: mkru...@cfwebtools.com Skype: markakruger -----Original Message----- From: Michael Dinowitz [mailto:mdino...@houseoffusion.com] Sent: Tuesday, September 03, 2013 7:19 AM To: cf-talk Subject: Re: Good jvm ram settings for 64 bit ColdFusion Mark, To start, the assignment of 6 gig to maxheap and 2 gig to maxperm is working perfectly. I'll expose my ignorance here and ask about that 80%. Is there some inefficiency in jvm ram assignment in 64 bit in comparison to 32 bit? Is there more overhead in ram assignment? Just wondering what's up with the difference. Thanks On Mon, Sep 2, 2013 at 2:13 PM, Mark A Kruger <mkru...@cfwebtools.com>wrote: > > Michael, > > Rule of thumb is 80% for equivelancy. In other words, a 1.8 gig 64bit is > equal to a 1gig 32 bit. Heap. Or you can just multiply times 2 (which is > usually what I do). So a 6 gig heap is roughly 3 or more times the size of > your 32bit 1 gig heap. If your server is dedicated to CF I think your 6 gig > heap is great. Takes most of the physical memory for CF and leaves 2gs for > the OS, monitoring etc. > > -Mark > -----Original Message----- > From: Michael Dinowitz [mailto:mdino...@houseoffusion.com] > Sent: Monday, September 02, 2013 12:18 PM > To: cf-talk > Subject: Good jvm ram settings for 64 bit ColdFusion > > > Does anyone have any suggestions for good workable ram settings on a 64 bit > CF10 install on a box with 8 gig of ram? I understand I can use more than > the standard 1 gig but has there been any work on the best settings? > Otherwise I'm thinking to start with 6 gig maxheap and 4 gig maxperm. > > > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Order the Adobe Coldfusion Anthology now! http://www.amazon.com/Adobe-Coldfusion-Anthology/dp/1430272155/?tag=houseoffusion Archive: http://www.houseoffusion.com/groups/cf-talk/message.cfm/messageid:356696 Subscription: http://www.houseoffusion.com/groups/cf-talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/groups/cf-talk/unsubscribe.cfm
