Re: JVM Sizing

Tim Slattery wrote:
Lew <lew@xxxxxxxxxxx> wrote:

Tris Orendorff <triso@xxxxxxxxxxxxxxxxxxx> wrote:
Well, a 32 bit JVM could address up to 4 GB of memory. A 64 bit JVM could address up to 2^34 GB of memory.
Tim Slattery wrote:
Why 2**34, that's only two bits more address space than the 4GB 32-bit
version?
2^34 is substantially larger than 4. Both were in units of GB.

2**32 = 4,294,967,296. 2**34 = 17,179,869,184 or 4 times 2**32.
substantially larger, yes, but 64-bit Windows systems, for example,
implement a 37-bit address space which allows for 137,438,953,472
bytes (128GB). Far more than anybody today contemplates installing in
their computer, but miniscule compared to an actual 64-bit address
space.

I just wondered why a 64-bit JVM expands the address space by only two
bits? I agree there's no reason to go all the way to 64-bits at this
point, but 34 bits doesn't seem reasonable to me.

Read it again. He said:
32 bits gives 4 GB (4 gibibytes, actually)
64 bits gives 2^34 GB (again, it should be GiB)

2^34 GiB = 16,777,216 tebibytes = 16,384 pebibytes = 16 exbibytes

--
John W. Kennedy
"The blind rulers of Logres
Nourished the land on a fallacy of rational virtue."
-- Charles Williams. "Taliessin through Logres: Prelude"
.

Relevant Pages

  • Re: JVM Sizing
    ... Lew wrote: ... address up to 2^34 GB of memory. ... Tim Slattery wrote: ... I agree there's no reason to go all the way to 64-bits at this ...
    (comp.lang.java.programmer)
  • Re: Field size with ADO.NET (XML)
    ... John ... > crystal reports, that is consistently running out of memory when we run ... The reason, apparently is that ...
    (microsoft.public.vb.crystal)
  • Re: Check Engine light: A brilliant idea?
    ... John S. wrote: ... Memory is cheap - there's no reason that the message couldn't be ...
    (rec.autos.tech)
  • Re: Future of Risc OS
    ... In article <4ecb512b43john@xxxxxxxxxxxxxxxxxxxx>, John Cartmell ... that the reason you doubt my memory of events early in Select is that ... mailing list; People did ask elsewhere; Some of the requests have ...
    (comp.sys.acorn.apps)
  • Re: Memory overcommit
    ... When preparing KSM unmerge to handle OOM, I looked at how the precedent ... of the same size as physical memory, ... But this is then liable to hang the system if there's some reason ... But badness() reckons ...
    (Linux-Kernel)