[Info-vax] x86-64 VMS executable image sizes and memory requirements ?

[John Reagan]

On Thursday, December 19, 2019 at 6:32:22 AM UTC-5, clair... at vmssoftware.com wrote:

> BAXTER$ dir/siz ni$:[sys$ldr]rms.exe
> 
> Directory XFEJ_RESD$:[SYS$LDR]
> 
> RMS.EXE;1                  4855
> 
> Total of 1 file, 4855 blocks.
> BAXTER$ dir/siz nx$:[sys$ldr]rms.exe
> 
> Directory XFEI_RESD$:[SYS$LDR]
> 
> RMS.EXE;1                  3883
> 
> Total of 1 file, 3883 blocks.
> BAXTER$

Here's another example of why these type of comparisons are confusing.

There are many things in the .EXE file that are not code or data.  In particular, things like debug metadata.  Not all of an .EXE is loaded into memory at initial image activation time.

On x86, we (LLVM) is putting more into the debug line table than we (GEM) do on Itanium.  Here's better breakdown

Itanium:
F11BXQP.EXE, 753,008 bytes of code, 16,490 bytes of debug line table

x86:
F11BXQP.EXE, 660,568 bytes of code, 168,754 bytes of debug data

I think we are currently asking LLVM to generate more debug data by default.  Or it just might be more accurate?

And yes, the Itanium instruction set isn't very dense.  But memory is cheap, right?  (you still have to load it from the disk however).