[Simh] EXT : newbie question about simh performance

[gérard Calliet]

Hello,

Thanks for all the answers. I do a grouped answer.

First:
MicroVAX 3900 simulator V4.0-0 Beta        git commit id: 8810571d
sim> show version
MicroVAX 3900 simulator V4.0-0 Beta
          Simulator Framework Capabilities:
                  64b data
                  64b addresses
                  Ethernet support
                  Idle/Throttling support is available
                  Virtual Hard Disk (VHD) support
                  Asynchronous I/O support
          Host Platform:
                  Compiler: DEC C 7.3-289
                  Simulator Compiled: Dec  4 2013 at 15:29:58
                  Memory Access: Little Endian
                  Memory Pointer Size: 32 bits
                  Large File (>2GB) support
                  OS clock tick size: 1ms
                  OS: OpenVMS I64 V8.4
          git commit id: 8810571d

Next (technical):

* [Hittner] Memory disk (LD) : I think about it, but I see the compiler 
doesn't do much IO itself (seen in the emulated VAX). But about 100% of CPU.
It seems it is simh itself (on the Itanium host) which does huge number 
of IO to do its cpu job.
I will test also the memory disk.
I thought that perhaps installing the simh image should help, but I 
cannot see any paging on the process running simh.
The core running simh is full used. There is however a  lot of Interupt 
Stack Mode activity.

* [Pizzolato,Billquist] 512m : To do that, I have a huge Working Set for 
the process running simh. No paging seen.

* [Pizzolato] asynchonous I/O : I see my version support it. Is it 
necessary to activate it with a parameter?

* [Khandy] set cpu_idle = VMS : I will try it.


Next (philosophy):

* [Hittner] compiler port: It is possible to port the compiler. It is an 
old Ada cross compiler. The issue is we need to create exact bit and 
byte images cross-compiled for 68k cpu equipments. If we port the 
back-end, the images will be functionnaly equivalent, but not bit and 
byte equivalent.

* [Hittner] i7 will outperform: Yes. But with VSI help, in the future, 
OpenVMS will become more usable to host anything (at least for all VMS 
specific hardware versions), and OpenVMS will run on standard x86 
hardware. For some customers, it is very important to be able to do 
anything on OpenVMS. Some projects need very long time integration 
cycles, and OpenVMS is about the only OS which yet knows about strong 
backward compatibility. And with VSI new company, we can see at least 
ten years of availibility in the future. So I want to be able to do 
anything on some projects only on OpenVMS, if possible. simh on OpenVMS 
here can do the work, because it seems the tuning for only a compiler is 
possible. And the need for pure VAX is only a few part of the global 
project.


Thanks,

Gérard Calliet

---
L'absence de virus dans ce courrier électronique a été vérifiée par le logiciel antivirus Avast.
https://www.avast.com/antivirus