Re: Good OpenVMS presentation (slides # powerpoint)

DDR-II and PCI-E are both transparent to users except perhaps for
performance. Just as NUMA is generally transparent at the user application
level except for performance.

"JF Mezei" <jfmezei.spamnot@xxxxxxxxxxxx> wrote in message
news:43D7F09D.F68921E5@xxxxxxxxxxxxxxx
> Robert Deininger wrote:
> > BTW, DDR-II and PCI-E are both in the pipeline for Integrity systems and
VMS.
>
> Does VMS *need* to be aware of the memory architecture ?
>
> I realise that in the case of GS/Wildfire class machines, knowing about
> the intimate interconnect design can help VMS take advantage of the
> memory system and avoid bottlenecks and increase performance, but woudl
> it be strictly necessary if you just had independant instance(s) of VMS ?
>
> Also for regular systems, would VMS be aware of the memory system at all
> ? Wouldn't that be all hardware with VMS seeing just one large
> continuously accessible memory ?


.

Relevant Pages

  • Re: Good OpenVMS presentation (slides # powerpoint)
    ... Just as NUMA is generally transparent at the user application ... But for uniform memory access, does VMS ...
    (comp.os.vms)
  • Re: Good OpenVMS presentation (slides # powerpoint)
    ... Does VMS *need* to be aware of the memory architecture? ... the intimate interconnect design can help VMS take advantage of the ... would VMS be aware of the memory system at all ...
    (comp.os.vms)
  • RE: Multicore Is Bad News For Supercomputers
    ... While VMS may have had some NUMA-aware logic back when it was on Alpha, ... capability relevant to the "new" NUMA that CSI will bring? ... Kerry Main ...
    (comp.os.vms)
  • Re: Multicore Is Bad News For Supercomputers
    ... While VMS may have had some NUMA-aware logic back when it was on Alpha, ... is this capability latent in the IA64 version of VMS? ... capability relevant to the "new" NUMA that CSI will bring? ... what happens when you have a NUMA hardware that runs HU-UX upon ...
    (comp.os.vms)
  • Multinets _BG: Driver and IPsec
    ... Multinet already advertises this functionality for VMS. ... Is it truly transparent to the application/socket programmer? ... Why didn't they do the same thing with TCPware? ... Process Software would rather die then allow a behavioural inconsistency ...
    (comp.os.vms)