Conference bulova::decw_jan-89_to_nov-90

I don't understand. Did you get this performance problem rumor from "official
channels"? Or is it just based on hearsay? I believe the VMS X server has
great performance compared to other vendors. Do you have hard facts backing
up this statement?

Re .1:

Nope, certainly NO hard fact; that's what I'm looking for.  To be
clear, this note  is  explicitly  NOT  to  report  a problem with
performance, but to receive  some  hard facts that will hopefully
prove the rumors to be incorrect.

Yet, although I don't have  hard  facts,  I  do keep running into
cautions    regarding   performance  (at  least  in    the    VMS
implementation)  and  would like to get other  peoples'  opinions
regarding that as well as the ULTRIX implementation.

I'm looking for information.

Jeff

    The VMS implementation works very well. Performance on standalone
    systems with less than 8mb can vary, careful tuning will improve
    matters,
    
    - A

As far as I know the VMS and ULTRIX implementations have
similar performance, both of which are high quality
for current X-11 implementations.

In the early baselevels there were rumors which escaped the restricted
few folks with DECwindows baselevels that VMS DECWindows had poor
performance. This was despite and in direct contradiction to stated
policy regarding disclosure outside the restricted baselevel
list/engineering proper. When x11 was released from MIT the very first
VMS implementation BASELEVELS which ran were somewhat less then
lightning fast in many respects. There are STILL a number of folks in
and outside of engineering who have not CLOSELY followed the progress of
the product who upon occasion have contributed to the perpetuation of
this RUMOR.

I can personally attest that there is virtually NO perceptible
difference in the performance of the Ultrix DECWindows implementation
.vs. the VMS implementation. In addition there are certain areas in
which the VMS implementation seems to outperform it's Ultrix
counterpart. The converse is true however, and I would not hesitate to
implement on either platform based on any discussion of DECWindows
performance.

Ultrix has a set of beneficial attributes in regard to system
services/interprocess techniques, and so does VMS. As far as the
decision to move towards one or the other, my vote would be to attempt
to develop PORTABLE CODE. Code which runs on both is not overly
difficult in the sense that there is a basic set of one for one
relationships between system services on both platforms. There are
exceptions but there are already a host of examples of portable code
here and in the DECWindows examples conference on ELKTRA.

One final note : WE ARE SELLING AND MARKETING INTEROPERABILITY BETWEEN
VMS AND ULTRIX PLATFORMS. REGARDLESS OF THE HOLES WHICH REMAIN WE HAVE
COMMITTED TO TRUE PEER COMPUTING BETWEEN VMS AND ULTRIX. EACH INTERFACE
HAS A VALUE AND BENEFIT, AND NEGATIVE RUMORS TEND TO ESCAPE NOT ONLY THE
RESTRICTED COMMUNITIES WITHIN DIGITAL, BUT THE COMPANY AS WELL. I WOULD
PROPOSE THAT YOU CORRECT WHOEVER TOLD YOU THAT THERE MIGHT BE
PERFORMANCE PROBLEMS WITH THE VMS IMPLEMENTATION OF DECWINDOWS, AND MAKE
YOUR DEVELOPMENT DECISION BASED ON CRITERION OTHER THAN PERFORMANCE
(I.E. WHAT DOES THE CUSTOMER WANT, NEED, WHICH INTERFACE ARE THEY
FAMILIAR WITH, ETC, WITH A GOAL OF PORTABLE CODE).

	Regards,

		Mike Weaver

    One more testimonial, for what it's worth: Over the past two months I
    have used *both* Ultrix and VMS DECwindows, every day, on 9 Mb GPXs. 
    There is no perceptible difference in performance.