| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 381.1 | | MU::PORTER | what's in a name? | Fri Mar 10 1989 22:06 | 21 |
| 4) No
3) Don't know
2) No
1) Seems ok to me (well, it was a vague question!)
Because DECwindows is X windows.
Or, to be more specific, DECwindows is our implementation of the
X window system, plus our own toolkit for the development
of applications.
Is DEC's implementation of X higher performance than someone
else's implementation of X? Maybe - I once saw some paperwork which
claimed that we'd done a good job in improving performance.
|
| 381.2 | | DECWIN::FISHER | Burns Fisher 381-1466, ZKO3-4/W23 | Fri Mar 10 1989 22:23 | 22 |
| Let's keep our terminology straight please...DECwindows is one of
several implementations of windowing systems based on the X11 protocol.
Your question is sort of like, "Is VAX Fortran better than Fortran?"
I presume that what you are really asking about is DECwindows as
compared to MIT's sample implementation of X. In that case, part of
the answer is, they both use the same graphics protocol, so one is
no more nor less efficient of ethernet bandwidth than the other.
However, there is more to it. For example, which set of widgets makes
the most efficient use of the protocol, which display server is most
efficient, etc. I can't answer these questions. You should go through
the standard sales support channels to find the approved answers.
As to local vs remote: This is not an official answer; I have not
measured anything. However, you would be surprised at how little
difference there is between local and remote. In fact sometimes remote
is faster (you have two cpu's working on your behalf, not just one!)
Burns
|
| 381.3 | I vote for remote... | POOL::CLABORN | There's no "'" in the possessive its | Tue Mar 14 1989 09:07 | 18 |
| Again, a very subjective answer in the same vein as Burns':
I use DECwrite quite extensively (a very intense application from a CPU and
memory point of view), and wouldn't dream of running it locally on my 9Mb
GPX. I get to use the 128 Mb of our 8800 boot node, and its CPU power as well
so image activation and response is extremely crisp.
About the only thing I don't run remotely (if at all) is Paint. Because all this
application is is essentially tracking the mouse, communication between client
and server is very intense. Run remotely, it's "ok" but there is a noticeable
lag in action as the client plays catch-up with the mouse. Run locally, it's
much crisper.
However, I feel that this perceptible "lag" when running remotely is the
exception, not the rule; especially when you can exploit the resources of a
large remote node.
- George
|
| 381.4 | What resource was that? | POOL::HALLYB | The Smart Money was on Goliath | Fri Mar 17 1989 16:46 | 16 |
| .0> Several software vendors on the DECstation 3100 have mentioned to
.0> my customer that X-Windows is a resource intensive hog, which will
.0> effect the performance of the networked workstations.
That's "affect", not "effect". Anyhow there is some truth in this
statement as you've phrased it. X-based windowing systems are big
memory hogs. No doubt about it. But MEMORY is the porcine resource.
A recent test run in my area used a VS2000 running DECwrite remotely,
with an HP LAN analyzer. For what it's worth, one remote user used
about �% of the Ethernet, over half of _that_ for remote paging.
Even though that's a rather limited study I suspect you'll find
similar results for most interactive operations.
John
|