Subject: Re: Extremely slow printing from Mac to Linux/netatlk
From: Ron Chmara (ron@Opus1.COM)
Date: Sun Sep 10 2000 - 22:33:35 EDT
Matthew Keller wrote:
> Marc Miller wrote:
> > Or maybe it has to do with the speed of the machine of the network, and of
> > the printer.
> I'm running 100mb switched everywhere, ATM backbone, on new G4's (some
> G3's) and HP4000's. :) BTW: PPR runs at "normal" speeds, and quite
> nicely.
I didn't say anything when this popped up, I figured it would get sorted
out on it's own (and it has "flamewar" written all over it).
Since it didn't, here's some reasons _why_ there's often a speed difference
between Mac and PC printing, which is *often* unfairly stacked in one direction
or another, usually moreso in a netatalk environment.
1. Native OS printing engines. When a machine seeks out a printer to output
to, it will often use or select a printing engine (i.e., on a Mac,
Postscript or Quickdraw, on a PC, ASCII, HPGL, Postscript). Some engines
are faster than others due to their complexity and capability. This means
that a simple text page is a good measure of the simplist printing engine
available for a given printer "entity" on an OS. The performance of these
engines will also be different depending on the size and complexity of a
print job.
2. Printer queuing settings. On Macs and PCs there are various settings to
tune print drivers for foreground and background printing, print as spooled
or print after spooling is completed, etc. Since most folks don't navigate
all the way through, their settings may be differently optimized for
forground and background printing. So, in a given printer "test", a PC
which puts printing as the highest priority may be put up against a mac
where printing is at an extremely low priority. (Which is pretty close
to the respective OS's defaults).
3. Data formatting. This messes with a lot of people, who assume that
TCP/IP printing is always faster than appletalk printing. In this case,
they're usually dead wrong. TCP/IP printing on the Mac formats data for
7-bit clean data streams, usually making the job bigger, and slower,
than leaving it as appletalk (as re-parsing an entire job may add a
whole lot of time to an output job).
4. TCP/IP vs. Appletalk. First see #3. Then add in the general packet
window sizes of TCP/IP vs. Appletalk, and you come up with a whole set
of odd speed issues. 8-bit clean TCP/IP printing streams from a PC will
always be faster on 100Mb than Appletalk on that 100Mb or even TCP/IP
from a mac.
5. Print driver "features". Macs are big on features, some of which
the PC world has, some of which it doesn't. A Mac taking a simple
text document, running it through a desktop printer stream, remapping
all the data for color matching, formatting the data for a universal
Postscript PC printer driver, and than printing that, is going to be
much slower than a PC spitting ASCII out over a virtual port.
So how do you fix this, in, oh, say, a print-heavy-prepress-workflow
where print jobs _aren't_ single page text notes, but 1.2GB print
jobs? Easy. For the small print jobs from a Mac, print in the
foreground. For massive jobs, print to a postscript file, and then
send it into a hot folder, or use Appletalk/NetBeui streams on a highly
switched, highly zoned, network. I've set up high-data volume
service bureaus, who do 5-10Gb of printing per day, from 10 machines,
with 4 different *physical* network segments, to keep data pipes
available. That's 100Mb switched ports to _every_ desktop. If your
livelihood depends on print speeds, switched 100Mb to the desktop
isn't a feature, it's a requirement.
To summarize: Macs will always print much slower, *if* the test
is a simple wordpad ACSII file vs. A PostScripted teach-text file.
To truly compare speeds, you'd need to use a variety of documents,
using printers which were *designed* for their respective
environments, rather than ones that had appletalk or NetBeui or HPGL
tacked on to a "native" print mode, and you'd have to use whatever
applications are used most frequently in the environment (did the
test compare a Corel doc vs. a Teachtext doc? No? Why not? :-) )
I've seen PC's print 5-10 times slower than Macs on the basis of
8-bit vs. 7-bit data on Appletalk.... but that's not an issue
of raw print capability, that's an issue of a stacked test. If your
numbers are closer to PC's being 120-130% faster (that's 10 seconds
vs 12-13 seconds), that's when you are seeing the age of Apple's
print architecture. Much more than that, and it's a stacked test,
where you need to take a closer look at why you're getting those
numbers.
-Bop
-- Brought to you from iBop the iMac, a MacOS, Win95, Win98, LinuxPPC machine, which is currently in MacOS land. Your bopping may vary.
This archive was generated by hypermail 2b28 : Wed Jan 17 2001 - 14:32:08 EST