2011 VERY slow lisp performance!

[Opie]

Have you thought about turning off the REGENMODE variable? I'm not certain this will solve your problem, but it is something else to check. Other than that, you may have to look into optimizing your code.

[marko_ribar]

INTEL CORE QUAD CPU Q8400 (4x2,66GHz)
4GB RAM
Windows 7 Ultimate

-AutoCAD 2008 (invisible) : 00:00:07:722
-AutoCAD 2008 (visible) : 00:00:08:740
-AutoCAD 2011 (invisible) : 00:00:12:100
-AutoCAD 2011 (visible) : 00:00:30:870

M.R.
It seems that you're in wright... Something here is wrong with ACAD 2011... I even used update 1.1 for AutoCAD 2011 x64
The reason for these slow performance lies because CMDECHO is set to 1 (on)

[ccowgill]

yes the update installed while setting up Autocad for the first time.
can you please do me a favor and run the code below and report the elapsed timer results that you will get. I'm executing the code on a p4 3.2 ghz computer,

autocad 2008 execution invisible: Elapsed timer (on): 0 days 00:00:00:953
autocad 2008 execution visible: Elapsed timer (on): 0 days 00:00:01:062
autocad 2011 execution invisible: Elapsed timer (on): 0 days 00:00:01:219
autocad 2011 execution visible: Elapsed timer (on): 0 days 00:00:19:531

I'm running 2011:
Elapsed timer (on): 0 days 00:00:02:460 invisible
Elapsed timer (on): 0 days 00:00:02:440 visible
_VERNUM = "E.115.0.0 (UNICODE)" (read only)
AMD Phenom X4 (3.4ghz), 8GB Ram, Win 7 Ultimate

sorry, I dont have 2008 to test in.

[BlackBox]

Intel Core i7 (4x2.67GHz)
4.0GB RAM
Windows XP Pro

AutoCAD Land Desktop 2009 (Invisible) Elapsed timer (on): 0 days 00:00:01:047
AutoCAD Map 3D 2009 (Visible) Elapsed timer (on): 0 days 00:00:12:703
AutoCAD Civil 3D 2011 (Visible) Elapsed timer (on): 0 days 00:00:12:484
AutoCAD Map 3D 2011 (Visible) Elapsed timer (on): 0 days 00:00:05:250

[dgorsman]

yes the update installed while setting up Autocad for the first time.
can you please do me a favor and run the code below and report the elapsed timer results that you will get. I'm executing the code on a p4 3.2 ghz computer,

autocad 2008 execution invisible: Elapsed timer (on): 0 days 00:00:00:953
autocad 2008 execution visible: Elapsed timer (on): 0 days 00:00:01:062
autocad 2011 execution invisible: Elapsed timer (on): 0 days 00:00:01:219
autocad 2011 execution visible: Elapsed timer (on): 0 days 00:00:19:531

A) P4 processor, 3 GHz or otherwise, is going to be noticeably slower with AutoCAD 2011. Period. More so if you have a more recent OS such as Vista Win 7. And more so if you are not maxed out on RAM. If you haven't started saving for new hardware get started NOW.

B) Check all your PLINE system variable settings, you may be running with "heavy" PLINEs enabled.

[marko_ribar]

I optimized perf.lsp for a little faster performance (on my machine it's now twice faster then before in slowest mode). Sysvar REGENMODE does not influence on performance on my machine...

- AutoCAD 2011 visible (around): 00:00:12:500
- AutoCAD 2011 invisible - below 1sec
- AutoCAD 2008 visible - below 1sec
- AutoCAD 2008 invisible - below 1sec

(defun c:perf()
  (setq osm (getvar 'osmode))
  (setq cmde (getvar 'cmdecho))
  (setvar 'osmode 0)
  (setvar 'cmdecho 0)
  (vl-cmdf "'time" "r" "")
  (setq i 0 p0 (list 0 0) p1 (list 0 1000))
  (repeat 1000
    (setq p0 (list (* i 150) 0) p1 (list (* i 150) 1000) )
    (vl-cmdf "pline" p0 "w" "10" "10" p1  "" )
    (setq i (1+ i))
  )
  (setvar 'osmode osm)
  (setvar 'cmdecho cmde)
  (vl-cmdf "'time" "d" "")
(princ)
)

M.R.

[Serhan_BAKIR]

Hi,
thanks all for your feedbacks! I've written that benchmark code in 5 minutes just for preliminary tests. Moreover I repeated the process not only with polylines but also with dimension and text objects (with same poor performance).

that simplistic object generation approach gave us various results. it is really interesting to have longer execution times with newer cpus. and I assume that better hardware doesn't always mean improved results. since my p4&2008 couple gave shortest time which is very odd!!

I see that reasonable results are only obtained with 2011 by Mr. "CCOWGILL" with his phenom cpu. so it seemed to me that speed of execution is not solely dependent on cpu but more on system settings. thus I wrote a script to compare system variables between my 2008 and 2011 versions. with SYSVDLG command I got an output of sys.var. file (.svf)

2011 had 687 variables while 2008 had 480.
in my setup I have 42 variables which differentiate among these versions.
reading the manual for everyone of them, relevant ones seemed like:
shortcutmenu, selectionarea, regenmode, grips, gridmode, griddisplay, dimzin

changing those variables didn't give me any different result in execution speed. I also tried hardware acceleration on/off (which slowed me down 1sec. while on) so I conclude that sys var differences between "versions" don't play a role on this problem. I have equaled them all and nothing changed.

Now I need to compare sys.var.s between same versions. So Mr. CCOWGILL I will be very happy to compare my setup with yours since your results are consistent and fast (visible/invisible). Can you please attach your svf file if it is also appropriate for you.

ps: I'm attaching svfcomp.lsp so you can make a comparison if you ever need it.

[marko_ribar]

SVF file won't help... I compared 2008 & 2011 with your lisp and difference is only in, by my opinion, unessential things for lisp execution performance... Only thing that will help to run lisps faster in ACAD 2011, also by my opinion, is to set zoom to very small factor, and in the end of lisp to set to zoom extents... I've noticed that if zoom factor is larger, performance is lower and inverse if zoom factor is smaller performance is greater... I don't know why is this, but ACAD 2011 has more commands and variables and it's bigger program than ACAD 2008... I also tested perf.lsp on ACAD Architecture 2010 and it's performance is somewhere in between these two, and it is also big program like ACAD 2011...

M.R.

[Serhan_BAKIR]

yes you are right, differences between system variables (2008 vs. 2011) do not effect lisp execution. Zooming in and out effects visibility therefore has influence on performance, as you stated.

I assume, comparing system variable differences between same versions, such as comparing Mr. CCOWGILLs sysvar file with mine can give some clue about why his autocad is faster than mine.

[marko_ribar]

I assume, comparing system variable differences between same versions, such as comparing Mr. CCOWGILLs sysvar file with mine can give some clue about why his autocad is faster than mine.

Maybe Mr. CCOWGILL haven't zoomed out as we did (he only wrote that plines were visible), but I still don't know how his time is faster when visible, probably he only panned drawing to be invisible and times were almost same...

BTW, when you serhanbakir298323 used perf.lsp that you wrote, was your CMDECHO variable turned on or off? I still have undesirable slow times when CMDECHO is on (1)...

M.R.