Sunday, September 06, 2009

Pay No Attention To That Last Post

My last blog post may have expressed some negative views toward software upgrades:
I know I'll take some slack for this, but: why did you all go and grab an OS update so fast? What have you gained? The down-side is lost time and application compatibility problems...was it worth it?
That blog post was of course not my real opinion, but rather a recreation of the evil Ben. Always upgrade your software to the newest version whenever it comes out!


krz said...

gg pissing of the customers and then playing smart about it? uhm im not so sure anymore what this blog tries to achieve

Paploo said...

I think you had some good points before Ben, but I think the truth lies somewhere between the extremes.

With each update comes new fixes, new features, and new bugs. Major updates also come with the risk that some part of the software that you depend on breaks. This can be *really* bad for some users; which is why I recommend that most people wait a month or two before upgrading.

On the other hand, some users (like myself) are developers and power users, and often see a bigger benefit from the new features than the problems we face for those first few weeks while everything gets ironed out.

In the end, though, there has to be some group of users that adopts the updates first. If *everyone* waited, then it would just slow the march of progress without a single benefit to anyone. If *no one* adopted, then progress would stop. And while change can be scary, I seem to think that there is agreement that computing is in a better place now than it was 10, 15, and especially 20 years ago.

Benjamin Supnik said...

Hi Paploo,

You're absolutely right -- the point of the post was really only to draw attention to the hidden assumptions of software economics...all software decays to "free" so companies intent on selling things are dependent on being able to identify and sell something better.

I became aware of the problem with this in my previous work with Avid/DigiDesign - they sell very high-end (and very nice) audio and video editing systems to "industry" (TV, recording studios, etc.)

The problem is that they hit feature saturation...once you have 128 channels of audio and broadcast quality picture, you don't really _need_ a new system, and it became very hard for the company to sell new units into a saturated market at their standard price point.

As I've ranted about in the past, operating systems have reached that point. Apple is at least recognizing that by selling an OS for a lot less than they used to. To me this makes sense - there isn't that much exciting that the new OS can do. It's at best nice to have, not must have.

Microsoft has a get-out-of-jail free card this once: Vista is so bad that people will pay full price to escape. But if Windows 7 is even remotely acceptable, it'll be the last full price OS they sell.

In the case of X-Plane we've been lucky so far: users have (so far) had an unending appetite for a more visually realistic flight experience, and that ties directly into hardware that is still following Moore's law. Those things combine to mean that we can find ways to make the next X-Plane "pop" more visually than the last.

(It would be interesting to see a montage of final-version X-Plane's, e.g. 599, 670, 763, 863, etc.) each at their max settings...)

Dave Duck said...

Heck, I agreed with your original comment. I'm sitting this OS out until a compelling reason shows up to get it.

For one thing, I'll be damned if I'll upgrade to CS4 just because Adobe is pushing it as the "solution" to a possibly broken CS3.

For another, 10.5.8 on my MacBook works brilliantly. It's taken me a couple of decades, but "new" is not the catnip it once was to me. On almost any front.

Edward Glowworm said...

Hiya Ben,

I agree with your curiousity about why one would upgrade the OS (and possibly be disappointed by the X-Plane performance).

OK, so I wanted to understand the performance difference of OS and
architecture (in order to find the optimum performing X-Plane on my rig), so I've did some testing with 32-bit and 64-bit version of Ubuntu 8.10, 9.04 and 9.10, together with XP 9.22, XP 9.31 and XP 9.40. (Even tried Win7 Enterprise w/ XP 9.40). Basically each OS is a fresh install and rebooted prior to the test of a given XP. No further tweaking performed prior to test. nVidia driver come in different
versions, and default settings have been used (e.g. Anti Aliasing
settings are not forced, and handled by application).

Detailed results of 45 benchmarks and rig/driver info, here:

I used following command to run the standard bench mark in terminal (run in Gnome) and to show the results in the terminal:
./X-Plane-i686 --fps_test=1; cat Log.txt | grep "FRAMERATE TEST"

(my) Conclusion
(sorted by FPS hi-lo):

Avg_FPS XP OS Fastest by OS or X-Plane
216 9.22 Ubuntu-810-32 2.6.27-15 >> Fastest XP 9.22 + Fastest Hardy
215 9.22 Ubuntu-904-32 >> Fastest Jaunty
215 9.22 ubuntu-910-32 2.6.31-14 >> Fastest Koala
208 9.22 Ubuntu-904-64 2.6.28-16 >> Fastest Jaunty-64
185 9.22 Ubuntu-910-64 >> Fastest Koala-64
182 9.40 ubuntu-910-32 2.6.31-14 >> Fastest XP 9.40
181 9.31 Ubuntu-904-64 2.6.28-16 >> Fastest XP 9.31
179 9.40 (2) Ubuntu-810-64 2.6.27-15 >> Fastest Hardy-64

95 9.40 Win7 Ent-64 no Aero
88 9.40 Win7 Ent-64 Aero >> Slowest Win

(2) indicate the 2nd time I've run the test.

I did the test in Xterm as well, I noticed 1 FPS improvement, so I didn't bother to test further in a console only environment.

I'm somewhat surprised there is no real linearity between the various Ubuntu distros and XP versions. 1 version of XP could perform well in 1 distro and suck in the next. One thing that get the highest results seems to be XP 9.22. (Was this because the graphics format changed from 9.22 to 9.31?)

Anyways, if there's no compelling reason in X-Plane versions I tend (understatement) not to upgrade, same for OS. My X-Plane rig is only to drive the flight simulation, and nothing else (I use Mac for regular computer needs.)


Benjamin Supnik said...

When comparing x-plane versions you might want to try --limited_glsl in 940 vs regular 922. That would eliminate the role of quality improvements in the shaders that aren't carefully regressed in the settings of the fps test.