PDA

View Full Version : LW [8] Slower than 7.5c



Exception
05-15-2004, 06:04 AM
Ok, I just received [8] and i was very happy to get it. I bought it although there is only one (honestly) thing in it that I find usefull over 7.5: render speed.
This turns out to be false. LW[8] is not faster than 7.5c at all. In fact: it is slower. Not by much, but those 'render enhancements' as stated in the features list turns out to be not true. I addressed the fact that for the architectural visualisation community lw[8] had nothing to offer despite our many requests before, and Chuck reply was: well you wanted faster rendering, and lw[8] is faster. Well it is not. What does newtek have to say about this misadvertisement?
What did I buy LW[8] for? stability? Newtek should have fixed 7.5c first. im sure there a many people out there that did not upgrade.

I tested scenes from the new lightwave distribution:

Ninjasample.lws:
7.5c: 1m 6 sec
[8]: 1m 8 sec

Sunsetsample.lws:
7.5c: 3.1 sec
[8]: 3.7 sec

Moonbase.lws:
7.5c: 13m 42s
[8]: 14m 5s

and from the Old lightwave (7.5c) distribution:

Glass & cigarettes:
lw7.5c: 36.4 sec
lw[8]: 41.5 sec

radiosity_reflectivethings.lws:
lw7.5c: 48.5 sec
lw[8]: 49.1 sec

Tracer-NoRadiosity.lws:
lw7.5c: 5m 53 sec
lw[8]: 6m 24 sec

All tested on a P4 3.2 Ghz, 2GB of RAM, fair testing performed with all settings identical. Tests have been repeated to verify.

mrunion
05-17-2004, 08:25 AM
By default my render options was set to 4 threads instead of one. For simple scenes one thread is supposed to do better usually. You may want to check the number of threads and try again.

G3D
05-17-2004, 11:33 AM
I have noticed the same slowness, and the number of threads don't make much difference on my dual dell system. Haven't tried out the macs yet. Everything else is very nice, though.

Exception
05-23-2004, 11:26 AM
Of course I made sure the number of threads was the same. I tested with all settings identical, as stated in the first post.

[edited by the moderators]

Hervé
05-23-2004, 11:30 PM
That pushes the need for F_Prime supporting all shaders even more then.... he he

digimassa
05-24-2004, 02:19 AM
[removed by the moderators]

Aegis
05-24-2004, 02:37 AM
I dont understand why none really cares about this, they have been lying to us, and dont even respond to this message.

Well, to be honest I hadn't really expected any attention to be paid to the renderer with this update so the fact that [8] is slightly slower comes as no surprise to me - I guess the added animation features and dynamics come at a price - even if you're not using them...

theo
05-24-2004, 06:03 AM
The only thing that I can think of concerning the LW claims is that maybe the renderer has been enhanced in a very unobvious way.
In other words. code-level changes have been adopted to increase rendering efficiency but not in a way where we will all notice a dramatic difference in speed per se.

I do think though that this section of Lightwave's marketing needs to be toned down a bit due to this almost unnoticeable difference in speed.

The renderer, and it could be me psychologically coming to this assumption due to a bias towards LW, does seem to be slightly better, but not in a way I can put my finger on which may be due to what I said in my first paragraph.

wacom
05-24-2004, 10:10 AM
There was a post of someone using huge shadow maps on a very large scene, and he said the render time changes were huge from 7-8. Eight was much faster.

If you're sweating a few seconds you really need to support Worley and start shave'n MINUTES and HOURS off of those render times.

Still NewTek needs to tell us WHERE the renderer is faster/better.

redlum
05-25-2004, 09:37 AM
not wishing to start a flame war . . .

However I am using an old 400 mhz G4 with a 64 meg gforce 2 card and 1 gig of ram. I have found LW 8 to run very smooth and is so much faster than 7.5c. I've only had it wink out on me once. Rendering is a lot faster too.

theo
05-25-2004, 09:50 AM
Redlum... Are you just pushing around simple models with one texture and a specular highlight? How many hours and days in a continous cycle is the program being used? How many motion paths and what kind of special effects are being employed in terms of columetrics or PFX or dynamics? Just to post what your results are isn't that simple friend- what are the qualifications of the statement.

Trust me on this- the scene I am working on right now would freeze your entire system much less create a determinable environment as to the stability of LW8.

What you need to do is load up the heaviest scene your system can handle, play with that for about ten hours then report back.

Exception
05-25-2004, 10:17 AM
Theo: that would be a nice idea, you know, saying that some render improvements in some specialist field like 10000x10000 sized shadow maps will be speeded up or whatever, this does not warrant the render enhancements and render speed enhancements on the product features list, and CERTAINLY not when the rendering is actually slower in all normal circumstances, including newteks own benchmark tests.

Here you have something to read that Chuck told me after my notification that lw[8] had nothing to offer the architectural visualisation community whatsoever and that our threads, which has been circling this board constantly for a long time now were blatantly disregarded.

I quote from Chuck (readable in this thread : http://vbulletin.newtek.com/showthread.php?s=&threadid=21261&perpage=15&pagenumber=5 ):


"If I'm not mistaken, faster rendering was among the requests of interest for the Architectural community, and we have been able to provide some improvements in first release of [8]."

Well I guess thats not true is it?

And furthermore after I said this:



I wouldnt mind paying 100 or 200 dollars more for lightwave, really, and I think many people will agree with me, if it only would have the features everyone's been asking about. It really comes down to those few features and stability that makes us users dissatisfied.

Chuck replied this:



I think you'll find LW[8] a step in the right direction.

Step into what right direction? The direction of Newtek's wallet?

So this enahncement in a very unobvious way as Theo states is is not valid since it is projected as of use for the entire arch. vis. community. Nope, we have to deal with yet another few minutes added to our already astronomical render times.
These features were claimed before the release, and put on the product features list, the only thing that I find any use for, but no it turns out to be a lie. and no one from Newtek even takes the effort to reply to this thread which has been around for long enough to reaf for everyone who takes the least notion of whats going on here.
As to supporting worley, that is all fine with me, but if nwtek hadnt claimed this render speed increase then we would have bought fprime instead of the 8 upgrade.
That would leave us with a 7.5c which is a mess of a program, bugs flying out of it on every corner, and its tech support taking months to answer.

Indeed I AM very angry!

Dave Davies
05-25-2004, 10:42 AM
Ninja's - 7.5c = 26.1 seconds, 8.0 = 28.8 seconds. By my reckoning, that's almost a 10% increase in speed. Fairly respectable, I'd say.

Dual 2.0Ghz, 1GB RAM

Dave Davies

Lynx3d
05-25-2004, 11:07 AM
Err...Dave, an increased rendertime means decreased speed...
or did you mix up the numbers?

And well, perhaps ons should benchmark a scene that takes at least a few minutes, possibly the renderer spends more time to setup/optimize the scene to speed up raytracing or whatever in the long run.
Unfortunately, i STILL DIDN'T GET MY LW8!!

Dave Davies
05-25-2004, 11:20 AM
DOH! Brain fart. (Although it may explain why I drove all the way from town in reverse!)

I'm embarrassed! I'll shut up now.

Dave Davies:

theo
05-25-2004, 12:47 PM
Exception- I think you need to relax friend. I don't disagree with you if you'll read my post properly. I was just agreeing in a less critical way of Newtek, that's all.

I, too, wanted to see noticeably faster rendering and up to this point I can say that it could be better. I just didn't want to knock Newtek too much in this thread because I like them.

redlum
05-25-2004, 12:49 PM
Originally posted by theo
Redlum... Are you just pushing around simple models with one texture . . . .

What you need to do is load up the heaviest scene your system can handle, play with that for about ten hours then report back.

I don't know if this is the heaviest scene my system can handle, but it's the largest one I've built with it so far.

Here is what's on the statistics window.

Points - 91182
Polys - 102596
Surfaces - 159
Images - 8
Lights - 5
Camera - 1

here's a fly through (http://homepage.mac.com/dmazanec/movies/warehouse.htm)

and here's what it looks like (http://homepage.mac.com/dmazanec/art/warehouse.jpg)

Exception
05-26-2004, 05:51 AM
Theo I didnt mean any hardship towards you. Im sorry if you took it that way. [edited by the Moderators]

Lynx3d: The last scene I rendered was more than 5 minutes.
here I post it again:

Tracer-NoRadiosity.lws:
lw7.5c: 5m 53 sec
lw[8]: 6m 24 sec

Which is also an increase of almost 10 percent.

If they speeded things up with 10 percent, that would be fair, but slowing it down?!

digimassa
05-26-2004, 06:09 AM
:cool:
and still no commnet from NT on this thread; this the way how to care for the sorrows of paying customers^^

Exception
05-27-2004, 07:34 AM
Yep its getting more and more obvious. In every department I am afraid.

Exception
05-27-2004, 07:41 AM
Launched a formal Customer Support inquiry into this matter. See what they have to say if rubbed under the nose instead of given the chance to discuss the matter here.

Chuck
05-27-2004, 01:15 PM
As I've mentioned elsewhere, I and everyone else at NewTek who would normally be participating in the forums are working on tight deadline projects and we are spending little to no time in the forums at present. Patience with this situation would be appreciated, but if anyone is frustrated that's understandable and we certainly regret that we can't be as present at the moment.

I've directed the attention of the development staff to this thread and queried as to what the status should be of "faster rendering" in regards to the feature list. Marketing receives the list of features from Development and does not depart from that or embellish it, and Development does not list features unless they've put them in.

I also took a look around for a thread where the user had done some architectural renderings with radiosity and gotten faster rendering times with [8] than with [7.5], but I can't locate the post. If someone could and could link to it from here that would be appreciated.

digimassa
05-27-2004, 02:30 PM
:cool:
Chuck, its nice to hear your words here,
but kinda amazing is, that you have to look into threads to know about rendertimes of LW with or without radiosity^^

Chuck
05-27-2004, 02:34 PM
Originally posted by Exception
Theo: that would be a nice idea, you know, saying that some render improvements in some specialist field like 10000x10000 sized shadow maps will be speeded up or whatever, this does not warrant the render enhancements and render speed enhancements on the product features list, and CERTAINLY not when the rendering is actually slower in all normal circumstances, including newteks own benchmark tests.

Here you have something to read that Chuck told me after my notification that lw[8] had nothing to offer the architectural visualisation community whatsoever and that our threads, which has been circling this board constantly for a long time now were blatantly disregarded.

I quote from Chuck (readable in this thread : http://vbulletin.newtek.com/showthread.php?s=&threadid=21261&perpage=15&pagenumber=5 ):



Well I guess thats not true is it?

And furthermore after I said this:



Step into what right direction? The direction of Newtek's wallet?

So this enahncement in a very unobvious way as Theo states is is not valid since it is projected as of use for the entire arch. vis. community. Nope, we have to deal with yet another few minutes added to our already astronomical render times.
These features were claimed before the release, and put on the product features list, the only thing that I find any use for, but no it turns out to be a lie. and no one from Newtek even takes the effort to reply to this thread which has been around for long enough to reaf for everyone who takes the least notion of whats going on here.
As to supporting worley, that is all fine with me, but if nwtek hadnt claimed this render speed increase then we would have bought fprime instead of the 8 upgrade.
That would leave us with a 7.5c which is a mess of a program, bugs flying out of it on every corner, and its tech support taking months to answer.

Indeed I AM very angry! [/QUOTE]

You seem to be upset with me in particular, and if so, my apologies for upsetting you, but in proper context it would be clear what my remarks meant and your presentation of them ignores that context and meaning in favor of being able to use the bit quoted as a setup for a denigrating and sarcastic comment. Yes, I think LightWave [8] is a step in the right direction for all LightWave users but I certainly acknowledge, as do all of us here at NewTek, that how much of a step depends on your primary usage. Please rest assured that there will be extensive improvements in those areas which were not emphasized in 8.0 as we proceed in the 8.x cycle.

Development has confirmed that the performance should be faster than the results you are getting, that they have identified the issue that has resulted in the expected performance increase not being present in the current version, and that they expect to resolve the issue before LightWave [8] officially ships.

Chuck
05-27-2004, 02:41 PM
Originally posted by digimassa
:cool:
Chuck, its nice to hear your words here,
but kinda amazing is, that you have to look into threads to know about rendertimes of LW with or without radiosity^^

I don't think it should be surprising that I would want a set of outside reports to discuss with the development staff - but hey, I could be wrong! :)

Lynx3d
05-27-2004, 03:04 PM
Hm, Chuck, i can't find that thread about that architectural test either...just can't remeber where or when it was posted, probably in one of those 20 page long threads :)

but found another person getting faster renderings from LW8:
http://vbulletin.newtek.com/showthread.php?s=&threadid=21386&perpage=15&pagenumber=4

There were some other people reporting improvements too and some reporting mixed results...

Anyway, to me it seems Newtek is communicating more than in the past...however i haven't got an answer to my email about a SDK issues yet...
really, i do want to help with my modest possibilities... :)
But i guess you now are even more busy than before the first shipped LW8 CD...

Exception
05-28-2004, 06:24 AM
[edited by the Moderators] ...it is very upsetting to pay for a product which does not do what it promises, especially if it makes the purchase of the product senseless.

I eagerly await a response from Newtek in regard to this matter.

theo
05-28-2004, 06:54 AM
Sounds like Newtek needs a few more bees in the hive.

Chuck
05-28-2004, 08:51 AM
I eagerly await a response from Newtek in regard to this matter.

Exception, that was in my message; apologies, my preceding comments in that post may have distracted you such that you missed the information.


Development has confirmed that the performance should be faster than the results you are getting, that they have identified the issue that has resulted in the expected performance increase not being present in the current version, and that they expect to resolve the issue before LightWave [8] officially ships.

Chuck
05-28-2004, 11:19 AM
To repeat:

Development has confirmed that they have identified the issue in the version released for pre-upgrades, and expect to correct the issue prior to the official release of LightWave [8].

Any rendering using raytracing options should be faster in [8] than in [7.5c]. P4 systems should render about 5% percent faster, with P3 and Athlon systems seeing even larger gains, up to about 20% for Athlon in some situations. Speed gains will vary depending on a number of factors, however, so please bear in mind that YMMV.

A caveat is that errors in the Caustics and some other routines were fixed, with the result that while caustics and some other elements now render without the problems the errors were causing, they also render slower. The team felt that "slower and as intended" was the better option as opposed to "faster and not exactly what we're trying for."