[Danukem]

The Commander, on 16 January 2012 - 01:50 PM, said:

In fact, shouldn't the above link really direct to http://dukeworld.duk...ke32_latest.zip and not http://dukeworld.duk...e32_current.zip

Every once in a while, a really nasty bug gets committed, so there is always that danger. On the other hand, if current.zip is not going to be updated on a regular basis, then it's probably better to link to the latest snapshot.

[Helixhorned]

Plagman, on 16 January 2012 - 12:26 PM, said:

I always had Mouse4 bound to quick kick and double-Mouse4 to steroids. Not sure why it wasn't working for you; was that with sdlayer or winlayer?

Ah yes... I misread the winlayer source, where mouse4 was already passed along; the change affects only the SDL one then.

[Jblade]

Just saw the new snapshot with the doubled max tile amount! helix you are a living legend, thanks so much for this!

[Helixhorned]

Hehe, you're welcome!

While I'm pretty sure that the increasing doesn't break anything, I did the testing only by taking an AMC TC art file and offsetting the tiles into the new range. I didn't test voxels or models, so give me a call if there are problems.

Also, I see that you're using dynamictileremap in AMC TC. Is that because you were running out of tiles or does it have a higher meaning? I'm asking because I never got the fuss around that feature and IMO it's only causing problems.

... Further on the topic of weird features, would anyone shed a tear if adult lock was removed? That's also dozens of lines of dead code, if you ask me. (Not implying that I would remove dynamictileremap without hearing the arguments.)

[Hendricks266]

Dynamicremap is very useful if you want to organize your art in ways other than the seemingly arbitrary tile numbers scattered throughout the range that have hardcoded uses. For example, the title screen is 2493 while the loading screen is 3281. Don't remove it.

There is a case to be made for the parental lock, like if you want to expose someone to the gameplay without the gore or sexual content attached. We should not remove features because they are "causing bloat" (which is a dubious reason at best for a project of this small size and the small footprint of the code in question) or similar reasons.

-------------------------------------------

Has anyone had success building EDuke32 with LTO=1 on MinGW32-GCC? I get these errors when linking the executable:

http://pastebin.com/7WbUqcNN

A quick grep of the source reveals that the problem symbols are used in inline asm code:

http://pastebin.com/90jjQC4D

[Jblade]

I'm not even sure dynamic remap even worked for me, but maybe other people are getting use out of it

As for parental lock; it's one of those things that seems redundant but then maybe some person somewhere is still using it so it's hard to say.

[Danukem]

James, on 19 January 2012 - 11:22 PM, said:

As for parental lock; it's one of those things that seems redundant but then maybe some person somewhere is still using it so it's hard to say.

The reason to keep parental lock is to to emulate as much as possible about the original game, not because people actually use it.

[LeoD]

Could someone please re-enable crosshair-coloring? [eduke32.cfg : CrosshairColor = "255,174,0"]
Btw, I guess the color of a rocket's smoke trail should turn white again in some distance.

Attached thumbnail(s)

• 
• 

[Plagman]

Probably another of the many problems caused by the sprite caching thing... it doesn't look like it's re-culling the lights as the sprite moves, I'll check it out in more detail.

[Micky C]

When Duke is underwater, instead of his spotlight shadow looking like Duke, it appears to be that of the enforcer shooting his chaingun when stationary, and an enforcer running when moving. Looks like the tiles are mixed up when the sector lotag is 2.

[Daniel]

I have send to helixhorned another email regarding OpenGL Shading (this time its issue about Shading Scaling with distance and shade parameters):


==================================

[Kyanos]

I have a question on what would be the most efficient way to make some repeating models.
example.
Say I have a small md3. It currently has 4 different versions, using 4 different tiles and 4 different skins, yet 1 model.
Would it be better (FPS/loading time) to make 1 skin, with no more than 4Xs the area up to a max of 1024^2 (where each old skin is moved to a corner or something) then make 4 md3s (once again small like 1-20 kb) each having different UV texture coords?
ADDED:

A pic is worth a million of my words
Is 4 SMALL md3s sharing 1 skin more efficient than 1 md3 using 4 skins?

[Plagman]

Is there a reason you can't just use one model on one tile that has 4 skins defined and cycle through them with palettes? Sorry if I missed something in your post that invalidates that.

[Micky C]

I believe he's asking which method out of

Quote

4 SMALL md3s sharing 1 skin

and

Quote

1 md3 using 4 skins?

is less computationally intensive, both in regards to framerate and loading time.

[Danukem]

It's impossible to answer Drek's question without more information.

As for Plagman's question: I think the reason he would want the skins to be on different tiles, rather than on the same tile with different palettes, is that with different tiles you can make images for the mapster tile selector showing the different skins. That makes it easier on the mapper, but you can't do that if they are all on one tile (plus the mapper would have to remember which pals the skins are on).

[Mia Max]

Is there an extra thread that explains more about polymer?
I guess many questeions have alraedy been asked about it in this thread, but there are too many posts in here to find answers without searching an awful long time, taking even more time when you don't understand english very well like me.
For example, what does limitate the fps?
I just looked on GPU and CPU-usage during playing and realized that GPU-usage is at 20% and CPU-usage at 50%, but in some scenes I got under 30 fps.
It would be much easier to bulid maps if I knew more about it to avoid low framerates.
I know that these questeions have been asked before and I was searching for answers, but didn't find good ones.
I also have visited the eduke32 wiki page to learn more about mapster32 and some of its new features like TROR, but I found nothing about that.
I know that Plagman and TX have better things to do than answering stupid questions, so an extra thread with more informations about Polymer would be awesum.

[Danukem]

Mia Max, on 26 January 2012 - 04:30 AM, said:

Is there an extra thread that explains more about polymer?
I guess many questeions have alraedy been asked about it in this thread, but there are too many posts in here to find answers without searching an awful long time, taking even more time when you don't understand english very well like me.
For example, what does limitate the fps?
I just looked on GPU and CPU-usage during playing and realized that GPU-usage is at 20% and CPU-usage at 50%, but in some scenes I got under 30 fps.
It would be much easier to bulid maps if I knew more about it to avoid low framerates.

For now, the only way you can avoid low fps is by keeping your map fairly simple in construction, and also making sure that there are only a few lights (especially spotlights) which are shining on rendered surfaces at any given point in the map. It's fine to use detailed textured and models, though. I asked about Polymer optimizations a while ago in a different thread, and here is how Plagman replied. It's the most detailed information that we have about it to date, and I believe it all still applies. I underlined the part which seems most relevant to explaining the low fps.

http://forums.duke4....dpost__p__88492

Plagman, on 31 May 2011 - 09:39 PM, said:

The problem isn't that we need to brainstorm for optimization ideas, it's that I have to break everything and put it together in a different fashion. It's been known from the start; I've done it like that at first because it's always easier to get something that works right than something that works right and well. Too bad that only applies to the short term.

There are several different problems; one aspect a map is made of a lot of planes that are all drawn separately from front to back; materials are switched between each plane, and each time a red wall is crossed a "portal" is drawn to determine if the sector that follows needs to be queued for rendering or not. This constant material switching and drawing only one plane at a time isn't really efficient. The former causes a lot of state validation and thrashing, while the latter doesn't utilize the GPU in an efficient fashion. It's constantly submitting tiny pieces of work and reaping it, causing a lot of idle time waiting for the results. This also causes big maps with lots of red walls to be insanely complex to draw.

Another problem is that the multi-pass drawing approach uses a lot of fillrate as soon as a fairly big plane gets a few too many lights on it.

Things that needs to be done:
- Move the diffuse modulation from the material to the vertex attribute. That would cause the amount of different materials in a map to be all the different tiles used, instead of the more complex tile X shade.
- Move plane vertex attributes from several unique vertex buffers to a single storage or several depending on the staticness of the data.
- With a more reasonable material count and stuff sharing buffers, we can start batching planes together. To leverage that, throw planes into material buckets instead of rendering them right away. When we're done building index buffers for the buckets, submit them in one big go (one per material).
- Change the HSR walking to do a first pass before doing the actual shaded drawing. This means that in the event that occlusion queries have to be reaped, we can get the result a lot faster since we're pushing a lot less pixels using a passthrough fragment program.
- For rendering the shadow maps, we only need materials for sprite silhouettes. The rest can be batched in a single draw call after building the right index buffer, same as above. This will be a lot faster.

(Even) bigger projects:
- Defer the lighting model to shade from a fat buffer instead of inline. This eliminates two very expensive steps, light culling and multi-pass drawing.
- Instance models together.

[Mia Max]

Thank you, DeeperThought.
Now I understand.
I already avoid lights.
But I didn't know that many sectors/red walls are a big problem.

So it's not that important which hardware you have as it is limited by the idel times.
In my new map I have still 30 fps in the biggest area, so other people with less good hardware should have nearly the same framerate as my CPU and GPU are at 50% and 20% ?
I just want to make sure that the map is playable for most people.
I played my old map Desertbase with Polymer and it isn't enjoyable due to 15 fps and less.

[Helixhorned]

LeoD, on 22 January 2012 - 04:29 PM, said:

Could someone please re-enable crosshair-coloring? [eduke32.cfg : CrosshairColor = "255,174,0"]

Seems like a configuration reader/writer issue to me...

Daniel, on 24 January 2012 - 07:43 AM, said:

I have send to helixhorned another email regarding OpenGL Shading (this time its issue about Shading Scaling with distance and shade parameters):

I don't think that shades with large negative values are used very often, and if they are, it's to make something visible all the way...
So, such perfectionism is of little practical relevance, IMO.

[Kyanos]

Plagman, on 25 January 2012 - 05:52 PM, said:

Is there a reason you can't just use one model on one tile that has 4 skins defined and cycle through them with palettes? Sorry if I missed something in your post that invalidates that.

It's like DT said, it's easier to map with when each different version is a new tile #, also some md3s need to be assigned to unique tiles for sprite like special effects to work.

Quote

Myself,
Is 4 SMALL md3s sharing 1 skin more efficient than 1 md3 using 4 skins?

I think I now know the current answer to my ?

Every tile being defined loads it own skin, no matter if it has been loaded already or not.
Therefore, it is more efficient (for now) to use 4 different small skins than 1 large skin.

Am I right? I've made assumptions based on DTs reply to Mia Maxs question about FPS and polymer.

Would using Alt Pals be any faster? Each alt pal is loaded at start up also, no?

[Hendricks266]

Hendricks266, on 19 January 2012 - 04:14 PM, said:

Has anyone had success building EDuke32 with LTO=1 on MinGW32-GCC? I get these errors when linking the executable:

http://pastebin.com/7WbUqcNN

A quick grep of the source reveals that the problem symbols are used in inline asm code:

http://pastebin.com/90jjQC4D

Even after r2271 I still encounter this issue.

Quote

r2271 | helixhorned | 2012-01-20 05:23:53 -0800 (Fri, 20 Jan 2012) | 4 lines

Use ATTRIBUTE((used)) for symbols used only in inline asm.

This may fix compilation for optimized builds with GCC or clang where
the compiler would otherwise decide that those symbols are unused.

[LeoD]

LeoD, on 22 January 2012 - 04:29 PM, said:

Could someone please re-enable crosshair-coloring? [eduke32.cfg : CrosshairColor = "255,174,0"]

Helixhorned, on 26 January 2012 - 09:46 AM, said:

Seems like a configuration reader/writer issue to me...

Outcome of source code rape between r1624 and r1627.

[TerminX]

It looks like nobody ever set anything up to add the crosshair color information to the new style cfg... should be pretty easy to fix.

[Helixhorned]

Attention, DON'T use r2299 to save games. I had disabled the formerly non-functional dynamicremap feature (TX re-enabled the fixed one in r2300), but that wasn't entirely without side-effects. Sorry!

edit: the real breakage comes when trying to load an r2299 save with a later build.

[LeoD]

Today I somewhat restored my broken compile environment. But building the debug version fails for the backtrace DLL. Any ideas?

Attached File(s)

•  make RELEASE=0-4.5.2.log (292bytes)
  Number of downloads: 628
•  make RELEASE=0-4.6.2.log (377bytes)
  Number of downloads: 637

[Hendricks266]

I get similar errors, but a whole lot more of them. The solution I found was to add "-lintl" to the makefile entry for ebacktrace1.dll. Only add this if you get errors.

Index: polymer/eduke32/Makefile
===================================================================
--- polymer/eduke32/Makefile	(revision 2290)
+++ polymer/eduke32/Makefile	(working copy)
@@ -336,7 +336,7 @@
 
 # RULES
 $(EBACKTRACEDLL): Windows/src/backtrace.c
-	if $(CC) -O2 -shared -Wall -Wextra -o $@ $^ -lbfd -liberty -limagehlp; then $(COMPILE_OK); else $(COMPILE_FAILED); fi
+	if $(CC) -O2 -shared -Wall -Wextra -o $@ $^ -lbfd -lintl -liberty -limagehlp; then $(COMPILE_OK); else $(COMPILE_FAILED); fi
 
 $(OBJ)/%.$o: $(SRC)/%.nasm
 	$(COMPILE_STATUS)

Both Helix and synthesis don't have this problem, and we do not want to make this change upstream because we don't want any dependencies for the dll. I am at a loss.

[TerminX]

Hendricks266, on 29 January 2012 - 11:14 PM, said:

I get similar errors, but a whole lot more of them. The solution I found was to add "-lintl" to the makefile entry for ebacktrace1.dll.

Index: polymer/eduke32/Makefile
===================================================================
--- polymer/eduke32/Makefile	(revision 2290)
+++ polymer/eduke32/Makefile	(working copy)
@@ -336,7 +336,7 @@
 
 # RULES
 $(EBACKTRACEDLL): Windows/src/backtrace.c
-	if $(CC) -O2 -shared -Wall -Wextra -o $@ $^ -lbfd -liberty -limagehlp; then $(COMPILE_OK); else $(COMPILE_FAILED); fi
+	if $(CC) -O2 -shared -Wall -Wextra -o $@ $^ -lbfd -lintl -liberty -limagehlp; then $(COMPILE_OK); else $(COMPILE_FAILED); fi
 
 $(OBJ)/%.$o: $(SRC)/%.nasm
 	$(COMPILE_STATUS)

Both Helix and synthesis don't have this problem, and we do not want to make this change upstream because we don't want any dependencies for the dll. I am at a loss.

Same problem here, actually.

[LeoD]

Hendricks266, on 29 January 2012 - 11:14 PM, said:

+ if $(CC) -O2 -shared -Wall -Wextra -o $@ $^ -lbfd -lintl -liberty -limagehlp; then $(COMPILE_OK); else $(COMPILE_FAILED); fi

Works. Thank you.

LeoD, on 22 January 2012 - 04:29 PM, said:

Could someone please re-enable crosshair-coloring? [eduke32.cfg : CrosshairColor = "255,174,0"]

Helixhorned, on 26 January 2012 - 09:46 AM, said:

Seems like a configuration reader/writer issue to me...

LeoD, on 28 January 2012 - 05:59 PM, said:

Outcome of source code rape between r1624 and r1627.

TX, on 28 January 2012 - 07:45 PM, said:

It looks like nobody ever set anything up to add the crosshair color information to the new style cfg... should be pretty easy to fix.

Yup. I simply reactivated the code block from line 716-741 in config.c (r2290). No idea about possible side effects, though.

[LeoD]

Proposal:
What about adding an echo or version command/directive to the CON and DEF languages?
If these files could put out a version number it would be easy to identify outdated Mods or HRP versions when someone posts his eduke32.log.

[Hendricks266]

Echo would be simple enough. Maybe I'll work on it.