Slow animation

Sgt. Rick · Post by **Sgt. Rick** » Tue Jul 18, 2000 4:05 am

Forgive me if this a much asked question, but ever since I upgraded 2.3, the game animations have been pretty slow. I tried to find a message about this, but have failed so far. Please help me!

Thanks for not slamming me...

Rick

P.S. AWESOME GAME!

Tomanbeg · Post by **Tomanbeg** » Tue Jul 18, 2000 4:22 am

I think there is a thread back in june saying to set the speed controls to 'XXX'. Look around June 23rd, maybe. The 'ol RAM ain't what it used to be. And I havn't seen much flaming here. Unless, of course, you want to be flamed. In which case you will find a overabundance of verbose volunteers.
T.

Wild Bill · Post by **Wild Bill** » Tue Jul 18, 2000 7:58 am

Correct, Tomanbeg...On the money!...WB

------------------
In Arduis Fidelis
Wild Bill Wilder
Coordinator, Scenario Design
Matrix Games

Sgt. Rick · Post by **Sgt. Rick** » Wed Jul 19, 2000 5:56 am

Thanks guys...

ASDN · Post by **ASDN** » Fri Jul 21, 2000 4:39 pm

That's not quite right.

I have the K6-2-333MHz & 64MB & S3 VIRGE 2MB and it takes 3-4 seconds to move one OT-34 through 12 hexes in the Last Stand at Seelowe
scenario (map zoomed out to the max).

At the comp.sys.ibm.pc.war-historical the guy named "schalken" reported the same times on K6-3-450MHz & 256MB computer (and probably he has a quite decent GFX card). It seems quite slow.

I slowed down my computer to 200MHz and get 6-8 seconds for this move - which is unacceptable slow. So moving 10 tanks takes about 1 minute on 200MHz system

There were some suspection, the problems comes from the code optimized for Pentium processor, which works slow on AMD machines.

Any comments from the guys with Intel Pentium machines ?

[This message has been edited by ASDN (edited July 21, 2000).]

Arralen · Post by **Arralen** » Fri Jul 21, 2000 5:22 pm

Originally posted by ASDN:
That's not quite right.I have the K6-2-333MHz & 64MB & S3 VIRGE 2MB and it takes 3-4 seconds to move one OT-34 through 12 hexes in the Last Stand at Seelowe scenario (map zoomed out to the max). At the comp.sys.ibm.pc.war-historical the guy named "schalken" reported the same times on K6-3-450MHz & 256MB computer (and probably he has a quite decent GFX card). It seems quite slow.I slowed down my computer to 200MHz and get 6-8 seconds for this move - which is unacceptable slow. So moving 10 tanks takes about 1 minute on 200MHz system There were some suspection, the problems comes from the code optimized for Pentium processor, which works slow on AMD machines. Any comments from the guys with Intel Pentium machines ? [This message has been edited by ASDN (edited July 21, 2000).]

Hohoho.
I'll try on my old machine - K5/166 on Chaintech 5IFM0 with 64 MB.

But I'm pretty shure you just forgot to set the movement delay ot minimum (preferences!)

Arralen

[This message has been edited by Arralen (edited July 21, 2000).]

Nikademus · Post by **Nikademus** » Fri Jul 21, 2000 5:46 pm

When i upgraded to 2.3 i found all the delay settings had been set back to default which slowed the game down considerably.

just go back into the preferences and recustomize

speaking of customize....any word on possibly creating a seperate delay setting for AFV hits (with all that juicy location and pen information!)

ASDN · Post by **ASDN** » Fri Jul 21, 2000 7:32 pm

I posted my message because I had set the movement delay to XXX

And schalken had set it, too.

And, of course, we tried to turn off the graphics acceleration in the DirectX (as the Paul Veber had suggested) and it didn't help.

[This message has been edited by ASDN (edited July 21, 2000).]

schalken · Post by **schalken** » Fri Jul 21, 2000 10:27 pm

Hello everyone,

About this movement speed issue. Voriax
says the speed has always been fast enough
on his K6, but what is fast for you. Paul
Vebber said it takes less than half a second
to move a tank the full allowance on his
P3/667. This is nearly ten times faster than
I've heard any K6 owner confirm in this
little test. So how fast is it Voriax? Less
than one second or more than three?

Video card speed is irrelevant, moving one
bitmap around. 99.9% of time is probably
consumed by various CPU level calculations.
I wonder how slow some more stressing
operations like scrolling would be if mere
unit moves take four seconds due to poor
video card. Swap file size has no effect.
This is Windows 95 OSR2.

And consider - in a 200 unit scenario of
30 turns, with an average delay of just 2
seconds, unit moves take about three and
half *hours* in total.

I'd really like to hear what Michael
Wood has to say.

Voriax · Post by **Voriax** » Fri Jul 21, 2000 11:53 pm

ASDN, I have a K6-2/300@336, 96 megs memory and a TNT2 video card and no difference in speed between 2.2 and 2.3 versions. Also the speed has always been fast enough.

I guess you have noticed how the speed varies depending if enemies see your unit or not? I think they do a bunch of opfire checks each time you move a hex so that causes some jerkiness.

Also as microsoft claims the DX7 is AMD(3dnow) optimized that probably isn't a problem....now what version of M$ofts memory hog operating systems you use? 64 megs of memory with a win98 for example isn't very much...even less with that latest thing they foisted on us.

Or then the video card...with 800x600 resolution that 2MB is getting full..think you could borrow a TNT card or similar from somewhere and try? If that solves the problem such card should not cost very much nowadays.

Voriax

ASDN · Post by **ASDN** » Sat Jul 22, 2000 12:29 am

Originally posted by Voriax:
I doubt the SPWAW code is optimized for either of the major CPU's but just for the Direct X. Michael Wood could tell us more about that.

The code can't be optimized for DirectX (I can't figured out what it means). But in the case you are right and it is possible the code can be optimized for DX, that the SPW@W is likely not optimized for Direct X because it is conversion of DOS game (Paul Veber confirmed it indirectly on war-historical newsgroup by pointing out that one can't compare native Win32&DirectX game like TOAW with the converted one like SPW@W).
But for sure the code can be optimized for the particular CPU and SPW@W looks like optimized for the Pentium machines or similar ones, because there is a non-logical order of instructions in the code (to help keeping busy both pipelines in Pentium CPU). It can be also optimized for PII machines - it is not clear from the code to me (and I have no time to dig in it more).

As for swap file, have you thought about setting a fixed size swap file? About 2,5 times the memory amount. My WIN95 started behaving much nicer after I set that swap file.

Maybe my answer wasn't clear enough. When I am playing SPW@W with the mentioned scenario I have about 14-16MB free RAM with ICQ and WinPopup running in the background. My Windows 95 SR2 doesn't even try to use the swap file yet so its setting is irrelevant.

Maybe it seems like a flaming to Voriax but I swear it is not my intention. Let's only make things clear.

But today I tested SPW@W on the Intel PII 350MHz machine with 64MB RAM and 8MB AGP Riva 128ZX GFX card, under Win98. I got the same results 3-4 seconds for the move (in the same conditions as above).
So it is not the K6 problem.
How much faster can be the P3/667 ? Six to eight times ? Especially comparing to the K6-450 with Matrox Millenium II card (schalken).

And I agree with schalken that the video card speed is irrelevant. But maybe not in 99.9%.

I want to ask MG if there exist a chance to reduce the movement time ? Maybe even by factor of 2 or 4 ?
It helps also with PBEM replay time

schalken · Post by **schalken** » Sat Jul 22, 2000 3:02 am

ASDN,

Interesting... But Pentium II doesn't support
SSE instructions. K6-2 & K6-3 do support
3DNow. DirectX supports both SSE & 3DNow.
So if the code is optimized for Intel, but
DirectX 'favors' K6 in this case, P2 & K6
are equally fast. (Where is Michael Wood?)

P3/667 is up to 50% faster than K6-3/450
as far as integer is concerned. Significantly
more in floating.

Voriax · Post by **Voriax** » Sat Jul 22, 2000 3:08 am

ASDN, you know where the shortie 'IBM' comes from?

No?

Well, it means 'It's Black Magic'

the bugs and strange behaviour of computers are numerous. i'm one of the persons/reasons why you don't have sound with fast arty, as I got frequent crashes with sound on during bombardment. Latest drivers, DX, everything.

I did time a full movement of OT-34 at the Seelowe scenario, it was slightly under 2 seconds with the movement rate a default. It is fast enough for me, should it be faster I'd want to slow it down. No jerkiness whatsoever, just smooth movement.

Btw ASDN, no offense taken. And I seem to run the same Win version as you.

This is really baffling. I assume you have run a pile of benchmarks etc to verify your computer works normally and performs at expected level with other programs?

i might throw in a one more guess. As you may have noticed I have a slightly overclocked CPU, namely the system bus and memory run at 112 MHZ. If SPWAW likes fast memory this might be a partial cause. some programs like seti@home are really happy if they get fast memory and get a 20% performance gain from it.

Voriax

ps. IBM may also mean 'Idiots Built Me'

Kluckenbill · Post by **Kluckenbill** » Sat Jul 22, 2000 3:36 am

On a slightly different issue . . . . is there some way to speed up the replay of my opponent's turn in a PBEM game? My video is just fine but the replay takes forever. I have fast artillery on and all options to XXX but its still interminably slow.

schalken · Post by **schalken** » Sat Jul 22, 2000 12:14 pm

Voriax, does your system beat these?

Moving the OT-34 at 77,13 to 90,12 (13 hexes)

Movement Delay set to XXX

Map zoomed in - less than 1.5 seconds
Map zoomed out - less than 3.5 seconds

Movement Delay set to 1

Map zoomed in - 4 seconds
Map zoomed out - 5.5 seconds

Movement Delay set to 2

Map zoomed in - 6 seconds
Map zoomed out - 8 seconds

According to benchmarks (like WinTune),
2D video, memory and CPU performance is
just as fast as K6-3 is at 450. And Quake1
scores above 30 fps at 640x400.

schalken · Post by **schalken** » Sat Jul 22, 2000 12:31 pm

Voriax, does your system beat these?

Moving the OT-34 at 77,13 to 90,12 (13 hexes)

Movement Delay set to XXX

Map zoomed in - less than 1.5 seconds
Map zoomed out - less than 3.5 seconds

Movement Delay set to 1

Map zoomed in - 4 seconds
Map zoomed out - 5.5 seconds

Movement Delay set to 2

Map zoomed in - 6 seconds
Map zoomed out - 8 seconds

According to benchmarks (like WinTune),
2D video, memory and CPU performance is
just as fast as K6-3 is at 450. And Quake1
scores above 30 fps at 640x400.

ASDN · Post by **ASDN** » Sat Jul 22, 2000 12:55 pm

Voriax,

Short answer:
1. I moved unit in the direction opposite to the enemies and the front line is full of smoke (after a barrage) so probably no even a single enemy saw my movement. Yes - I use "jerkiness" to detect enemies, I hasn't spot (just look witch hexes become visible after move - it is so hard not to take it into account).

2. DX7 seems optimized for K6-2. I got some speed boost after installing it. But we are talking about optimization of the SPW@W code.

3. I have 14MB free memory with SPW@W with the scenario & ICQ & smaller utils and no swap file yet.

4. SPW@W apparently uses 800x600x256 color mode (I know how to set the refresh rate for SPW@W

). This mode uses only 450KB of the video memory. I don't know if SPW@W uses the rest of it. Maybe Mike can give us some tips (I know how to use the off-screen memory in the DDraw programming) ? But it is still pretty much (1.5MB free) for icons.

5. I have no AGP slot. This is PCI card, but Schalten seems to have more decent AGP card (look at his configuration). But the S3 Virge is *VERY* fast 2D card.

kao16 · Post by **kao16** » Sat Jul 22, 2000 1:09 pm

Originally posted by ASDN:
Kao16,

Could you measure the time of the movement described above? You don't need to be *very* precise Just if it is 1 second, 2 seconds or 3-4 seconds or so.
If you have the faster movement than I have probably the delay calculation is invalid on K6 machines

Tried Last stand at Seelowe.
Live delay - 160
Scroll delay - 1
movement delay - 1
message delay - 150
(took some 58 minutes to complete the initial bombardment- next time I will reduce all delay to 1 before I start)

Movement of OT-34 moving maximum number of hexs (on road, requiring turn to retreat) about 8 sec.

JSIII max movement forward - 4 sec (couldn't cross the river, a lot of changes of direction to go around troops and tanks)

OT-34 moving maximum allowance NNE (up map) 4-5 sec.

So, it seems to be as slow as you are getting but it is a running on a P120 (pre-MMX), 80 MB (70 ns) RAM, and Grafixstar 700 with 4 MB.

Voriax · Post by **Voriax** » Sat Jul 22, 2000 1:39 pm

Schalken, here are my results for the same tank and same destination:

Delay XXX
IN: less than 2 secs
OUT: less than 3,5 secs

Delay 1
IN: Bit over 4 secs
OUT: Less than 6 secs

Delay 2
IN: Less than 7 secs
Out: About 8,5 secs.

As for Quake benchmarks, I don't even have 1 installed but at Q2 I got something around 45fps at 1024x768

So the movement speed in my system is roughly same as in yours. I normally play at XXX speed setting and map zoom is at one of the 2 middle steps, depending of the situation.
For me the speed is okay, I don't like tanks zipping around

Voriax

kao16 · Post by **kao16** » Sat Jul 22, 2000 2:10 pm

I'm running v2.3 on a pre MMX pentium 120 with 80 MB RAM and a 4 MB video card.

It's noticeably slower than DOS based SP games, but I'm not experiencing the delays people are reporting (have speed settings on 1).