AUG HBAR rate of fire

[KingVaroon]

Den,

Im using a stop watch to record how fast the AUG fires off one full clip (42 rounds). I have tried several times now and each time it seems to take 4 seconds to fire off a full clip.

Doing the math (42/4)*60= 630 RPM...

The charts you have put up on the home page says 650 RPM. Would you mind double checking the fire time yourself and making necessary corrections please?

[Den]

I did the AUG and RPD extensively due to the 42 round mag and the in-game attributes reading identical rates of fire.

Both of them have the same FireTime.

In fact, they may be even faster as both a 0.091 and 0.090 delay macro got successful bursts of fire.

The magazine empty time is only accurate up to 0.01 of a second. The thousandth place is difficult to measure with that means.

[KingVaroon]

You are correct, i just ran several tests with the RPD and it is also coming out to 630 RPM. But if you are aware of this then why did you put them both as 650 RPM?

[Den]

Because that's what they are.

666 RPM would be more accurate than 650.

[fred]

I also used the stopwatch method for measuring ROF using my PS3 in multi-player private match.

My figures for the all the Machine Guns including AUG are:

MG	Mag Size	TimeToEmpty	Calculated RPM	Den's RPM for PC
L86	100	8.4	714	750
RPD	100	9.8	612	650
MG4	100	8.3	723	800
AUG	42	4.1	615	650
M240	100	6.7	896	850

I got Den's figures used above here .... denkirson.xanga.com/715966769/modern-warfare-2/

As you can see there are some substantial differences. I don't know whether the PS3 has different fire times than PC, whether there is some kind of display issue, etc. But I'd say the time measurements for the 100 shot MG's in particular should be pretty accurate because even if my reaction time is a little off it doesn't make much difference in 7-10 seconds it takes to unload the mag in full auto. I also took multiple measurements and averaged them.

Any ideas?

[n1gh7]

fred

because both of the console systems are locked at 60fps they cannot accurately show guns firing above a certain RPM. you need to be above 60 fps or so. They cut out bullets or time.

Something like that.

[KingVaroon]

@night: so does that mean the guns fire slower on the consoles or it just appears slower than its firing?

[mannon]

It appears to actually slow the time to empty the magazine and thus actually slow the rate of fire according to tests, stupid as that is. I still contend that the game should be able to generate all the necessary hitscans to fire a full auto weapon without any slowdown at all even if you got dropped to 10 frames per second, but the game doesn't appear to be programmed that way.

My only theory is that the game actually waits for a frame to be drawn for each hitscan to borrow data from the rendering to find out where that pixel at the middle of your screen is pointing, thus preventing the game from producing hitscans between frames and costing you increasing amounts of time the fewer frames per second and the more hitscans per second.

But even if that's true it could delay the actual hitscan but not delay the timer that ticks off how long to wait before firing off your next shot, so I don't quite get it.

I suppose the game's actual timers could be getting choked by the poor system performance, but I don't understand why they wouldn't scale to the current FPS. On PC I almost always set my games to run with a variable and uncapped framerate and they run fine. But then... this is a highly modded version of the IdTech (Quake 3) engine... There's no telling what sorts of quirks are hanging around in there. Quake engines have traditionally had some very strange framerate related oddities.

[n1gh7]

KingVaroon

Both are correct.

The gun fires slow because it appears it is firing slower.

I'm just going to expand on what mannon said.

The game can only update when a new frame is drawn. This means that you will constantly be bottlenecked at the Graphics card. For example, when you jump the time that you spend in the air is preset and will continue counting down but you only go as high as the frame that was drawn at the highest point in your jump.

This also explains why you cannot update to the "server" faster than your FPS. This, and the fact that the quake 3 engine has certain "sweet spots" (like 125 fps) are the basics of tweaking. (which i don't think you an do anymore because of IW being stupid.)

That's how I see it. Something liek that.

[ecomni]

That's quite interesting. So this effect isn't just visual? I ask mainly because of the drastic difference between the M240 and MG4 "perceived" RPMs.

[fred]

Dec 30, 2009 3:30:56 GMT -5 KingVaroon said:

@night: so does that mean the guns fire slower on the consoles or it just appears slower than its firing?

A while back I calculated ROF for almost every automatic weapon (but not including single shot or triple burst).

Here is the methodology I used to calculate:
* Used PS3 in multi-player private
* Hand Timer
* Unsighted (e.g. Not pressing L2)
* No attachments on weapon
* Magazine Full
* Simultaneously Press start/stop on timer and pressing and holding R2 (fire button).
* Press start/stop on stopwatch again to stop time when magazine is empty. I am watching ||||||||||||| indicator to empty as well as sound and muzzle flash.
* Formula for ROF = (mag size * 60) / Time to Empty Full Mag
* Make at least 3 good measurements and average
* Discarded measurements that I think I spazzed out on.

As you can see below my figures are consistently having ROF 3-6% lower than Den's. The places where they agree less closely then the 3-6% lower appear to be: very high rate of fire weapons and the machine guns, especially the MG4 and M240. For the very high rate of fire weapons that are lower I am not too surprised given all your comments about frame rate caps. However, It is surprising that some ROF for the high rate of fire weapons are measured higher then Den's and that the MGs are off so much. Especially as I am most confident about the measurements for the the 100 round MG's. This is because I would expect measurement errors (systemic or otherwise) to be proportionally smaller. For example, my measured TMP and AA-12 times are 1.1 and 1.2 seconds to empty. If I am systemically off +/- .1 second the % error might be expected to be close to 10% ... or like +/- 80rpm for the TMP. That's a lot. However, for the RPD with the same measurement error of +/- .1 second and 9.8 seconds to empty would be a % error only around 1% or +/- 6-7 rpms.

Weapon	Mag Size	Time To Empty	Calc ROF	Den's ROF
M4	30	2.3	783	800
SCAR	20	2	600	630
TAR	30	2.5	720	750
ACR	30	2.5	720	750
FN2000	30	2.1	857	900
AK	30	2.7	667	700
MP5	30	2.25	800	850
UMP	32	3.2	600	630
KRISS	30	2	900	1000
Uzi	32	2.1	914	888
P90	50	3.33	901	850
L86	100	8.4	714	750
RPD	100	9.8	612	666
MG4	100	8.3	723	800
AUG	42	4.1	615	666
M240	100	6.7	896	850
PP2000	20	1.65	727	700
G18	33	2.2	900	1100
TMP	15	1.1	818	900
AA-12	8	1.2	400	400

Den seems to have used a much more sophisticated method to measure minimum fire time and hence ROF; but on a different platform than my PS3. I really would expect the results from each method to agree on a given platform.

If the figures agree on PC then PS3 would appear to have different ROF for certain weapons.

If the figures don't agree on PC then maybe there is a problem with the methodology outlined above.

So I'd be interested in seeing if the manual timing method performed on PC agrees with Den's fire rates on a PC.

Could anyone check?

[chyros]

Argh, AGAIN someone is not taking into account that you need to calculate for n-1 bullets. In addition, you're not using slowing macros like Den is which may give rise to some human error.

[n1gh7]

I think to dial in the fire time Den uses a delay macro. I think what it does is it "pulls the trigger" with the delay of the fire time. He makes it a 3 bullet burst. If all 3 bullets fire then the fire time is less than or equal to that.

Something liek that.

I thought that this was all explained... you guys need to stop questioning Den on the same thing over and over again.

Also the n-1 thing. and read what mannon and I said farther up. it explains why you cannot measure the FireTime on console.

[mannon]

I'd put more trust in Den's methodology. But it may be a good idea to check the consoles for consistency since they tend to have a few odd weapons with different stats. Still, it's usually damage that is different rather than ROF I believe.

Given that the processing seems to me a major factor I would suggest finding a very simple area to do your testing, put yourself in a corner somewhere so that it's repeatable and aim down at your feet in exactly the same spot for every weapon.

I'd also suggest using extended mags to increase the time to empty and thus minimize human error. And I'd average more than just 3 times. Of course, that's a lot more work. ;p

I also wonder how accurate the sound is. At a guess it's probably as accurate if not more accurate than the video. (The sound isn't limited to the frame rate.) But lots of games wind up with timing issues with the sound. Hell a lot of them actually use a burst of several shots fired for the full auto weapons rather than trying to actually kick off a sound effect for each round fired, which often results in weapons having vastly different sounds vs their actual ROF. I don't know how well the sound meshes with the shots fired in MW2, but it could be either less or more accurate than the video, so it may be useful to try turning it off and seeing how the weapons test without it. I suspect it won't make much difference anyway, because what you'll actually be doing is watching the ammo indicator empty out and anticipating the moment of empty. The only way to measure by sound would be to wait until you don't get the next expected shot which would incur a delay at least equal to the firetime before you would even have the stimuli (lack of shot fired) and on top of that would add your own reaction time as another delay.

Recorded audio might be more telling, though. But of course, the usefulness of that depends upon how reliably IW coded the audio, and my experience has been that game developers usually do audio by feel rather than precision. As long as a gun sounds good whether you heard 29 bullets or 31 instead of 30 they could usually care less. ;p

[Den]

The sky. Aim at the sky.
If you're shooting your feet, that's a lot of impact particle effects.

[n1gh7]

mannon

I think that the audio is generated the same way that hit scan are, saying that there is the graphics card bottle neck. When I record the audio has about a 2 frame delay from when I shoot, but that could be intensional.

About sooting at the sky, do it on a map where there isn't a bunch of dust clouds. Those are a pain. And no rain. I hate that stuff.

[ssog]

Dec 31, 2009 18:47:53 GMT -5 n1gh7 said:

And no rain. I hate that stuff.

All I can say is that my life is pretty plain. I like watching the puddles gather rain.

[fred]

Thank you for your comments! You have increased my understanding about the complexities of automatic fire in the MW2 game engine.

My key insight from this thread is that in practice a variety of effects interfere with the theoretical weapon ROF making it a slippery thing in practice, both to measure and in actual gameplay
This is unlike finding weapon damages, for which you can derive concrete figures for and make corresponding accurate predictions. I recognize Den used a sophisticated and controlled method on PC to measure Fire Times and I suspect that he has gotten very close to the theoretical figures encrypted in the fast files for PC. However, from your comments those theoretical ROF figures are effected by the situation and don't necessarily agree with the reality of of the IW MW2 game engine once you add in multiplayer, explosions, gunfire, movement, etc. In addition, the PS3 console could have different stats, but that will be difficult to tell.

Den: The sky. Aim at the sky. If you're shooting your feet, that's a lot of impact particle effects.
N7gh7: About sooting at the sky, do it on a map where there isn't a bunch of dust clouds. Those are a pain. And no rain. I hate that stuff.

Good suggestions. I tended to be shooting at walls at short/medium range when I did the measurements. I re-measured the 100 round MG's shooting into clear skies with no rain or blowing sand (Skidrow looking straight up). My measurements didn't change more the .1 seconds (not significant result) But I also tried the measurement in a more complex situation, shooting up a lot of windows and hitting moveable stuff at close range. (Greenhouse on Estate). Again no significant difference.

Chyros: Argh, AGAIN someone is not taking into account that you need to calculate for n-1 bullets.

Ah, the fence post problem. This is a good point, but:
(1) n-1 moves the error the wrong way. Most of my measurements are lower than Den's. For example the RPD measurement goes from 613 to 606 if I use 99 rather than 100. For smaller mags this is more significant. The SCAR measurement goes from 630 to 570 when I use 19 rather than 20.
(2) My biggest cue to stop the timer has been watching the "|||||||||" bullet indicators gray out. I have not really been sure where in cycle of a FIRETIME that the last white bar grays and or the video and audio stops. It is uncertain. If it is at the end then n is more correct than n-1.

Chyros: In addition, you're not using slowing macros like Den is which may give rise to some human error.

I recognize there is human error using my simple method. I thought I emphasized this in my previous posts. However, this error should be minimized when I shoot slow firing, large magazine weapons, so the 100 shot MG's should match Den's figures closer, right? They don't. This implies the above effects are in play or the PS3 has a different ROF for various weapons. It is very hard to tell which.

Further questions:
1) Do you think the effects that interfere with the theoretical weapon ROF affect the higher ROF weapons to a greater degree?
2) Do you think the effect always is to REDUCE the ROF? Or do you think it is possible these effects can sometimes INCREASE the ROF?
3) Do you think this effects all players on a console multiplayer online game equally, assuming equal weapons? Or do the ones with more "action" hindered?

[mannon]

1. Yes Higher ROF weapons suffer more at low framerates.
2. Yes Having high framerates does not push the ROF beyond it's coded rate, low framerates merely reduce it.
3. It seems to be dependent upon your local performance. In other words if you're getting a low framerate then it's your gun that is slowed down. Enemies won't be affected by your framerate, but by their own.

As for the fence post thing... It doesn't matter which way it moves the numbers. n is wrong and n-1 is right. You can test this with the Intervention with it's extremely long firetime. The ammo is removed from the ammo counter as soon as the shot is fired and does not wait for the firetime after the shot to remove it, which is kinda nonsensical. The firetime only happens n-1 times because the first bullet is fired immediately and the ammo counter hits zero as soon as the last round in the magazine is fired. The game does not make you wait the firetime after the last shot fired before you can reload.

PS. Thank you Den! lol I hadn't even thought of the sky!

PPS. Since the consoles already have their framerates capped they may actually be running at a pretty consistent rate, thus you wouldn't get as much of a fluctuation in actual ROF. It may take quite a bit to really drop the framerate appreciably and it may not be possible to really do much damage to the framerate alone. It may take a lot of people all shooting at once to really drag the framerate down. It should be noted, however, that even if consistent the framerates will already be lower than possible on PC due to the cap and thus ROF will be affected by that.

[chyros]

Jan 1, 2010 5:06:18 GMT -5 fred said:

Chyros: Argh, AGAIN someone is not taking into account that you need to calculate for n-1 bullets.

Ah, the fence post problem. This is a good point, but:
(1) n-1 moves the error the wrong way. Most of my measurements are lower than Den's. For example the RPD measurement goes from 613 to 606 if I use 99 rather than 100. For smaller mags this is more significant. The SCAR measurement goes from 630 to 570 when I use 19 rather than 20.
(2) My biggest cue to stop the timer has been watching the "|||||||||" bullet indicators gray out. I have not really been sure where in cycle of a FIRETIME that the last white bar grays and or the video and audio stops. It is uncertain. If it is at the end then n is more correct than n-1.

Chyros: In addition, you're not using slowing macros like Den is which may give rise to some human error.

I recognize there is human error using my simple method. I thought I emphasized this in my previous posts. However, this error should be minimized when I shoot slow firing, large magazine weapons, so the 100 shot MG's should match Den's figures closer, right? They don't. This implies the above effects are in play or the PS3 has a different ROF for various weapons. It is very hard to tell which.

Fence post problem, I like that name ^^ .

1)This is true, you are correct in this. However you still need to do it xD . If the error gets bigger because of this, it just means that the error was bigger to begin with.
2)You can test this easily. If you press the fire button and a | disappears immediately, then it's at the start.
+
Hmmm, you have a point there.

[qupie]

Dec 31, 2009 18:33:47 GMT -5 Den said:

The sky. Aim at the sky.
If you're shooting your feet, that's a lot of impact particle effects.

so does that mean that a silencer might increase the ROF? and FMJ may slow it down?

[Den]

Silencer won't make much of a difference, but those FMJ sparks definitely chew up my frames when up close.

[toysrme]

in a private MP match the ps3 and 360 will both run constant 60fps.

any FPS is accurate to display the ROF of these weapons.
fact of the matter is that the game engine itself should be running full speed on any system, ther eis more than enough CPU power to do it. the only practical difference FPS makes is at what exact points are you shown an image.

the only displayed change FPS makes to ROF is a very, VERY minor change in the accuracy of the first and last shot and at what point in MP your packets are being sent/recieved.

Ill take a 20fps video stream where you know the exact frame information VS a 120fps hand timed piece of shit every day of the week accuracy wise.



gunna say it again... you can time this doo-doo by hand all you want, it wont be accurate on any system LoL

/thread

[toysrme]

fred, den said he did a screen capture on his PC with fraps, the rounded to arbitrary numbers.

[azrael1911]

Dec 30, 2009 9:50:21 GMT -5 mannon said:

It appears to actually slow the time to empty the magazine and thus actually slow the rate of fire according to tests, stupid as that is. I still contend that the game should be able to generate all the necessary hitscans to fire a full auto weapon without any slowdown at all even if you got dropped to 10 frames per second, but the game doesn't appear to be programmed that way.

My only theory is that the game actually waits for a frame to be drawn for each hitscan to borrow data from the rendering to find out where that pixel at the middle of your screen is pointing, thus preventing the game from producing hitscans between frames and costing you increasing amounts of time the fewer frames per second and the more hitscans per second.

But even if that's true it could delay the actual hitscan but not delay the timer that ticks off how long to wait before firing off your next shot, so I don't quite get it.

I suppose the game's actual timers could be getting choked by the poor system performance, but I don't understand why they wouldn't scale to the current FPS. On PC I almost always set my games to run with a variable and uncapped framerate and they run fine. But then... this is a highly modded version of the IdTech (Quake 3) engine... There's no telling what sorts of quirks are hanging around in there. Quake engines have traditionally had some very strange framerate related oddities.

ha.
that's not even half of it.

having higher FPS allows you to jump considerably higher, fire your gun faster.

and here's the kicker.

on Q3 engine games, if you manage to somehow reach 1000fps you can walk up any surface with a 30% slant.

that means you can literally sprint up near vertical surfaces provided you have the FPS for it.

[mannon]

Yeah, good times... ;p

[fred]

n1gh7: because both of the console systems are locked at 60fps they cannot accurately show guns firing above a certain RPM. You need to be above 60 fps or so. They cut out bullets or time.

I assume this rate would be something like: 60fps * 60 sec = capped at 3600 rpm? None of the weapons I measured has anything like this kind of ROF. The highest I measured was VECTOR - Bling w/ Extended Mags & Rapid Fire at about 1150 rpm

mannon: Still, it's usually damage that is different rather than ROF I believe.

But did anyone ever actually test the consoles for ROF? If so, did measurements actually agree with the configuration settings Den found? I can tell you that people seem to have tested damage much more than ROF ...

toysrme: gunna say it again... you can time this doo-doo by hand all you want, it wont be accurate on any system LoL

Sure, from many of the comments, I see the ROF is "slippery" to measure. However, if the figures are relatively accurate this is good enough for me as this would yield tactically useful guidelines for weapon selection. Of course you could just try the weapons and see how you play with each, but that would be ... too easy.

mannon: I'd also suggest using extended mags to increase the time to empty and thus minimize human error. And I'd average more than just 3 times
mannon: But it may be a good idea to check the consoles for consistency since they tend to have a few odd weapons with different stats.

I adopted these suggestions: using extended mags, n-1 calcuation, averaging 4 to 5 measurements and using times measured to the hundreth of a second. Again this is on a PS3 console.

The weapons were iron sights with either extended mags, rapid fire, or akimbo or blinged with both extended mags + rapid fire. See Mag Size Used column to determine when I used extended mags.

Highlights are:
* not all the SMG's fully benefit from Rapid Fire: Uzi and Vector are good, UMP Ok, MP5 minimal and P90 nothing.
* the M4, TAR and ACR fire about the same rate. From the sound I always thought M4 was fastest.
* The MGs continue to test close to my earlier measurements.

	Mag Size	# Measure	Average Time	Calc ROF	Comment
M4	45	4	3.72	710
SCAR	30	4	2.99	581
TAR	45	4	3.72	710
ACR	45	4	3.73	708
FN2000	45	4	3.00	881
AK	45	4	4.42	597
MP5	45	4	3.26	810
MP5 -RF	30	4	2.01	867	Only 7% increase for RF
MP5 - RF	45	4	3.03	873	Only 8% increase for RF
MP5 - Akimbo	30	4	2.28	765
UMP	48	4	4.75	594
UMP	32	3	3.19	583
UMP - RF	48	4	4.11	686	Only 16% increase for RF
UMP - RF	32	4	2.64	705	Only 21% increase for RF
UMP - Akimbo	32	4	3.23	577
Vector	45	4	2.98	887
Vector - RF	45	4	2.30	1149	30% increase
Vector -RF	30	5	1.55	1123	27% increase
Vector - Akimbo	30	5	2.13	818
Uzi	48	5	3.21	877
Uzi-RF	48	4	2.48	1136	29% increase
Uzi-RF	32	5	1.68	1105	26% increase
Uzi-Akimbo	32	5	2.24	830
P90	75	5	4.98	891
P90	50	3	3.31	887
P90-RF	75	4	5.00	889	No effect from RF attachment
P90-RF	50	4	3.28	896	No effect from RF attachment
P90-Akimbo	50	5	3.42	860
L86	200	4	16.66	717
RPD	200	4	19.65	608
MG4	200	4	16.64	717
AUG	63	4	6.21	599
M240	200	4	13.35	894
PP2000	30	4	2.44	713
G18	50	4	3.34	882
G18 Akimbo	33	4	2.14	896
TMP	25	4	1.83	789
AA-12	16	4	2.35	383

[n1gh7]

fred I misspoke. What I meant to say is that because a console cannot stay at or above 60 fps all the time while firing you cannot show accurate RPM. I think the limit of RPM on console is 1200RPM because of the 60fps cap. In the quake 3 engine that MW2 is based on there are certain quirks and limits that are based on FPS. I do not claim to know or understand all of them and if you really what to know all of them I recommend you ask Den, a tweaker, or someone that is extremely familiar with the quake 3 engine.

[xenon]

this only affect console? pretty sure rapid fire empty my mag faster at PC - even with P90...?

edit, actually. its easy enough to make a rough test with a crude stop watch.

@pc:

P90 empty mag in ~3.2 sec. 50 bullets give ~900 rpm
P90+Rapid empty mag in ~2.6 sec give ~1150 rmp

UMP empty mag in 3.1 sec. ~32 bullets give ~600 rpm
UMP+Rapid empty mag in ~2.4 sec give ~800 rpm

[KingVaroon]

i thought the P90 was 850 RPM.. and the UMP was 630..