View Full Version : "Negative Acceleration?" ugh...

12-16-2008, 02:50 PM
Ok I'm having a problem.

When I flick the mouse very fast in CS: Source, the mouse seems to slow down significantly than when I move it a bit slower, but th same distance. I know this is called negative acceleration, because negative accel means just that: you have to move the mouse physically farther, to turn a certain # of degrees, if you flick the mouse faster, than if you flicked the mouse slower. This kicks in at a certain "physical" movement rate.

Now I know that you don't need to do 180 degrees very often (which is why some of the best CS players use really low sens like 45 cm for 360 degrees), but I'm used to fast flicking in Unreal Tournament 2004. But CS:Source does not seem to like it very much, and it's difficult to adjust to it.

Hardware is Razer Deathadder mouse (newest firmware and drivers).

I already know it's NOT a mouse problem because I can flick the mouse as fast as my arm will physically let me, in Unreal tournament 2004, and there is no negative acceleration at all. And I'm talking about unrealistic fast flicks, which I tested alot (both with and without drivers installed). But in Source, there is pretty massive negative accel if you flick too fast. I'm using sensitivity of 0.8, and windows at middle notch of course, and 1800 DPI.

Oddly, using a formula someone mentioned on a mouse forum, if I use 450 DPI, and sensitivity of "3", I do NOT see any negative acceleration at all, (which you would never do in a game), although due to a hardware limitation at 450 DPI, you stop gaining faster turning after a certain point (ESReality's Deathadder review mentions this).

Acceleration is completely removed from XP with Aion (anir's) patch. Tested it with accel enabled (in XP, patch removed, but Enhanced pointer precision off, of course), not in Source) and no difference.

Only reason I'm asking if there's a fix is because using 450 DPI is really a step back in precision and accuracy over 1800 DPI.

-noforcemspd -noforcemparms -noforcemaccel does NOT help anything either.

Anyone have some sort of solution? I am pretty sure that CS:Source IS using "directinput" (CS original does NOT, and the negative acceleration is even WORSE in CS original (quake3 and ET players also have it)).

How come 450 DPI isn't showing negative accel but 1800 DPI is? And how come UT2004 is not affected ?

12-16-2008, 03:12 PM
well to be honest the hardware limitation @ 450 dpi is beyond human ability, so if you don't want negative accel you're going to want to use 450 dpi. after all i believe the deathadder has built in 1000hz polling rate, so the difference is negligible. low sensitivity + high dpi = negative accel

12-16-2008, 07:26 PM
I'll do that, but the problem is, i don't get negative acceleration in other games, only in CS (and Source).

I actually just tested this in soldier of fortune 2.
With /In_Mouse -1, I got negative acceleration and it was quite noticable.
I basically flicked the mouse hard, about 6 inches, and I only turned like 90 degrees :( (sens was set to the lowest), when if I moved the mouse slower the same distance, I would turn a bit over 180 degrees.

With /In_mouse 1, it said "initializing directinput", which means directinput was being used, and I did NOT get any negative acceleration anymore, (I remember reading that there was some sort of drawback to using directinput in these games, (Quake 3, Wolfenstein, ET) although I don't remember what they were. I was able to flick the mouse as hard as I could and got the 180 degree turn.

But this isn't about soldier of fortune 2 :)
Does this mean that CS: Source is NOT using directinput?

If it does NOT use directinput, is there a way to make it use directinput?

Based on my searching on google, ALL non-directinput games WILL have negative acceleration at high DPI.

12-17-2008, 01:22 AM
CSS uses direct input, direct input won't(shouldn't) be affected by Windows mouse accel tweaks or whatever and doesn't suffer from inherent acceleration issues.

Take a look at what fps you're getting, low or fluctuating fps might cause this... make sure the death adder is at 1000hz too, etc

12-17-2008, 05:27 AM
i HATE when this happens.

it seems random at times like alt tabbing and then going back in usually solves it when i have encountered it.

my friends and i call it anti mouse accel

12-17-2008, 08:58 PM
CSS uses direct input, direct input won't(shouldn't) be affected by Windows mouse accel tweaks or whatever and doesn't suffer from inherent acceleration issues.

Take a look at what fps you're getting, low or fluctuating fps might cause this... make sure the death adder is at 1000hz too, etc

Guys i've done everything here.
I'm 100% sure that either CS:S is NOT using directinput, or something is really wrong with its mouse code.

I tested this in both Wolfenstein and in soldier of fortune 2.

When I set IN_mouse -1 (which uses windows settings), I noticed the EXACT Same type of negative acceleration that you would expect, in those 2 games, and the EXACT SAME type of negative accel that *BOTH* CS, and CS:Source is giving me.

When i set /in_mouse to 1 (after i had to turn off my anti virus which thought a switching to directinput was an attempt at a keylog, lol), there was NO negative acceleration at all (until i flicked the mouse (diamondback on that computer) so fast, that it exceeeded the ability of the mouse. (I was not able to exceed the Deathadder mouse at all, no matter how fast i threw it around).

But in *BOTH* counterstrike AND counterstrike source, I get the exact same negative acceleration as if the two quake3 engine games were using in_mouse -1.

Which tells me that Cs:Source is not using directinput *correctly* (if at all).

I know its doing SOMETHING with directinput, because it doesn't turn on enhanced pointer precision by default, like original CS Does when you run it (which is why you need those windows XP acceleration fix patches and those parameters).

However I have a theory here and I'm going to post back:
I'm going to try removing the acceleration patch, enabling enhanced pointer precision and then running counterstrike, to see if the mouse starts acting funky, during slow movements, along "close to diagonal" movements.

If I notice CS:Source doing the same thing that original CS did, with flaky slow mouse movement (due to windows xp acceleration being turned on), then its not really using directinput or its not using it correctly.

Because everyone knows that NON directinput games suffer from negative acceleration with high DPI mice, as well as pointer movement problems in some games (IF windows xp acceleration is enabled). But everything current (except crappy mickey mouse free games) uses directinput.

I'll post back with the findings. i'm going to remove the accel fix patches and then force acceleration in XP.

12-17-2008, 09:56 PM
Take a look at what fps you're getting, low or fluctuating fps might cause this... make sure the death adder is at 1000hz too, etc

Read this real careful before you embark on your conspiracy theories

12-18-2008, 01:42 AM
My DA is set to 500 hz.
And my framerate was rock solid at 85 FPS. I do have a high end system--FPS is not a problem for a QX9650 @ 3.7 ghz and a Radeon HD 4870.

Anyway, i noticed something, which is what i spent the last 3 hours testing.

My 2nd computer, which is running a pentium4 3.4 ghz, has zero negative acceleration at all, in CS:Source, except at the Diamondback's sensor limitation, which the Deathadder far exceeds on that system. Running ATI X850, btw.

Wondering what could be the difference in what was causing it, after checking all cvars, and doing all other sorts of things (like uninstalling mouse drivers, etc), I noticed first: g...the 2nd computer was running 100hz refresh rate.

so I decided to up the refresh rate on this system, and noticed slightly less negative accel...but it was still quite noticable. Not quite as bad as 85 hz, but it went from being terrible (at 85 hz) to being "bad" at 100 hz.

Hmm...so I jacked up the refresh rate as high as the monitor could go (at 1024x768)-150 hz.

At 150 hz, there was still negative acceleration, but it was very slight. Nowhere near enough to affect any sort of gameplay.

I then underclocked my HD4870 to 250/400, and noticed no change, so this wasn't anything to do with the video card (vsync always fixes FPS related problems, and this wasn't a FPS problem).

That's when I suddenly remembered something about Quake3 and Unreal tournament '99 completely spazzing out if the FPS was too high....but those were all solved by turning on vsync. I then realized my 3.4 ghz pentium4 was over twice as slow as my 3.7 QX in single core programs....

So I underclocked my QX CPU to 1100 mhz, rebooted, and BAM
ZERO negative acceleration at 1100 mhz @ 100hz refresh rate, unless i flicked the mouse so fast I risked ripping something in the mousepad.
note: turning the refresh rate down to 75 hz started producing annoying negative accel.


Guess this is yet ANOTHER bug report Valve needs:
"Negative accel with high DPI mice on extremely fast CPU's and low refresh rates--something to do with internal engine calculation".

But it's extremely odd that BOTH faster CPU speed (bad) and the higher refresh rates (good) of a monitor affects negative accel somehow...something is really quirky in the Source engine then...

Bug in the game engine. No fix except to underclock my CPU or run my mouse at 450 hz+higher mouse sensitivity... not enjoyable to lose the precision of 1800 DPI...

No one may really care that much now, but when CPU's get 3 times as fast as the current QX9650's and Core i7's, a lot more people are going to start noticing....

12-18-2008, 06:52 AM
Nice work Falketyne, nice work. :)

12-18-2008, 12:44 PM
Set your Death Adder to 125hz, cap your fps at 125, v-sync off, sens at a whole number, monitor hz at 120 or 160, (something close to 125), res at something low like 800x600 and try again. Should remove most rounding the game preforms, see if it's still broken...

12-26-2008, 12:40 AM
I'll try this soon. i wanted to take a break for awhile because, you know, it took HOURS for me to just test the things above...and...I'm sort of lazy.

I'll post back with results later, but I don't anticipate much with this QX9650....

01-19-2010, 02:19 AM

There's recently been a thread (http://www.overclock.net/computer-peripherals/639905-negative-acceleration.html) at another forum, where the guy posted two videos of negative acceleration examples:

There was also this thread (http://razerblueprints.net/forum/index.php?topic=8572.0) at another forum and some sort of explanation:
Source engine games do not use direct input (because its gay) which means that your mouse movement ingame is dependent on how your mouse moves in Windows Desktop.

When you move your mouse ingame, your cursor in the background (windows desktop) moves as well. So everytime that your cursor moves to the edge of your desktop, windows will automatically recenter your mouse to the middle of your desktop screen again. The heavy negative acceleration that you’ve got is due to the fact that your windows cursor has to constantly recenter itself once you hit the edge of the screen.

So essentially: high dpi on lower resolution = negative mouse accel when windows sensitivity is set to 6/11 – it will depend also on the sensitivity you use in game aswell ofcourse.

To fix this stupid problem either:

a) Lower your dpi and keep windows sens on 6/11,


b) Keep high dpi and use higher resolution to compensate,


c) Keep high dpi and your low resolution, but lower windows sensitivity below 6/11 till you find the right spot.

YOU MUST ALSO increase you’re mouse hz to 500 for this to work effectively.

- – – – – – – – – -

Finally You’ll notice games like quake live do not have this problem because they have what i mentioned initially, a choice between:

c) in_mouse -1 (the gay source way)
a) in_mouse 1 (direct input – preferred and default setting)
b) in_mouse 2 (raw input – best, but sucks on some setups hense why it’s been added as optional in a recent update)

I’ve basically had to use 450dpi in tf2/css and 1800dpi in cod/quake at 1024768
to do the math:

if you are at 450 and want 1800 for example:

( S / 450) x 1800

and vice versa from 1800 to 450:

( S / 1800) x 450



..which suggests that software/system does influence this (although taking into account the CPU speed theory there's some disagreement on how - maybe it influences it in various ways) and therefore it seems appropriate to make this known and put this in the CS:S bug list.
I personally haven't tested this myself yet, because I'm busy these days.
But if these findings are true we should let Valve know that we want this fixed, since we want nothing but perfect mouse behavior in CS.

Ideally, we'd also like to hear from Valve directly regarding this, but let's be realistic.. ;)

01-20-2010, 02:43 AM
So apparently the guy from the other forum (http://www.overclock.net/computer-peripherals/639905-negative-acceleration-updated-steams-confirmation.html) got a confirmation that what was suspected is true. That is, that high DPI in Source causes negative acceleration, due to the fact that Source doesn't use direct input, but is influenced by Windows settings etc.
The quote I posted above is said to be right.

That means that the old CS:S mouse optimization guide (http://www.overclock.net/computer-peripherals/173255-cs-s-mouse-optimization-guide.html) (which recommeds the highest DPI always) is no longer correct for all DPIs/resolutions.

06-19-2010, 03:43 AM
i've been using 2000dpi at .8 sens, 1680x1050. what should i try, maybe 1000dpi at 1.6 sens?

06-20-2010, 09:04 AM
I have a razor deathadder and use 1600 dpi I think (either that or 1800) at 1680x1050 and I dont get this at all in css. All I've done in windows settings is disable pointer precision. I am using 1000hz polling rate. I generally get like 100-190 fps in game. My monitor has 60hz refresh rate.

01-04-2011, 12:21 PM
While speaking to someone who makes mousepads / mice for a living, we both concurred that anything over 1300 dpi is actually not used by any professional people. I am not trying to start a flame war or huge argument, just to say that don't assume higher is better. Dpi Does not equal sensitivity. I tried over 20 gaming mice and did all the testing in the world. I found around 700 to be the sweet spot with 1600 by 1200, but because I chose to go with some of the razer mice I had to go between 450 and 900.

In any event. At this point in pc gaming you would think that mice problems would be a thing of the past. Controllers, and consoles have really gimped the mouse progress. But.... atleast our community will usually help everyone figure out how to play perfectly if there is a way.

Pc Forever.

stabby stabby
01-04-2011, 12:34 PM
Wow, epic necro...

You know they've implemented raw input, right? Enabling it should resolve the game's role in any neg accel experienced.

Still, I'm always interested to hear a professional's take on the mystical dpi vs sensitivity conundrum.

Personally, I've been struggling between choosing a lower dpi (350 would be ideal, but I must settle for 300) and sensitity 1.0 to avoid the game mucking up the input with multipliers and "pixel-rounding", or using my g9x's absurdly high max dpi of 5700 for a sampling of as many reports as possible.

Why is 700 dpi the sweet spot for your setup, and what was your in-game sensitivity? I'd love to know how you came to that number.

And, of course, different polling rates create different a different "feel" given different dpi/sensitivity configurations...ugh.