It’s a somewhat common practice in PC FPS to overclock the USB Polling rate of certain mice to get a performance boost. Is such a thing worthwhile to do with a fightstick? Has anyone tried and got any “safe” numbers? The fightstick I’m using is a MadCatz SE.
If you don’t have any input lag from when you’re activating the switches on the stick, then what would the “performance boost” be?
Default Windows USB settings have a small lag on mouse inputs. If you up the USB polling you decrease the lag. Helps accuracy and response for FPS gamers.
PS3 and 360 don’t run on Windows (obviously), hence don’t use the same USB settings, and I really doubt they have the same lag (up to 8ms on Windows).
It’s a non-issue.
it’s worth noting i’m playing the pc version. the performance boost you get by upping your polling rate on a mouse can make certain mice such as the intellimouse optical which is nearly a decade old as good if not better than most recent gaming mice. I don’t know what the polling rate is for the xbox360 and PS3 hardware, but I’d imagine it’s higher than the default 125hz/8ms rate that you’re stuck with on the pc.
The reason people increase USB polling rate on mice is that the laser takes more measurements and the analog motion is recognized more accurately.
As joysticks are digital, your game only takes input once a frame, and the default polling rate is already several times a frame, if this mod is possible on a joystick I don’t know, but it’ll only be a placebo effect.
^Exactly, this is not worth the effort.
Actually, you may call it a placebo effect, but the fact of the matter is that the difference between the 125mhz the USB port usually runs on and the 1000mhz it can run on make a definitive difference when trying to play on joystick on Windows. Games don’t specifically take one input per frame; there can be a very large number of inputs in an absurdly low space of time. This is actually magnified when you have more than one person playing on separate joysticks on the same computer.
There was a program that allowed you to up the polling rate on Windows XP but I haven’t found anything comparable for Windows 7, which I have a problem with right now. The stick is laggy in 7, while it was just fine in XP (where I had my polling rate adjusted; before that it was laggy as well).
Wow.
This thread was inactive for exactly a year before you resurrected it. .
i am sorry dude but you really speak nonsense, you obviously only remotly understand that.
and first off just so you know you can (try to)overclock a ball mouse too, it has nothing specific to “optical” devices, it is just more noticeable an improvement on a high end optical mouse compared to a low end optical/ rubber ball mouse. Plus low end devices MCU may not work properly if overclocked.
you are mixing up read accuracy and polling of the optical device by the MCU inside the mouse and the polling of the usb device (the other end of the MCU) by the USB host (the PC)
what you speak of is the diffrence between a 800 dpi sample and a 4000 dpi sample on an optical mouse, meaning that the MCU has more processing power and can analyse reliably high accelerations and speeds of the mouse over the surface while a lesser one will just give up and reset, giving sketchy reads and handling them to the PC.
What the OP speaks about is the USB polling rate, you can have the most sensitive optical device in the world it is still going to be polled every 8 miliseconds by the windows HID device driver, so even if the data is very accurate and smoothed out it still is handled to the game with up to 8 miliseconds of lag,
now 8ms might seem like a puny number… it is not, in a FPS that is kind of making your handicap twice or thrice the amount of a correcly geared opponent. in a fighter this is important too because it is SNOWBOWLING.
Once again in fact once again you are wrong, if you overclock your mouse (optical or not) you actually have fewer mesurement because you only have 1 or 2 ms of measures instead of 8, what you improve is the diligency at which those measurements reach the application (in our case the game),
It is also hard on your CPU and that is why microsoft chose the default to 125Hz at the time XP was made, having a core doing only mouse read decyphering wasn’t an option, but a modern gaming computer doesn’t get bottlenecked by an overclocked mouse. most gamers set their polling rate to 500 hz which is a good compromise.
now this was the easy stuff most hardcore gamers know and this is the only thing a FPS player has to care about. unfortunately it gets more complex when it comes to fighter.
as you said the time reference for a game like street fighter is a “frame” which is basically a one calculation of the iterative process of the game, (and has very little to do with graphic output at the point we discuss, though we use the same word for both).
in real time
-the driver when he gets some love from the thread handling of the CPU core polls the USB host. (this is normally negligeable unless you have an old computer and tons of shit running)
-the USB host polls everything plugged to it by order of priority (usb keys, webcams, printers, scanners, cameras, card readers, external drives, keyborad, mouses, gamecontrollers… etc) and receives so many answers (0-500ms)
-when he finally gets some time to poll your controller it gets in return a report of the states of the layout on that controller (that where the 0-8ms of the OP factors in)
-then the USB host handles that back to the software level (the HID driver)
-the Driver when he gets some love from the CPU sends it to the application( SFIV) after transforming it from states to presses/holds/releases codes
-SFIV when it gets some love from the CPU runs its input system factors in to compare it to previous data and detect special motions, buffered charges/motions etc…, eligible moves (that is the level where the so called “options selects” take root) then it handles instructions to the game engine.
-game engine sends data to rendering engine (16.6 ms currently calculated frame)
-that sends data to graphic device drivers,
-the GPU does whats asked of him buffers the “frames”(the graphic ones this time) in the ramdac and unfolds them at the graphic card output refresh rate.(0-16 ms per frame depending on refresh rate)
-ramdac is polled by the TV at the TV refresh rate for the images “feed” ( from there we have a minimum of 0 to 120 ms of what is called input lag but should be labeled display lag)
so if you add everything up you have on average a pretty much impressive amount of lag of around 50 ms on the input/game side and another 50ms from the TV, human reaction time to a expected event while being focused is around 120ms (neuroscience 101) so you basically doubling that, but still as all of you have pointed it out so far 8ms is not accounting for much (it is still around 4% overall.)
BUT we would miss the real problem if we stopped the analysis here.
the main problem is that lost of stuff happen in 8 ms on a controller
case A - Lets say you do a quick shake of the joystick front back to trigger a charge back-> forward motion and start charging again straight away, if you are not lucky everything happened whithin the 8 ms second( or more) slot between 2 polling the front input is never handed to the PC by the controller MCU because it is not active anymore at the moment of the report, and it wasn’t yet at the previous report. Your special just never happens and in 100 ms nothing is going to happen on the screen where you were waiting that antiair and you 'll eat the jump-in. now poll your controller every 1 ms and that is not likely to happen EVER.
case B - you do a hadoken motion and press light punch, you are a mere human so you press high punch 1 miliseconds earlier than the down microswitch is released, and in that time frame the USb report is sent, 100s later you see a nice lp.shoryu coming out and not the expected fireball because what was originally a 1 milisecond mistake turned into an 8 ms mistake during which maybee the game engine had to finish the calculation of the frame.
now poll you controller every 1 ms and this may still happen but is is going to be offset into another 1ms mistake, and the probability of a frame calculation to be closed during these 1 ms is much lower than that during 8ms.
the list is endless, unconsistent piano inputs, unwanted negative edge, over the top leniency, inconsistent Plinks, inconsistent karathrows, 1 frames links turning into half a frame links or 1 and a half frame links.
I beg to disagree
It may look like nothing but 7 ms even if it is not going to save the day, it still is that much less handicap on your side.
I don’t know anybody playing this game even remotely seriously who would play on a wireless gamepad and this is exactly the same argument and similar lag values that motivate the choice for a wired controller. Why go all out against wireless communication lag, but keep silent when it comes to polling lag, these are just both sides of the same coin.
Heck some even go as far as replacing their brand new happ/crown/knock-offs buttons with lighning fast response buttons from sanwa, and that is for a milisecond gain… at best.
I totally agree to say that all that sh** isn’t going to turn a lamer into an ace, but as one said :
~ Victory is won not in miles but in inches. Win a little now, hold your ground, and later, win a little more. ~
Having a very responsive controller might not remplace skill, hard training and heavy practice, but it is nowhere nearly as negligeable as some of you guys 'd like to make it sound.
That program never worked for me even on XP, ( i solved that by using high end ps/2 keyboard encoder, they use virtually asynchronous protocol, and lag is around the 1ms just like overclocked usb),
The fact is on top of all described above, controllers firmware are not always properly optimized and sending all data when they are polled. That 's because they have anologic axes and triggers to report for as well and a truckload of buttons so the whole description of the pad may or may not take several iteration of 8ms, effectively demultiplying the above stated problems to 16ms or even 24 ms
Firmwares may or may not buffer inputs.
I very much hope so, though i found no information whatsoever about that on the web, it would only make sense since they had to compensate for the 2.4Ghz relay induced lag on the wireless controllers, In any case I doubt i’ll ever play the game on such a ghetto gaming system as long as we are provided with a superior build of the game to enjoy on PC.
Serious gamers/typists use hi-fi mechanical keyboards with high speed keystroke filter and Ps/2 interrupt base protocol to play/work, or overclocked proprietary driver based USB keyboard.
There is always input and display lag on SFIV the all accounted measured minimum was around 50ms if i remember correctly, for 95% of the playerbase it is somewhere between that and 150ms. with online play the sky is the limit.
A normal being can consciously notice the diffrence between 50ms lag 250ms and 500ms, but making out between 50ms and 80ms is impossible, but it is not without consequences, people can “feel” it during gameplay.
Well if you are still playing the game (and reading these forums) you probably have worn off those SE buttons, keep the nice madcatz box, buy a high end Ps/2 encoder (35$) and replace your stick and buttons with japanese gear( 40$), and while your at it get a bit of heavy stuff glued inside to weight it so it is more steady on your desk/lap.
You’ll solve many a problem all at once, some of them more of a handicap than usb Hid input lag.
^^lol you lost me on the first line…
Most games only poll for inputs 60 times per second (16.6ms). The default USB polling rate should be more than adequate.
tl;dr this is pointless when you play on console regularly besides PC because it’ll all feel laggy?
When you play something like stepmania even a 5ms lag difference is noticable. A 30ms delay you can even hear almost clearly.
Wow talk about elitist haha, but I don’t think you’ll ever see MvC3 coming to PC. I think SF4 was more of an experiment considering melty blood and Guilty Gear were released on PC as well. I still have hopes for the final build of MBAA coming to PC though.