Main > Main Forum
Encoder Comparison Pages Back Online
patrickl:
--- Quote from: RandyT on August 14, 2004, 06:27:48 pm ---
--- Quote from: patrickl on August 14, 2004, 03:11:17 pm ---
I'll just recap my (deleted) posts:
--- End quote ---
I think you mean "re-write history". See the quotes in my posts for what was deleted (I knew there was a reason I do that all the time.)
--- End quote ---
Again you completely misinterpret my post (buffers have nothing to do with transmission times and I don't say keystrokes are lost, but that they arrive late), then you make fun of your own idiotic interpretation and to top it all off you try to sidetrack the discussion with subjects that have nothing to do with the discussion at hand.
Tiger-Heli:
--- Quote from: RandyT on August 14, 2004, 06:27:48 pm ---Ummm...I still use one of those systems with a big ol' AT to PS/2 keyboard adapter as part of my development system :D
--- End quote ---
Forgot to mention, actually on my PC I use an AT keyboard, into a PS/2 adapter, into a USB converter . . .
RandyT:
--- Quote from: patrickl on August 14, 2004, 06:45:49 pm ---Again you completely misinterpret my post (buffers have nothing to do with transmission times and I don't say keystrokes are lost, but that they arrive late), then you make fun of your own idiotic interpretation and to top it all off you try to sidetrack the discussion with subjects that have nothing to do with the discussion at hand.
--- End quote ---
Well that was one part of the two+ pager I just wrote. How about the rest? Everything I wrote has something to do with this topic.
I did misunderstand the point of that statement, but what I said is still true and the buffer is still important there. Let's look again just for fun.
--- Quote ---So with PS/2 you could release at max 8 keys within a frame before you actually start suffering more than a whole frame delay due to transmission speeds.
--- End quote ---
The software is responsible for retrieving the keyboard data from the buffer. It might do it every 17ms, as you stated, or every 35ms or ? ? ?. It's up to the software and has nothing to do with the interface.
But let's use your 17ms as though it were somehow set in stone (keeping in mind that it is certainly not). But before I go on, and for the sake of others who might not have a feel for the time frame of 17ms, do the following:
Speak the words "one mississippi". Did you do it? Ok, in the time it took you to do that, 17ms went by 60 times and was far less time than it took you to make the "wuh" sound in "one". Now let's continue...
Based on the maximum speed specifications of the PS/2 port, which is in reality about 1/2 millisecond per 11-bit frame, it would be possible to send 34 button presses or 17 releases or a combination of the two with the appropriate number differences.
Now, in order for any button "event" to be delayed by an entire "frame", umm... 17/1000ths of a second, it would need to occur at the precise instant the software was finished emptying the buffer and went on to the rest of it's business. If it occured 6ms afterward, it would be delayed by only 11/1000ths of a second, 12ms afterward it would be delayed by 5/1000ths of a second and so on. In order to be more than 1 frame late, you would need to have, on average, more than 24 button transitions that occur in 17/1000ths of a second. If you tried with every bit of determination you could muster, you, as an individual, would have virtually no chance of making 5 transitions occur in that amount of time, let alone 25.
4 players could be hitting or releasing 2 buttons each with all of them hitting or releasing diagonals (not just holding a direction), in a window of 17/1000ths of a second, and delivery of the data via PS/2 still would not exceed a 17ms frame. Again, exceeding the maximums merely by chance, even with 4 players banging like mad, is a virtual impossibility.
There have been encoders on the market (long fixed by now and certainly not mine!) that had problems with games requiring only 3 presses within a "frame". This was due to poor firmware, questionable hardware design or both. But as I said, the culprit has been fixed as far as I know.
The question is, what makes you think that the extra data and processing steps required by USB aren't going to give you similar or possibly worse performance?
But here's something that I found interesting during my research, and someone please correct me if I'm wrong, but I'm pretty sure it was from the HID white paper or something of similar reliability (MS?)
The USB keyboard data is delivered to the host in packet form, meaning that is a collection of bytes with header and data information. The packet is sent serially (one bit after another) and then "re-assembled" into a packet again at the host. The host then identifies the packet as being one from a keyboard, pulls out the keystroke data and puts it into a virtualized keyboard buffer. But here's the interesting part...as the host deals with the data as a "packet" there is no particular order in which it is placed into the virtualized buffer.
This would mean, if I'm not mistaken here, that if you and your competitor were in a situation where the first one to press the button were to win, and you both pressed close enough together that your keystrokes ended up in the same packet, the winner is pretty much picked at random :)
So it's quite possible that the FIFO system of PS/2 is superior in this regard.
Until you examine the full workings of a device, meaning all internal protocols used in communication with other components and the methods used in the microcode, you won't ever get a clear picture as to the real performance capabilities the device can offer. Regardless of whether the final output is USB or PS/2. Remember, the output of anything can only be as fast as the slowest part coming in or the slowest part of the process as a whole.
RandyT
krick:
--- Quote from: Bgnome on August 13, 2004, 07:04:07 pm ---i think that there are still a whole lotta people, myself included, bringing life back to old, outdated systems. dont leave us out of the loop..
--- End quote ---
This is precisely why I think there really needs to be a PCI version of the ArcadeVGA card.
Andy, are you listening?
Bgnome:
this talk of bandwith has got me worried..
if i have a splitter on my ps/2 port for my trackball and encoder, will i suffer performance issues? keeping in mind that i intend on using it in a 4-player situation and that the data/clock lines are all separate..