Bob, if you change injectors you must change chips. I use a RadioShack A/B printer switch so I don't have to plug and unplug cables, to switch between DirectScan and a Pocket Programmer, with an inverter plugged into the cigarette lighter to power both laptop and programmer, which sit in the passenger seat. Make a pass, park, review the ds log (<1 to 5 minutes for me depending on what I'm studying), record data in notebook and plan changes (<1 to 5 minutes), open chip image in winbin, make changes, save file (total <1 to 5 minutes), flip a/b switch, open file in eprom programmer program, insert blank eprom, program (2 minutes), insert eprom in carrier, insert in ecm, flip a/b switch back (another 1 minute), and go again. Not counting the notebook and thinking time, between 2 and 10 minutes depening on how extensive the changes are, with no extra hardware - fast enough so cooldowns are the rate determining step. Keep a bag of blank eproms and erase them in bulk that night. In reality, I made a little cable/board/adaptor so I have a flash ram chip sitting within reach, on the trans tunnel, instead of digging into the ecm each time, and it programs in about 10 sec and doesn't use the carrier, so all that is even faster. No on-the-fly changes but I can't write that fast in my notebook anyway
. The editing is probably about the same time as using the fast or accel software, their downloads are faster than my program eprom/insert, and to me the rest is a wash.
I guess I see three gm ecm configurations to consider, and I'm ignoring spoolup and ae stuff to focus on a 1/4 mile pass after the launch.
First is a stock maf or translator/lt1/ls1 but no extender. The maf will be pegged sometime during spoolup and all fueling after that is "blind". The programmer sets up some amount of fuel delivery and it is up to the driver/tuner to make the airflow match to arrive at some target afr. The stock O2 sensor and perhaps an egt meter can be logged and adjustments made after each run. This can be burning a new chip or turning a thumbwheel to a new setting, and includes the max effort setup (which also allows changes to be programmed in with key on engine not running, which is flexible but has its own learning curve for the tuner).
Second would be an extender setup, where the maf stays on scale for the run. Now the fueling can be made to match the airflow for the entire pass, during the pass, using what is essentially a preprogrammed target afr. However, the tuner doesn't know what that afr is going to be before a run. He can just richen or lean things out, and then has to see what the new afr was. To change that afr, I believe there are setup changes that can be programmed in with the key on and engine not running, so there is some learning curve associated with this. Again, data can be logged in one pass and then used to adjust the afr for the next pass.
Third is the mafless ME setup. This replaces the maf with a map/speed-density system to remove the inlet restriction of the maf. Otherwise I see it as equivalent to number two - airflow measurement stays on scale but tuning is open loop and it is up to the tuner to log a run and use that to adjust for the next run.
Number one is also susceptible to problems like boost creep, where the airflow changes but the ecm doesn't know it and the chipmaker didn't anticipate it, while two and three handle that okay. I'm focussing on fueling here because the rpm range after launch is so narrow that I'm assuming constant and equivalent timing from all ecms. Yes, an independent wideband o2 can be added but it will most likely have to be independently datalogged and so analyzing all the recorded data will be more time consuming (see the innovate! wbo2 thread in the scan tool forum for a great looking new unit).
The aftermarket ecms, fast and gen vii+, are both speed density so the airflow measurement is always on scale (yes, I'm assuming a correctly sized map sensor), so they are best compared to 2 and 3. The big advantage they offer that I see is the integrated wideband O2 sensor. At a minimum, it can be datalogged easily with perfect alignment between it and other engine variables like rpm and knock, and used for tuning the next pass. A good argument can be made that tuning using logged afr's is more intuitive and easier to do than tuning using stock O2 voltages and/or egt like the stock-based 2 and 3 above. Yes, buying a fast or vii+ is an expensive way to get a loggable wbo2 but if you don't mind paying for convenience it's your choice. However the wbo2 can also be used in closed loop mode at wot (at least with the fast; no one answered my question if this was possible with the gen vii+). To me this is the biggest advantage since not only is tuning done after a pass it happens during a pass, making each pass potentially less damaging and cutting the required number of passes to arrive at a final tune down by at least a factor of two (based on conversations with at least 10 fast tuners and a few other chip tuners besides myself). If it saves just one set of headgaskets while a new and impatient tuner is trying to optimize his car's performance, the fast or vii+ has paid for itself. Then, the tuner has the choice of leaving the setup in closed loop mode or going back to open loop. Closed loop may save the engine if something else fails in the fuel delivery system, like the fuel pump or filter, and can hurt the engine if the sensor itself fails in an "unlucky" manner. From talking with many racers the sensors very rarely fail so to me the choice would be to leave it closed loop, but each tuner can decide for themselves.
So to summarize, the stock based systems are cheaper, but if you add an independent wbo2 (about $400 for the innovate! and interface cable) and peaknhold drivers they will be about $1000 total (including translator/maf or map), vs $2300-2500 for a fast or vii+. My guess is that the learning curve for each setup with a wbo2 is fairly similar - two-thirds of it learning how to tune and one-third learning how to make the changes. Without the wbo2 to go with 1, 2, and 3 I believe that the learning curve is steeper and it will take longer to get a car tuned. Support for 2, 3, fast, and vi+ seem to be readily available from the vendors; making your own chips for 1 requires the most time spent hunting down all the tools you will need from various places. The stock based systems are only applicable to the tr's at present; each vehicle and ecm combination seems to be growing its own set of stock based systems (sy/ty, lt1, ls1, etc). The fast and vii+ are universal; they are used on many different engines so any tuning skills based on them will have a much larger market base so clearly, any tuner shops will prefer them over the limited customer base possible with a tr-specific setup. That limits support for 2 and 3 to just the vendors, plus other racers. A big chunk of the price difference between stock based and fast/vii+ is probably explained in the markups needed by the distributors, and a lot of this is used to fund the tuning support given by those distributors. So, every user has to decide for themselves how much help they will need and how many sources they want to be available for that help, and factor that into their decision as well as the price.
Anyway, that's how I see the debate, for drag racing at least.