Not a practical system
The system was never intended for use by a human. Aside from the impossibility of keeping so many different counts, the gain from the use of such a system is absolutely negligble, not more than a few hundredths of % in virtually any shoe game you are ever likely to encounter, amounting to at best a couple of extra cents an hour. For further information on the utterly negligible gains from the use of a computer see Bill Zender's "Card Counting for the casino executive".
When quizzed on this forum about the issue, Grosjean's argument was that the system was designed to be employed by a robot against an online game. Reading between the lines he must have been talking about the Stanley Acropolis game, since that was the only online casino that matched the specified table limits, rules and penetration of the time.
In retrospect I can only assume that Grosjean made a serious error of judgement trying to employ this elaborate technique at Stanley Acropolis. SA were offering a 20% sign-up bonus up to $1000 at the time, as well as a recurring 5% bonus. In addition, you could shuffle up at will on any negative counts. Finally, there were incredible one-off promotions such as a day's worth of 2-1 blackjacks which cost Stanley over a million dollars. In short, nothing terribly sophisticated was required to beat that game, not even count cards if you didn't want to.
Many new blackjack students are fascinated by the idea of computer perfect play but the reallity is underwhelming to say the least. Even if you can do it that will not materially affect your results except in single-deck games with very deep penetration. The brainpower would be better devoted towards some external technique such as shuffle-tracking.