Pile shuffling one time before a game starts is not slow play. Pile shuffling has a way of significantly transforming the starting configuration of your deck and washing away all patterns that used to exist due to clumping (while possibly introducing new and very different-looking patterns). Because it is deterministic, it can have a way of putting the final configuration of your deck into a predictable pattern if the starting configuration was known and easily memorized. This is why, if you rig your deck between rounds, and immediately go into a pile shuffle when you first sit down for a match, you can use it to stack your deck or mana weave (i.e. cheat). But if you perform even one riffle shuffle before you begin pile shuffling, then all of the deterministic aspects of pile shuffling are negated and it can't be effectively used in any cheat (other than possibly rigging the top / bottom card of your deck).
An aside about randomization: if the final configuration of a deck is even a little bit influenced by the starting configuration of the deck, then it is not randomized. For instance if a Dark Ritual was adjacent to a Phyrexian Negator, and they remained adjacent during your entire shuffling ritual, then it is not randomized. If your deck starts out with a better-than-random mana distribution, and you perform a series of riffle shuffles in which it consistently sustains a better-than-random mana distribution after each shuffle, then it is not randomized.
So anyways, my studying of deck randomization has revealed that dozens of riffle shuffles are less effective at washing away the starting patterns of a deck's configuration than a single riffle shuffle followed by a pile shuffle. This is especially so because some sleeves tend to stick together a bit, and are capable of sticking together through the course of quite many riffle shuffles, therefore tenaciously preserving information about the starting configuration of the deck. A lazy riffle shuffle that interweaves large clumps of cards can fail to erase certain deck configuration properties, such as weaved mana (i.e. if your deck is perfectly mana weaved, and then you perform a loose riffle that drops clumps of 8 cards at a time, you will likely find that the deck is still highly mana weaved) or combo cards that are adjacent to one another. That is why I would say that a mixture of riffle and pile shuffling is not a slow way of randomizing your deck, but actually a faster way of really randomizing it. If you can fully randomize your deck at the beginning of a match in under 3 minutes, then you should not be worried about violating slow play rules. However, magic's historical convention for sufficient randomization required at least 3 riffle shuffles, and I would not recommend any shuffling algorithm in which fewer than 3 riffle shuffles are executed (regardless of what additional actions you are performing instead).
Performing more than one pile shuffle creates numerous issues. First, it can be used to mana weave a deck. Second, there is a way to perform two pile shuffles in a row such that it restores the deck back to its original configuration before you started the pile shuffle, and if you also rigged the order of your deck before the match begin, then this would execute the ultimate cheat -- the fully stacked deck from top to bottom. Third, two pile shuffles is just as deterministic as one pile shuffle, and does not provide any greater randomness, nor any greater ability to separate adjacent cards than a single pile shuffle provides. Fourth, two pile shuffles takes twice as long as one pile shuffle.
Another thing to consider is the timing of a pile shuffle.
- Since a deck's state at the beginning of a match is not known to be random, and could in fact be rigged, significant randomization effort should be taken before the first game begins. This is the time in which a pile shuffle is most justified. A pile shuffle is more likely to be an authentic attempt at randomization and not a cheat if at least one riffle is performed first.
- Once players are done with their opening randomization rituals, provided that a player does not like their opening hand and proceeds to mulligan it, their follow-up randomization ritual should not be as long as the first one. This is because the deck has already been sufficiently randomized once (and if it wasn't, then they already violated a rule), the player has no way of knowing whether the rest of their deck is also badly clumpy, aside from being slightly skewed by subtraction of the 7 cards they did see (unless they do know, which means they know their deck wasn't sufficiently randomized, which violates a rule). So the follow-up shuffling ritual is starting from a fairly randomized point, and should not take significantly more randomization to get it to a new randomized state (in other words, a second pile shuffle at this point is not justifiable). Some players indeed get away with pile shuffling after taking a mulligan, but I think a judge would at least look at the total context of how the match is progressing, if either player is up a game, how long the entire shuffling ritual is taking, and the pace of play.
- Once a game begins, if a player is forced to shuffle their deck due to a game rule such as sacrificing a fetchland, then a pile shuffle is very much not justifiable, and could well be considered stalling. Some decks require to be shuffled dozens of times per match due to game rules like fetchlands, tutors, etc.; and a one-minute shuffling ritual for each game rule for each player could all by itself eat-up more than the match's allotted time, which is why you should strive to get some riffle shuffling done efficiently and very quickly.
- After each player performs sideboarding, it is likely that they have seen the exact configuration of their deck (but it is unlikely that they are capable of memorizing it). It is possible that they picked-up the cards from their previous game in a way that simulates a partial mana weave. The player might have inserted their sideboard cards into their deck in an even distribution. All of this puts their deck in a more favorable than random state. However, it is not a state that is rigged well enough to then mana weave it using a pile shuffle. Since it is currently in a more favorable than random state, two strategically executed pile shuffles would preserve the current order of the deck, and produce an end result that is more favorable than random (therefore, two pile shuffles at this point could be used for cheating). In other words, the deck is in a state that is less random than when players take their first mulligan, but more random than when they sit down for a match, so a reasonable shuffling ritual here should be one that is just slightly faster than the beginning of match's shuffling ritual (i.e. 0-1 pile shuffles and several riffles). There is no reason why a riffle shuffle would be obligatory before a pile shuffle at this point, provided that the player did not have an opportunity to rig their deck in all-lands and all-spells clumps. Depending on the total context of the match's progress, you can probably get away with a 3-minute shuffling ritual at this point, but my personal belief is that it should be capped at closer to 2.5 minutes.
On another note related to cheating with pile shuffling, I should say that at an event with a REL level, you might fail a deck check if you sit down for a match with a deck that is already in a rigged state (i.e. all spells on top and all lands on bottom).