I used to insist on doing every single question available and getting them all right until Dr. Gleim himself called me and told me I was focused too much on the details, to use the Gleim suggested approach instead, and that I would fail if I kept using my current approach. I was floored that he would take the time to review a random customer's study habits and discuss them. I was asking tons of questions and it was before they had the Q&A section so maybe I was the squeaky wheel.
I listened to him and started using the Gleim suggested approach (which states you keep doing random MCQs for that chapter until you get to your desired proficiency level, eg 75%) except I kept going until I could do around 85% or better consistently per chapter.
I still have 600+ questions in REG that I've never seen, not including questions never answered correctly. You can see my scores below, all first attempts. (Other sections were older, questions have been updated, so I can't say reliably how many I had unanswered.)
Gleim is high quality review material. Use their approach and you should do well (even though it initially did not make sense to take a MCQ on information I had never reviewed and score consistently around 25% on pretests, it actually helped tremendously when I studied the material). The only thing they are lacking for me is flash cards, so I made my own. They are working on making this part of the Gleim set so I sent them copies of mine to help with the process.
Also I did all the SIM wizards as well, so two SIMs per unit. They take a while, sure, but SIM 2 usually tests on things SIM 1 did not, and they are worth the time.
PS: There is a very big difference between doing a certain % of the total questions and doing as many questions as you can (randomized, so some new and old) and scoring 85%-90%. The latter is actually more difficult and better predicts how you will do. I may not have done 90% of the roughly 2,500 questions, but I've probably answered certain questions 10 times. I used randomized sessions almost exclusively.