The idea....
There is a synth /1/ on Rhodium for P2P from benzyl-magnesium-chloride and acetonitrile. And there is another synth /2/ from the same imine (what is produced from the benzyl-magnesium-chloride and the acetonirile) is to hydrogenated to amph.
On SM /3/ there was someone who tried it out, but he used THF instead on ether and it ruined the whole reaction. He got bad yields with the P2P and there was no other report except what is on Rhodium. Here the yields what were reported by Poodle /1/ seems more than good (+75%).
The reaction would go on this way:
React the benzyl-magnesium-halide with the acetonitrile in ether to obtain the imine as it is described in /1/. This product would be filtered out from the reaction mixture. It would be dissolved in MeOH and it would be hydrogenated by aluminum and methanol to amph.
Instead of the platinum aluminum and methanol was used. Easier and cheaper...
Experimental part....
The grignard reaction between the (25g) benzyl-chloride and magnesium started slow. The magnesium was activated with iodine in methanol. When the fizzing started and the MeOH started to boil, the magnesium was filtered and added to the ether/benzyl chloride mix. The grignard reaction stared slow, but after 2 hours it just boiled. The temperature was kept under 40 Celsius. Because of the time, just the half of the grignard reagent was prepared... So some benzyl chloride was left in the mix.
The reaction between the acetonitrile and the benzyl chloride was the best part. When the acetonitrile was added a white participate formed. This was filtered (somehow a part of this slurry went through the filter).
There was 3 solution.
Solution one was the liquid what was left after from the acetonitrile and benzyl chloride reaction. Almost no white participate was in it, but a lot ether was in the solution. It was quite dilute.
Solution two was the liquid what was filtered. The white/gray participate. This was dissolved in methanol.
Solution three was the liquid what was left after the filtration, much white participate in it.
Sol. 1 was reacted with MeOH + Al. This was used to hydrogenate it, instead of the sodium borohydride.
After the reaction occurred, the solution was poured on ice cold water (a lot Al(OH)3 formed), this was reacted with cc HCl and than NaOH was added to pH 14. A the honey was collected and the sulfate salt was made from it.
Sol. 2 was reacted with MeOH + Al. The reaction was going on 50 Celsius. And the yield was 0. It was poured it on ice, added HCl and NaOH until pH 14 but no honey.
Conclusion: just dilute solutions of this thing should be hydrogenated and not the concentrated solution.
Sol. 3 was poured on water and cc. HCl was added. P2P formed in 80% yield, awesome
References
/1/: https://www.erowid.org/archive/rhodium/chemistry/phenylacetone.html#grignard
/2/: https://www.erowid.org/archive/rhodium/chemistry/amphetamine.html
/3/: http://www.sciencemadness.org/talk/viewthread.php?tid=5519#pid64089
Pictures and everything else data is coming soon, this was just a "test".
There is a synth /1/ on Rhodium for P2P from benzyl-magnesium-chloride and acetonitrile. And there is another synth /2/ from the same imine (what is produced from the benzyl-magnesium-chloride and the acetonirile) is to hydrogenated to amph.
On SM /3/ there was someone who tried it out, but he used THF instead on ether and it ruined the whole reaction. He got bad yields with the P2P and there was no other report except what is on Rhodium. Here the yields what were reported by Poodle /1/ seems more than good (+75%).
The reaction would go on this way:
React the benzyl-magnesium-halide with the acetonitrile in ether to obtain the imine as it is described in /1/. This product would be filtered out from the reaction mixture. It would be dissolved in MeOH and it would be hydrogenated by aluminum and methanol to amph.
Instead of the platinum aluminum and methanol was used. Easier and cheaper...
Experimental part....
The grignard reaction between the (25g) benzyl-chloride and magnesium started slow. The magnesium was activated with iodine in methanol. When the fizzing started and the MeOH started to boil, the magnesium was filtered and added to the ether/benzyl chloride mix. The grignard reaction stared slow, but after 2 hours it just boiled. The temperature was kept under 40 Celsius. Because of the time, just the half of the grignard reagent was prepared... So some benzyl chloride was left in the mix.
The reaction between the acetonitrile and the benzyl chloride was the best part. When the acetonitrile was added a white participate formed. This was filtered (somehow a part of this slurry went through the filter).
There was 3 solution.
Solution one was the liquid what was left after from the acetonitrile and benzyl chloride reaction. Almost no white participate was in it, but a lot ether was in the solution. It was quite dilute.
Solution two was the liquid what was filtered. The white/gray participate. This was dissolved in methanol.
Solution three was the liquid what was left after the filtration, much white participate in it.
Sol. 1 was reacted with MeOH + Al. This was used to hydrogenate it, instead of the sodium borohydride.
After the reaction occurred, the solution was poured on ice cold water (a lot Al(OH)3 formed), this was reacted with cc HCl and than NaOH was added to pH 14. A the honey was collected and the sulfate salt was made from it.
Sol. 2 was reacted with MeOH + Al. The reaction was going on 50 Celsius. And the yield was 0. It was poured it on ice, added HCl and NaOH until pH 14 but no honey.
Conclusion: just dilute solutions of this thing should be hydrogenated and not the concentrated solution.
Sol. 3 was poured on water and cc. HCl was added. P2P formed in 80% yield, awesome
References
/1/: https://www.erowid.org/archive/rhodium/chemistry/phenylacetone.html#grignard
/2/: https://www.erowid.org/archive/rhodium/chemistry/amphetamine.html
/3/: http://www.sciencemadness.org/talk/viewthread.php?tid=5519#pid64089
Pictures and everything else data is coming soon, this was just a "test".
). So, the next idea was the MeOH/Al reduction. Simple, easy and cheap, What else is needed?
Chemists generally do some sort of qualitative testing to determine the identity of the product of their experiments, in the old days this was done via derivatives which had known melting points and nowadays it's usually done with various instrumental means of chemical analysis 
You can't just assume this reaction works the way that you think it will, you have to have data to support your hypothesis 
But it may also work? Previously I have used drying tube with calcium chloride but no soda lime and an inert purge, using a method nearly identical to the one above as the halide in question took a fair amount of ether to dissolve it, with a slower dropwise addition of the ether/halide to ether/magnesium with good results. But this procedure may work just as well?