[Help Thread] Petrus Discussion and Help thread

KYLE ALLAIRE DROPS BOMBS! said:
Maybe, but...
I have 11.2 AO5, low 12 AO12, 7.08 single
Yet, I don't think I will ever get sub-10 ao5. Ever. Seriously.
Click to expand...

mDiPalma said:
I also think sub10 avg5 is impossible, for me at least.

I have 12.xx avg5, 13.xx avg12, and a few 9.xx lolsingles.
Click to expand...

Just because sub-10 Petrus hasn't been done doesn't mean it's not possible for some dedicated Petrus users. Many CFOP users who are stuck at ~12 think they can never get sub-10. But it's easier for CFOP users simply because they know many people have achieved sub-10 with the method so easier to believe that they can too with enough dedication. I would like someone to prove that Petrus can be just as fast (sub-10) for the right people.
 
sneaklyfox said:
Just because sub-10 Petrus hasn't been done doesn't mean it's not possible for some dedicated Petrus users. Many CFOP users who are stuck at ~12 think they can never get sub-10. But it's easier for CFOP users simply because they know many people have achieved sub-10 with the method so easier to believe that they can too with enough dedication. I would like someone to prove that Petrus can be just as fast (sub-10) for the right people.
Click to expand...

I do too! We shall wait with bated breath, then.
 
sneaklyfox said:
Just because sub-10 Petrus hasn't been done doesn't mean it's not possible for some dedicated Petrus users. Many CFOP users who are stuck at ~12 think they can never get sub-10. But it's easier for CFOP users simply because they know many people have achieved sub-10 with the method so easier to believe that they can too with enough dedication. I would like someone to prove that Petrus can be just as fast (sub-10) for the right people.
Click to expand...

I've been using petrus for 8+ years.
With my technique, it just requires too many cube rotations to possibly be sub-10.
I have been using CFOP nowadays, with much better results.
 
KYLE ALLAIRE DROPS BOMBS! said:
I've been using petrus for 8+ years.
With my technique, it just requires too many cube rotations to possibly be sub-10.
I have been using CFOP nowadays, with much better results.
Click to expand...

This.

I average ~12 with ZZ, but ~14 with Petrus. The only difference in my solution style between these two is the point at which I do EO (and thus, the quantity of rotations).

One could try EO2x2x2, then expand to EO3x2x2...etc. It would perhaps combine the best of ZZ and Petrus.
 
Interesting... having an explicit EO step, I'd never have expected rotations to be a problem.

Is there no fast rotationless "technique" for the block building?
e.g. Roux 2nd block can be done by limiting the moveset to <M, R, r, U> (and inverses and double turns)
Maybe a similar technique can be applied to petrus?

Or is the problem not in the block? Perhaps the EO step itself?

-----

From my observations of the solve statistics of some top cubers (Faz & Mats for CFOP, Alex for Roux), they can hit >9 tps on "intuitive steps" (F2L, second block, LSE)

Alg steps are at around 7 tps (probably due to the recog pause).

Since afaik, petrus doesn't lose out in movecount and is even viable for FMC...

The question now is what is holding you back from sub-10. Is it the block? Can it not be built at 9-10 tps speeds?

Or is it something else?
 
Last edited:
bundat said:
Interesting... having an explicit EO step, I'd never have expected rotations to be a problem.

Is there no fast rotationless "technique" for the block building?
e.g. Roux 2nd block can be done by limiting the moveset to <M, R, r, U> (and inverses and double turns)
Maybe a similar technique can be applied to petrus?

Or is the problem not in the block? Perhaps the EO step itself?

-----

From my observations of the solve statistics of some top cubers (Faz & Mats for CFOP, Alex for Roux), they can hit >9 tps on "intuitive steps" (F2L, second block, LSE)

Alg steps are at around 7 tps (probably due to the recog pause).

Since afaik, petrus doesn't lose out in movecount and is even viable for FMC...

The question now is what is holding you back from sub-10. Is it the block? Can it not be built at 9-10 tps speeds?

Or is it something else?
Click to expand...

I only ever considered one change to the Petrus method, which dropped my move count to under 50 on average, but I was suffering from the amount of time it took to identify cases. ( Similar to WV ). For every improvement I've made, something else will start slowing me down. Practice doesn't help much with Petrus because even the "intuitive" steps are just seen as "cases" now. Even 2-3 steps in a row can be foreseen and solved quickly as if it were one single case. It's just a matter of rotations and fingertricks slowing people down for the most part.
As for TPS, It's very hard to track pieces during the first few steps. Most optimal blocks are very hard to do, so looking past the first 2x2x2 + corner alone is rare. Sure you can start at 6 TPS, for like... 1 second? Then you have to look for more pieces which were nearly impossible to track during inspection.
For me, I see the 2x2x2 + corner, solve it. Finish 2x2x3 ( very hit and miss on optimization ), followed by embarrassing cube rotations and a bad edge flip where I can hardly track anything because I'm too focused on an optimal bad edge flip phase. Maybe I'm just not a great cuber, but there are some steps that are so silly that I can't ever see myself being faster than 11 seconds average with GOOD scrambles. Too many steps have a 50/50 chance of being completely terrible for finger tricks/rotations. I suppose the main conclusion I've reached after all this time is that Petrus forces you to tunnel vision on steps where lookahead is simply impossible, due to unavoidable rotations. That's one of those problems I've talked about where every solution to that problem will just cause more problems later in the solve.
 
I tried some restricted schemes.

avg12,single with eo then petrus: 17.53,14.59
20.76, 18.08, 14.59, 16.58, 19.02, 18.10, 24.01, 16.43, 15.62, 15.96, 16.04, 18.68
avg12,single with 2x2x2, then zz: 17.84,10.84
21.29, 15.91, 17.67, 17.37, 20.92, 16.94, 17.21, 18.29, 17.78, 15.66, 20.61, 10.84

I honestly think, of these two restrictions, 2x2x2 followed by zz has more potential. You can enter a brainless mode 'earlier' in the solve.

Of course, the best variant would be:
2x2x2 while reducing 8 remaining possible bad edges to 0,2, or 4,
3x2x2 while completing eo,
solve.
 
bundat said:
Interesting... having an explicit EO step, I'd never have expected rotations to be a problem.

Is there no fast rotationless "technique" for the block building?
e.g. Roux 2nd block can be done by limiting the moveset to <M, R, r, U> (and inverses and double turns)
Maybe a similar technique can be applied to petrus?

Or is the problem not in the block? Perhaps the EO step itself?
Click to expand...

Right now I'm developing algs for this (although it's pretty much intuitive it would be good to have some example solutions). One of the algs I have is (r U R' U' r') for flipping UF/UR. I think it's better to have a 5-move alg that's fingertricky than to do a cube rotation, three moves, then cube rotation back. Using the move set <R,r,U,M> eliminates cube rotations at the expense of a couple moves. Also, this enables you to place corners or start blockbuilding on R during EO.

So i think this would definitely solve a couple problems with Petrus, like lookahead (i'm just going to call the step EOF2L), no cube rotations for seamless transitions, and solving a smaller block during inspection. This would be very easy for Roux users b/c it starts with a roux block and incorporates the <R,r,U,M> moveset in a less restrictive step. From there it would be all Petrus.

I'm probably going to switch to this method and call it NCR (No Cube Rotations). I've kind of abandoned the idea of early CP, unless someone figures out a manageable way to do CP during the first block which would be compatible with this method (dont tell me about zzporky, i already know). How it will go is Roux block => EO + D edges => right block + OLL => PLL. I'm thinking of throwing in some EJF2L or WV in there, but not sure if it will be too many algs.
-----------
This morning I finished the algs for 8 Edge Orientation, there are 128 unique cases, and 47 algorithms. The avg. occurrence for all the algs. is 7.22 STM (this factors in how often each case occurs). These algs will orient the 8 remaining edges after the first Roux block and solving the BD edge. After this step you will place the FD edge and continue the solve like regular Petrus. The EOF2L step should be intuitive, however I made the algs for some example solutions. They are not all optimized for <R,r,U,M>, but it's a pretty good start. Anyone who is interested can PM me for the algs.
 
Last edited:
Anybody still think Petrus is viable? It's all I've ever known. Seems fine for novices.
 
Petrus is usually acclaimed as a more efficient method for speed in comparison to CFOP. However, Lars himself said, "Going for speed, I use 60 moves on average. Going for few moves, I average 45." CFOP's movecounts are around 57, I think. So doesn't that make Petrus about as efficient as CFOP? Because without move efficiency, Petrus has few advantages.
 
Does anybody have a good way to learn full COLL or another way to improve Petrus method? I can get a good 27 second average with 7 COLL and the 4 EPLL algorithms and want to become faster.
 
HenryVIII said:
Does anybody have a good way to learn full COLL or another way to improve Petrus method?
Click to expand...

hi i'm averaging around 40 seconds (~55 htm) with Petrus now, but that's mostly because I don't own a physical cube anymore and I don't know any algs. I just sort of solve the F2L and the top edges intuitively and then solve the last four corners with 1 or 2 commutators. it seems really efficient for me, and maybe it can help you!!
 
Need help getting faster with Petrus?

Alright, I've recently gotten super-involved with the Petrus Method (for all y'all haters: PETRUS IS UNDERRATED). I average around 17-19 seconds (with an 11-second Petrus PB), which is just a few seconds slower than my CFOP solves. Note: I'm not color neutral (I only solve on white), but I'm probably going to switch to white-yellow sometime.

Solve breakdown:
2x2x2: 1-2 seconds
2x2x3: 3-6 seconds, including recognition
EO: 1-2 seconds, including recognition
F2L: 2-6 seconds, including recognition and LS stuff
COLL/Sune OLL: 1-1.5 seconds, including recognition
PLL/EPLL: 1-1.5 seconds, including recognition

Anyway, my main problem is the 2x2x3. A lot of my pauses come up during this stage, mainly because it's so hard to ignore the non-2x2x3 F2L pair (the pair that cannot be joined to the existing 2x2x2). Any tips from the pros? I might upload a video of an ao5.
 
You must log in or register to reply here.

Similar threads

Nukoca
Petrus Method Discussion Forum, by Lars Himself
Replies
3
Views
1,508
Johannes91
J
PetrusQuber
Petrus Colour Neutrality - Help Needed
Replies
3
Views
463
PetrusQuber
PetrusQuber
M
Gemini Flash AI continues to come up with new tutorial ideas
Replies
1
Views
462
0rcoraCuber21
0rcoraCuber21
acrystal
I need help with my method (i forgot to mention, im calling it "Orcora Petrus")
2
Replies
26
Views
2,363
abunickabhi
abunickabhi
acrystal
  • Poll Poll
Is petrus the best method for FMC?
Replies
11
Views
1,107
abunickabhi
abunickabhi
Back
Top