• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

How do You Think CFCE Compares to CFOP?


  • Total voters
    40
Status
Not open for further replies.

ProStar

Member
Joined
Oct 27, 2019
Messages
6,253
Location
An uncolonized sector of the planet Mars
WCA
2020MAHO01
SS Competition Results
======
Update
======

Due to some smarter and more experienced cubers giving their advice, I reconsidered using CLL instead of COLL. I did some more research, and I've decided that my previous argument...

After research, I decided I would COLL as my alg set to solve corners. My reasoning behind this is that if a cross is solved after F2L, CMLL or CLL will mess up the cross, while COLL will force a U, H, or Z perm; an easier case than most ELLs.

...has a flaw. As Fillipe Teixeira said:

IMHO if you want to get advantage that the cross is solved you are losing the advantage to use better algorithms when the other case is true. (whitch will occur more often)

After more research, I agree with him. I've decided to learn CLL instead of COLL. If anyone can recommend good algs/vids I'd appreciate it, as I can't find a good list.
 
X

Xtreme Cuber

Guest
That's right, you can't; you just replace them with something else that's hopefully better. OLLCP, 1LLL, fruruf into T ZBLLs, etc. You can do the exact same thing with CFCE. It's not an argument for or against CFCE, unless you're comparing CFOP-with-so-many-extensions-that-it's-not-even-really-"cross-F2L-OLL-PLL"-anymore with base CFCE, which is obviously an unfair comparison.
What makes that an unfair comparison? I believe the point was that CFOP is better at its current development, compared to CFCE's development. I think it's pretty obvious that more developed methods are better than less-developed ones. This isn't to say that CFCE shouldn't be pursued or developed further, nor that you can't get fast with it, just that at the present time and with present available resources, CFOP is faster.

In addition, last layer is a bit faster with CFOP than CFCE, especially because most ELL cases are essentially LSE cases with two edges solved, which means they use a lot of M moves, which are usually slower than RUFD moves, which is what most PLL cases are made up of. In addition, recognition is definitely a bit slower in CFCE. Both methods have been around since at least 1982. One all but vanished and one became the number one method for top cubers across the world. There might just be a reason for that. ;)

Again, I'm not trying to discourage anyone from learning CFCE or furthering its potential. I just think it's worth mentioning that when judging the merits of a method, one should take into consideration all aspects including maximum TPS and number of resources. I agree that CFCE isn't a lot worse than CFOP, but I personally don't see any good reasons to choose it over other faster and easier methods except for proof of concept.
 
Joined
Aug 12, 2013
Messages
5,086
Location
Brazil
SS Competition Results
YouTube
Visit Channel
As Fillipe Teixeira said: "..."
...If anyone can recommend good algs/vids I'd appreciate it, as I can't find a good list.
Thanks for the mention!

Recognition guides

I suggest using the videos to learn recognition and drilling the CLL algs that fit you the best.

Notice that in the end, you have to develop your own technique for recognition and execution... use what works the best for you!
 
Last edited:

ProStar

Member
Joined
Oct 27, 2019
Messages
6,253
Location
An uncolonized sector of the planet Mars
WCA
2020MAHO01
SS Competition Results
Thanks for the mention!

Recognition guides

I suggest using the videos to learn recognition and drilling the CLL algs that fit you the best.

Notice that in the end, you have to develop your own technique for recognition and execution... use what works the best for you!

Thanks!
 

xyzzy

Member
Joined
Dec 24, 2015
Messages
2,877
If anyone can recommend good algs/vids I'd appreciate it, as I can't find a good list.
I already linked Justin's CLL/ELL list in the first page…

especially because most ELL cases are essentially LSE cases with two edges solved, which means they use a lot of M moves, which are usually slower than RUFD moves, which is what most PLL cases are made up of.
Is this really true? Are you sure something like (M' U M U)3 isn't faster than almost every PLL case? What about ELL algs that are RUF, like R' U2 R U' R' U' R' F R2 U R' U' R' F' R2? What about the 3-cycle comms like [R' F R, S] or [R U R' U', M']?

There are good ELL cases and there are bad ELL cases… which is pretty much the same situation as PLL. If you're looking at old alg lists for ELL (like all the garbage on AlgDb), then yes, a lot of the cases have bad MU algs. We know better now.

What makes that an unfair comparison? I believe the point was that CFOP is better at its current development, compared to CFCE's development. I think it's pretty obvious that more developed methods are better than less-developed ones.
I agree that more development would generally mean that more of the method's potential has been exposed, and hence it would likely be faster. However, my point was that CFOP and CFCE are so similar that it's not necessary to do much CFCE-specific development—most of what we already know for CFOP can directly be used with CFCE, without any modifications. Instead of comparing CFOP with all its extensions to baseline CFCE, we should be comparing CFOP with all its extensions to CFCE plus the same extensions that can be used with CFCE; CFOP stops being an obvious win over CFCE if you do this.

It's not like we're comparing CFOP against some obscure method like Belt; in that case, one might legitimately make the claim that even if Belt had the same potential as CFOP (almost certainly not, but just for the sake of argument), it has had a lot less development and it's so dissimilar to any other popular method that one shouldn't expect Belt to be faster than CFOP at this point in time.

(Good CLL and ELL algs have already been generated. Sure, there might be even better algs that we haven't found yet, but these improvements are generally marginal—if the algs are hard to find, they can't be much better than what we already have. We have tools like AlgExplorer that make finding good algs much easier than it used to be in 1980-2015.)
 
X

Xtreme Cuber

Guest
Is this really true? Are you sure something like (M' U M U)3 isn't faster than almost every PLL case? What about ELL algs that are RUF, like R' U2 R U' R' U' R' F R2 U R' U' R' F' R2? What about the 3-cycle comms like [R' F R, S] or [R U R' U', M']?
I don't know about you, but I can perform Ua, Ub, Aa, Ab, Ra, and T perms faster than (M' U M U)3, none of which use M moves. I didn't want to come across as saying every ELL case is horrible and every PLL case is amazing. However, the fact that PLL cases have virtually no M moves means that, given the same length of algorithms, PLL cases are usually faster. (And yes, I have looked at the ELLs you linked in your first post.)

I agree that more development would generally mean that more of the method's potential has been exposed, and hence it would likely be faster. However, my point was that CFOP and CFCE are so similar that it's not necessary to do much CFCE-specific development—most of what we already know for CFOP can directly be used with CFCE, without any modifications. Instead of comparing CFOP with all its extensions to baseline CFCE, we should be comparing CFOP with all its extensions to CFCE plus the same extensions that can be used with CFCE; CFOP stops being an obvious win over CFCE if you do this.
Okay, that makes more sense. Out of curiosity, what CFOP subsets were you thinking of that could be applied to CFCE? However, even if we're comparing baseline CFOP and CFCE, CFOP is still faster because of better recognition and max TPS and I don't think the 0.28 move difference compensates for that. A while back I was actually looking to learn CFCE because I thought it would be just as fast, but those were the two things that made me switch back to CFOP.

Again, I don't have a problem with anyone believing the two methods are equal and I think it's great that people are branching out to try new (or old?) methods, but in my experience, I haven't found any compelling reason to choose CFCE over CFOP. :)
 

ProStar

Member
Joined
Oct 27, 2019
Messages
6,253
Location
An uncolonized sector of the planet Mars
WCA
2020MAHO01
SS Competition Results
(I made a new account, so I had to replace the old one in order to edit the main post)

Yes, I am switching my 3x3 method. I am switching to the method known as Cross, F2L, Corners, Edges, or CFCE for short.


Overview of CFCE

In CFCE, you solve the first two layers using Cross + F2L, the same as CFOP. The difference comes in the last layer. Instead of doing OLL and PLL, you orient and permute the corners with one alg(two for beginners), and then orient and permute the edges with one alg(two for beginners). Solving the corners in 1 look takes 42 algorithms. This can be accomplished by knowing one of three algorithm sets: CMLL, CLL(for 3x3), or COLL. After the corners are solved, you use one of the 29 algs in the ELL algorithm set, thus completing the cube.

Why Did I Make This Thread?

I wanted to document my journey through switching methods, and decided this would be a good way to do it. It will raise awareness of this obscure method and allow those reading this to give advice. I will structure this thread similar to the thread by @PetrusQuber, “My Journey to Sub 8 Petrus”. I will post how I’m practicing, my improvements, what I’m learning, and maybe some more info on CFCE.

Why Am I Switching?

Although it originated in 1981, CFCE isn’t a popular method, which is appealing to me. It is, in my opinion, just as good as CFOP. I also won’t lose the progress I’ve made in Cross + F2L, as CFCE shares those first two steps with CFOP.

There is another reason I’m switching, and it takes the form of a story. While messing with the cube at around midnight about a week ago, I decided to try to create a new method, or at least modify an existing one. I know CFOP and Roux, the basics of Petrus, as well as the basics of ZZ (like really basic, I can’t solve with it). Over the next couple of days, I came up with a few methods, including RouxFOP, Petroux, and an outlandishly unrealistic corners-first method (I’m lousy at names, I know). Eventually, I came up with an idea I liked. I first solved FB and SB, like Roux, then used M moves to finish F2L, then solved the LL corners, then solved the LL edges. (sound familiar?) I decided that this might have merit as a method, and began to develop it seriously.

The prototype name I decided on was “FBCL”, standing for “First 2 Blocks(F2B), Bottom 2 Edges(B2E), Corners of the Last Layer(CLL), Last 4 Edges(L4E). Later, I decided to rename the last step “Edges of the Last Layer(ELL)”. I went to the SS wiki and searched ELL, just in case it was the name for something already, and this page came up. I read through it, and something I saw caught my eye: “...normally after performing CLL to solve the last layer corners…”

Soo… yeah. The method already existed. I did some research, and found out about CFCE. After doing more research, I found CFCE appealing to me, and after trying it out, I decided to switch, hence this thread.

Where Am I at Currently?

After research, I decided I would COLL as my alg set to solve corners. My reasoning behind this is that if a cross is solved after F2L, CMLL or CLL will mess up the cross, while COLL will force a U, H, or Z perm; an easier case than most ELLs.

I've decided to learn CLL, as the time I'd save with COLL whenever a cross is already made is less than the time I'll save with CLL's better algs whenever there isn't a cross(which is more often)

I currently use a 4lll I made up using a few algs from the two algorithm sets I need to learn(CLL and ELL). Without CFOP, I averaged about 25-28 seconds, and after a couple months of practice I'm averaging about 22 seconds.

What I’m Learning + Times

What I’m Learning


CLL: 7/42
O: 2/2
H: 4/4
Pi: 1/6
U: 0/6
T: 0/6
Sune: 0/6
Anti-Sune: 0/6
L: 0/6

ELL: 4/29
Won’t start learning until after CLL(Only know the EPLLs)

Times

Best CFCE Single: 13.88
Best CFCE Mo3: 16.00
Best CFCE Ao5: 18.46
Best CFCE Ao12: 19.37
Best CFCE Ao100: 22.52

Current Session Average(115+ Solves): 24.76

============================================================================================================

Last Updated: January 11, 2020
 
Last edited:
Joined
Dec 18, 2019
Messages
144
Location
On the Rocks that make up Saturns Ring
Dude, I mean CFCE is good but CFOP last layer recognition is faster.
I'm not saying that CFCE is bad, I'm saying that CFOP is faster.
If CFCE was faster, all the pro cubers that use CFOP would be using it.
Also, there are almost as many algs (only a few less)

P.S. i once tried but just found last layer recognition for CFCE too slow compared to CFOP

Anyway good luck!!!!
 

PetrusQuber

Member
Joined
Jun 27, 2019
Messages
3,460
Location
my house, cubing.
YouTube
Visit Channel
And so the CFOP/CFCE argument begins. He is trying to use CFCE. He does not need extra discouragement from what he is trying to do. He just wants to post times, what he is doing, etc here. I’m pretty sure he doesn’t care that much about if this method is better or worse, etc.
Sorry if I misinterpreted your meanings in any way, I just thought it would be good to explain the point of this thread. (@ProStar, feel free to correct this)

Good luck anyway!
 

RedstoneTim

Member
Joined
Jul 21, 2019
Messages
227
Location
Germany
YouTube
Visit Channel
Since you stated that you might learn some OLLCPs, I'd also advise you to have a look on a method called "Zipper".
It's basically CFCE with an extra edge. You solve like usually until last slot and then insert one corner (you can usually insert the corner while solving another pair beforehand), then you either do OLLCP and L5EP (L5EP can be completely <RU>-gen and doesn't have a lot of cases) or CFRLL (just 3x3 CLL but with better algs since you don't have to preserve the FR edge, though you could also just use normal CLL algs if you wanted to) and then L5E (ELL is a subset of L5E, so you should already be familiar with some of the cases, the rest could be done semi-intuitively).
In my opinion, Zipper is better than CFOP/CFCE since you still end up with a 2 look ending and have to solve one pair less, but decide for yourself.
All in all I'm happy people are starting to use other methods than standard CFOP, so I wish you a lot of fun learning CFCE!
 

ProStar

Member
Joined
Oct 27, 2019
Messages
6,253
Location
An uncolonized sector of the planet Mars
WCA
2020MAHO01
SS Competition Results
Dude, I mean CFCE is good but CFOP last layer recognition is faster.
I'm not saying that CFCE is bad, I'm saying that CFOP is faster.

Recognition being slower has a lot to do with how much you'd practiced OLL/PLL vs. CLL/ELL. Recognizing CLL is the same as COLL.

If CFCE was faster, all the pro cubers that use CFOP would be using it.

You could say that about any method, like Roux. CFOP has definitely been developed more, but that doesn't mean other methods are slower.

And so the CFOP/CFCE argument begins. He is trying to use CFCE. He does not need extra discouragement from what he is trying to do. He just wants to post times, what he is doing, etc here. I’m pretty sure he doesn’t care that much about if this method is better or worse, etc.
Sorry if I misinterpreted your meanings in any way, I just thought it would be good to explain the point of this thread. (@ProStar, feel free to correct this)

Good luck anyway!

I solidly believe that CFCE can be just as fast as CFOP. Even if it has slightly worse LL recognition(the main argument against it), the difference is marginal; not large enough to make a difference for someone with my average, or beyond.
 

efattah

Member
Joined
Feb 14, 2016
Messages
711
CFCE might be pretty good, but I think a missing item here is skip probabilities. With CFOP you get your PLL skip 1 in 72.
With CFCE, you could get a CLL skip (1 in 152), or an ELL skip (for which I don't know the probability--perhaps someone can chime in).
I can try to guess ELL skip chance:
Orientations: 8, Permutations: 10, so I think 80 combinations of which 1 is solved, so ELL skip could be 1 in 80
 

Etotheipi

Member
Joined
Mar 21, 2019
Messages
860
Location
somewhere on the complex plane.
CFCE might be pretty good, but I think a missing item here is skip probabilities. With CFOP you get your PLL skip 1 in 72.
With CFCE, you could get a CLL skip (1 in 152), or an ELL skip (for which I don't know the probability--perhaps someone can chime in).
I can try to guess ELL skip chance:
Orientations: 8, Permutations: 10, so I think 80 combinations of which 1 is solved, so ELL skip could be 1 in 80
Skips don't effect averages much, which are what most people care about, so I don't really see why this is at all a noticeable flaw in the method, at most it could raise an average by a couple hundredths of a second.
 

xyzzy

Member
Joined
Dec 24, 2015
Messages
2,877
CFCE might be pretty good, but I think a missing item here is skip probabilities. With CFOP you get your PLL skip 1 in 72.
With CFCE, you could get a CLL skip (1 in 152), or an ELL skip (for which I don't know the probability--perhaps someone can chime in).
I can try to guess ELL skip chance:
Orientations: 8, Permutations: 10, so I think 80 combinations of which 1 is solved, so ELL skip could be 1 in 80
1/162 for CLL skip (I think you typo'd this) and 1/96 for ELL skip.

But I don't think skip rates should be a relevant concern anyway—they really favour methods with a step that doesn't do much (e.g. with EO-ZBLL there's a 1/8 chance of EO skip), and it's not like a LL substep skip completely determines whether a solve is good or bad since a 3×3×3 solve doesn't just consist of solving the last layer. (There's the converse relation, where it's true that extremely good solves (as in top 0.05% of a person's times) have skips disproportionately often (probably well over 20%), but this isn't the important direction. Bayes's theorem!)
 

efattah

Member
Joined
Feb 14, 2016
Messages
711
What I am trying to get at is to examine the chance of good and bad random events. For example, CFOP solves dread N-Perms, but there are no N-Perms in CFCE, even the worst CLL and worst ELL aren't too bad, even if recognition is debatable. It is also true that ELL algorithms are very fast to execute.
 
Status
Not open for further replies.
Top