A potentially faster OLL 23 algorithm

[Aergyl]

Recently, I've been experimenting with the 2-twist

(R' D R D' R' D R) U' (R' D' R D R' D' R)​

for OLL 23. It's slower than the 3-cycle

R2 D R' U2 R D' R' U2 R'​

but the PLL transition is faster when done in the following way:

1. Perform R' D R D' R' D R while looking at UFL, LU and LUB (which is easy since they don't move). They will move and become the front 3 colors used in PLL recognition later, so we can use that information to make it faster. Also, take a mental note on the exact color of UFL.
2. Perform U'
3. Perform R' D' R D R' D' R and 2-side PLL recognition simultaneously. This can be done since the stickers used for the recognition, except for FUR, don't move at all. As for FUR, just imagine the color we took a mental note of being there.
4. Transition into PLL without pausing and cancel moves if possible.

It takes some practice to get good at the recognition, but it's definitely possible. Also, I made a video about it if you want to see how it looks:

Now, the question is which option yields the fastest average LL execution time. Since the average PLL execution time is essentially the same for both cases, we can just compare the average time until PLL execution starts instead. For the 2-twist, that is simply equal to to its average execution time and for the 3-cycle, it's equal to its average execution time, plus the average PLL recognition time.

The average execution times are pretty easy to measure. However, to make the comparison fair, it's important to start from homegrip and do both executions with the same amount of risk. Also, the average PLL recognition time can be measured by setting up OLL 23 (with random PLL), and framecount how long you pause between OLL and PLL.

If you do that, you should be able to compare them and tell whether the 2-twist has potential to be faster for you or not. For reference, I tried it myself, and the 2-twist surprisingly won by about 0.4s. I also tried to do three ao50 of LL with both options, and I got 3.35, 3.09 and 3.19 with the 3-cycle, and 2.97, 2.82 and 2.70 with the 2-twist, which clearly suggests that the 2-twist is faster for me.

I haven't heard of anyone talking about this before, so I'm curious about how well it works for other people. Please let me know if you try it out

 

[Spencer131]

That's an interesting experiment! It seems like the 2-twist was about 0.4 seconds faster for you, which can be significant for a fast solver

CFOP is not my main method, so I don't know for certain, but wouldn't it still be relatively easy to predict PLL when the OLL is R2 D R' U2 R D' R' U2 R'? Since it's a 3 cycle, most of the pieces are unaffected. I'm just curious how much you have practiced predicting PLL from that alg

 

[Aergyl]

That's an interesting experiment! It seems like the 2-twist was about 0.4 seconds faster for you, which can be significant for a fast solver

CFOP is not my main method, so I don't know for certain, but wouldn't it still be relatively easy to predict PLL when the OLL is R2 D R' U2 R D' R' U2 R'? Since it's a 3 cycle, most of the pieces are unaffected. I'm just curious how much you have practiced predicting PLL from that alg

Thank you! Yeah, my framecounted average PLL recognition time was 0.55s, so I'm sure there are fast solvers with faster recognition, and the time difference would be smaller for them.

It's definitely possible to predict parts of the PLL with the 3-cycle (I do it sometimes when I see big blocks, but I haven't really practiced it) and some people use that to make PLL recognition faster. I haven't heard of anyone who can actually predict the entire PLL though, without pausing beforehand or slowing down. That being said, even if you could actually do that, there's still a potential hard regrip after the OLL execution that you have to account for.

 

[Spencer131]

In Roux, getting that case for CMLL is amazing. You can perfectly predict LSE because the edges are unaffected lol

 

[Thom S.]

I haven't heard of anyone who can actually predict the entire PLL though,

Basically every ZBLL user can.

 

[Aergyl]

Basically every ZBLL user can.

Did you read the entire sentence?

 

[Thom S.]

Did you read the entire sentence?

Yes. My point is still correct.

 

[Aergyl]

Yes. My point is still correct.

Do you mean consistently (close to 100% of the time) or just sometimes? I haven't heard of any ZBLL user that never pauses for ZB recognition.

 

[Thom S.]

The main point is that OLL recognition with PLL recognition combined and ZBLL recognition are pretty much the same lenght of pause.

 

[Aergyl]

The main point is that OLL recognition with PLL recognition combined and ZBLL recognition are pretty much the same lenght of pause.

That makes sense. It wouldn't even surprise me if ZB recognition can be faster in theory. It's still going to be slower than OLL recognition only though, which means you have to "pause beforehand" in order to predict PLL that way. So if you do that, you have to account for that extra recognition time when comparing it to the 2-twist, since the only pause in that case is the OLL recognition.

However, that doesn't neccessarily prove that it's impossible to predict the entire PLL while executing the 3-cycle, when starting the execution as soon as you recognize the OLL. As far as I know, many fast solvers can do it sometimes, but I haven't heard of anyone who can do it consistently, which is what I meant to say. Correct me if I'm wrong though!

 

[Aergyl]

The main point is that OLL recognition with PLL recognition combined and ZBLL recognition are pretty much the same lenght of pause.

I'm not sure, but I might've misunderstood what you meant by combined, lol. If you meant doing them at the same time, please ignore the second sentence in my reply...

 

[turtwig]

From what I've heard, many top cubers can predict some or all of PLL during OLL. ROLL and JOLL are useful for this purpose. I've learned a bit of it and it's not too hard to internalize. With mastery, you can reduce the number of possible PLLs by on average 6x (3 groups of corner permutation and 2 groups of edge permutation). I'm sure many top cubers can even recognize beyond that based on experience and "feel." Even I know PLLs from certain LL cases, just from encountering them many times.
One example I can think of is from one of Tymon's WR average (4.86) solves, where he does EO/ZBLL
r U R' U' M U R U R'
R U R D R' U' R D' R2'
cancelling the R' R between the algs (meaning he knew ZBLL before or during OLL execution).
The main problem I see is that the 2 twist is not as fast as 3 cycle (or some other ZBLLs that I think are actually faster than 3 cycle), but this trick could be useful for some people if the 2 twist can be finger-tricked well and PLL transition is done well. That being said, I think to be as fast as possible, ZBLL will trump any PLL prediction scheme (the OLLs and COLLs for this case are pretty trash so the ZBLL set is pretty worth imo).

 

[Aergyl]

From what I've heard, many top cubers can predict some or all of PLL during OLL. ROLL and JOLL are useful for this purpose. I've learned a bit of it and it's not too hard to internalize. With mastery, you can reduce the number of possible PLLs by on average 6x (3 groups of corner permutation and 2 groups of edge permutation). I'm sure many top cubers can even recognize beyond that based on experience and "feel." Even I know PLLs from certain LL cases, just from encountering them many times.
One example I can think of is from one of Tymon's WR average (4.86) solves, where he does EO/ZBLL
r U R' U' M U R U R'
R U R D R' U' R D' R2'
cancelling the R' R between the algs (meaning he knew ZBLL before or during OLL execution).
The main problem I see is that the 2 twist is not as fast as 3 cycle (or some other ZBLLs that I think are actually faster than 3 cycle), but this trick could be useful for some people if the 2 twist can be finger-tricked well and PLL transition is done well. That being said, I think to be as fast as possible, ZBLL will trump any PLL prediction scheme (the OLLs and COLLs for this case are pretty trash so the ZBLL set is pretty worth imo).

Yeah, I don't doubt that many top cubers can reduce the pause between OLL and PLL with the 3-cycle. And I don't doubt that they can predict the entire PLL sometimes. I don't know how often though, so it's hard for me to predict how big the pause is going to be on average. Also, the 2-twist is only about about 0.1s slower for me, which means the pause has to be very small in order to be worth it, for me at least. Therefore, I think it would be very interesting to see what those numbers are for other people!

By the way, you mentioned some ZBLL's being faster than the 3-cycle, and I'm very curious about that. Do you have any examples?

Also, I definitely agree that ZBLL is faster with enough practice.

 

[Aergyl]

For anyone interested, I framecounted the 3-cycle (including the regrip) performed by Ruihang Xu in the following video (at 1:30):

Then, I compared it with the 2-twist (excluding the last U move) performed by Tommy Cherry in the following video (at 0:31):

If I did it correctly, the 3-cycle was (17/30)s and the 2-twist was (19/30)s. In other words, the difference is only (1/15)s. Of course, this isn't a perfect comparison and doesn't neccessarily prove anything, but I think there's a possibility that the difference in execution time could be smaller than some people think.

Edit: On closer inspection, it was actually the 3-cycle, but with two wide moves (R2 D' r U2 r' D R U2 R), which makes the comparison slightly unfair. Also, I get the feeling that something is wrong with my framecounting (I tried doing it with my phone but it feels like some frames are skipped). I'll recount it when I can use a computer later!

 

[OreKehStrah]

For anyone interested, I framecounted the 3-cycle (including the regrip) performed by Ruihang Xu in the following video (at 1:30):

Then, I compared it with the 2-twist (excluding the last U move) performed by Tommy Cherry in the following video (at 0:31):

If I did it correctly, the 3-cycle was (17/30)s and the 2-twist was (19/30)s. In other words, the difference is only (1/15)s. Of course, this isn't a perfect comparison and doesn't neccessarily prove anything, but I think there's a possibility that the difference in execution time could be smaller than some people think.

Edit: On closer inspection, it was actually the 3-cycle, but with two wide moves (R2 D' r U2 r' D R U2 R), which makes the comparison slightly unfair. Also, I get the feeling that something is wrong with my framecounting (I tried doing it with my phone but it feels like some frames are skipped). I'll recount it when I can use a computer later!

What are the time stamps? I’ll frame count it too.

 

[Aergyl]

What are the time stamps? I’ll frame count it too.

They're inside the parenthesis just before the colon symbols...

 

[qwr]

A while back I semi-seriously proposed counting double moves as 1.5 turns. in 1.5TM the standard OLL is 10.5 moves vs 15 of the 2-twist. Triggers like sexy are definitely faster than their love count but I'm not sure about RD triggers.

 

[OreKehStrah]

They're inside the parenthesis just before the colon symbols...

Lmao I completely missed those somehow. I counted 19 and 22 frames for the 2 vids respectively

 

[Aergyl]

Lmao I completely missed those somehow. I counted 19 and 22 frames for the 2 vids respectively

Interesting! I'll try it myself later as well. Also, I forgot to mention that it might not be completely fair to count from homegrip in this case, since the regrip can sometimes be a soft regrip, depending on the pre-AUF. I'm not sure how to make the comparison as fair as possible, but maybe counting half the regrip is best?

 

[turtwig]

By the way, you mentioned some ZBLL's being faster than the 3-cycle, and I'm very curious about that. Do you have any examples?

To be fair, I've never timed them or anything, some algs just felt faster to me. The ones I was thinking of in particular are
R' D' R U' R' D R2 U2 R' U R U R'
R' U' R U' R' U2 R2 U R' U R U2 R'
I feel they're more finger-trick friendly, in particular the second one which is just 2 sunes. But they are 4 moves longer, so maybe I just thought they were faster because I could get higher TPS on them but overall they still took longer to execute. Anecdotally, I've only seen top cubers do the standard 3 cycle when they don't know ZBLL for this case.

PS: I didn't know BLDers did the RUD alg for 2 twist. For the 2 twist ZBLL case, I've always done antisune U2 back-antisune. I wonder if the RUD alg is just faster or if it's a difference of hand position, etc., between 3-style and CFOP.