Foreright’s road to sub-20 and learning CMLL

[Foreright]

I’m returning to cubing after a year and a bit off. Previously I’d learnt full CMLL and was just sub-20. Since coming back I’m averaging around 23.5 and have pretty much forgotten all of my CMLL algs aside from the usual 2L ones. This thread is to track progress in getting back below 20 and potentially further and also relearning those CMLL algs.

Unfortunately my main smart cube (Moyu AI - the old, good one!) appears to have stopped registering certain turns so I’m back to the Gan iCarry for tracked / times solves and my mains - Gan X / Tornado V3 for untuned / practice until next week when a fresh (old) Moyu AI arrives!

I started a new Cubeast session on 30th Dec 2023 and my average for to start over 75 solves was 23.55.

As of 2nd Jan 2023:

Start time: 23.55 @ 65.36 STM

CMLLs learnt:

O: all (2/2)
H: rows, columns (2/4)
Pi: right bar (1/6)
T: left bar (1/6)
U: back row (1/6)
L: best (1/6)
S: sune (1/6)
As: antisune (1/6)
Total: 10/42

 

[Foreright]

Just to reply quickly for the other two days I've actually done solves. I did 50 solves on 2nd Jan for an average of 22.12 and 30 yesterday for an average of 21.90 so I'm heading in the right direction already.

As it stands right now my last Ao100 is 21.82 @ 60.4 STM.

I've quickly identified that my major weakness is my second block and specifically I'm using WAY too many moves to complete it.

I've improved yesterday from 19.x moves for SB to 18.2 but that is still too high. I'm aiming (long term) for 14 or even less and am doing a fair few untimed / casual solves concetrating on efficiency to try to achieve that. My turn count IS lower when slow solving which is a positive. Taking that into speed solves is going to be challenging though. This Ao100 is actually fairly atypical in the SB time is a bit lower than usual but even so at 37% of the total solve time it's too much. I've been able to full plan FB a decent (ie. 80+) percentage of the time and have been incorporating tracking and even better, placing the DR edge during my first block so that second block simply consists of placing the two pairs but I'm finding that I'm a little slow in finding them some of the time.

PB so far for this session if 13.53 albeit that was a scramble that I saw posted on the Roux discord as an easy one. Fastest solve with a scramble generated by Cubeast so far is 15.99 which was a 39 STM @ 2.44 TPS solve (https://cbst.link/s/1tktAxjv1LAusE7auH3f2z). That's still a way off my all time PB which was 12.03 back in October 2022 (https://cbst.link/s/bFALW4GbcfLWxKb1rKVmmE).

I spent some of yesterday relearning some CMLLs - specifically the rest of the H and Pi sets. Those have been pretty easy to learn and a couple of hours drilling those in the OnionHoney CMLL trainer has resulted in those being successfully memorized and quite fast to recognize and execute.

5-Jan-2024:

Total solves so far: 200
Ao200: 22.51
Ao100: 21.82

Best Ao100: 21.82
Best Ao12: 20.16
Best Single: 13.53

CMLLs learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar (1/6)
U: back row (1/6)
L: best (1/6)
S: sune (1/6)
As: antisune (1/6)
Total: 17/42

Next sets to learn will be T as it seems to be coming up the most from my own stats (17.5%) so far. Might combine it with learning U as there's some crossover from what I remember but we'll see how adventurous I'm feeling tomorrow

 

[Foreright]

Not general able to do an awful lot of cubing at weekends due to me (theoretically!) being a responsible adult with (definitely!) 3 kids. Managed to sneak in a bit of time this morning though and re-learnt two additional T algs, those being right bar and front row. Will probably be able to do a bit of drilling with the CMLL trainer this evening hopefully but those two seem to be pretty natural / easy to learn so I think they're probably "in there" for the time being.

As an aside, I'm finding it pretty easy to learn algs but the difficulty for me is associating the alg with the case and it's orientation. I seem to remember finding that the most challenging thing last time around.

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns (4/6)
U: back row (1/6)
L: best (1/6)
S: sune (1/6)
As: antisune (1/6)
Total: 20/42

Attaching my CMLL algs for my easy online reference

Edit: Just realized that I actually do know an extra T case already as it's a really small modification from H row / Pi Columns - that is T columns which is U2 r U’ r2 D’ r U2 r’ D r2 U r’

 

[Foreright]

Spent an hour or so this afternoon drilling the algs I've learnt so far and they seem to be pretty well ingrained so far. Managed to get 20 solves in before the kids came home from a swimming lesson

Generated By excelAnalysis.xlsm on 2024-01-08
avg of 20: 20.90
mean time: 21.03
mean stm: 55.90
mean tps: 2.67

    Time    STM    TPS     Scramble
-------------------------------------------------------------------------------
1.    18.81    50    2.66    U2 R F2 U2 L2 D F' R2 U L F R2 D F2 R2 L2 U L2 U2 L2 U R2
2.    19.39    66    3.40    R' U' R' U D' L' F D B L2 F2 U2 R2 D' B2 R2 D' L2 B2 D' B2 R2
3.    19.51    52    2.67    U' L U2 F B' L2 D R' B L' U2 R2 D B2 D2 R2 U' B2 D' F2 B2 L2
4.    20.17    58    2.88    B2 U F2 D' F R F' B' L D' R2 F2 B2 U B2 R2 U' R2 D' L2 D2
5.    19.39    56    2.89    U' F2 U' L D' R2 D2 R F' L' B2 U2 B2 R2 D' L2 U F2 U' F2 L2 D'
6.    23.78    59    2.48    U' R' D F2 L' U' F D' B L U' L2 U' B2 R2 U' B2 R2 U L2 B2
7.    19.27    50    2.59    F' B' D2 L U' R' F' L D' R' B' U2 R2 U' L2 D' R2 F2 B2 D R2 D'
8.    22.43    72    3.21    U' B2 U' F R D' R2 L2 F' D2 L' U F2 R2 L2 U2 D' F2 L2 D L2 U
9.    23.13    46    1.99    F' B' D2 L U' R' F' L D' R' B' U2 R2 U' L2 D' R2 F2 B2 D R2 D'
10.   20.35    55    2.70    R D' R U L D R F' L B D2 F2 U D' B2 L2 U B2 R2 U F2 L2
11.   23.68    56    2.37    U2 L D' F R' B R2 D F R F R2 L2 U D2 B2 D F2 L2 F2 D R2
12.   (18.38)   46   2.50    D2 F2 R' F L U2 D' R D' F U F2 D' B2 U F2 L2 U2 L2 D' R2 B2
13.   22.10    59    2.67    F R L F U L2 F' D2 F' L' F2 R2 D R2 F2 U R2 U2 L2 B2 U2 B2
14.   19.30    47    2.44    R' U' R B U R L' D' F R' U F2 B2 U' R2 B2 L2 U2 R2 U D
15.   20.86    47    2.25    L' F D B' R' L B2 D F' U B2 U R2 D' F2 L2 B2 L2 B2 U R2 D'
16.   21.51    65    3.02    F' R U L2 F2 B U B' L F' D' B2 D2 B2 R2 D' L2 B2 R2 D2 B2 U2
17.   24.10    72    2.99    R' U L B' U R2 U' R2 F D2 B2 U' F2 D L2 F2 B2 D R2 F2 D2 B2
18.   19.85    58    2.92    U2 B U' F2 U R' B R2 D' F' R2 D R2 U2 L2 B2 L2 D' B2 D2 B2 L2
19.   18.50    48    2.59    D' B L2 B' R' L' B' U2 D' B' U2 F2 D' F2 U F2 D' R2 D2 R2 B2 U'
20.   (26.20)   56   2.14    U2 F' B' R U2 F D' F2 U' R' D L2 F2 U' F2 U' R2 U F2 D' F2 D

Total solves so far: 220
Ao200: 22.18
Ao100: 21.75

Best Ao100: 21.75
Best Ao12: 20.16
Best Single: 13.53

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns (4/6)
U: back row (1/6)
L: best (1/6)
S: sune (1/6)
As: antisune (1/6)
Total: 20/42

 

[mrrubicskuber]

Nice progress! Roux is such a fun method, I used to use it quite bit and I was averaging around the same as you are now (however I never learnt that much CMLL). Good luck on future progress.

 

[Foreright]

Haven't managed to do that much solving over the past couple of days due to family stuff but I did manage to learn and drill the final two T algs. Interestingly I think I've reached my CMLL limit for now as I'm starting to get a couple of the existing ones confused and my CMLL recognition has slowed a little. I'm going to put off learning any more for a couple of days to drill the existing ones properly and get them fully ingrained.

I pulled my Gan bluetooth timer out off the shelf today and all did all my solves using it. It adds a whole layer of difficulty to a solve I think - Given I have to actually put the cube down and let go of it before starting it's making full remembering my FB + SB DR solutions a little harder. On average, pickup and putdown is adding just about a second to my times according to the stats so I'm actually quite pleased with my progress. I might stick with using it all the time now.

For today:

total solves: 50
avg of 50: 22.10
mean time: 22.17
mean stm: 59.24
mean tps: 2.70

That includes 13 sub-20 times which is a positive. However, if I look at my stats for today (these are averages, ie. discard top/bottom 5%):

... we can see that I'm pretty much 2 seconds over my target but that almost all of that is attributable to my horrible second block. I've been experimenting with slowing the first block so that the recognition time is much lower and it "flows" more and that works to an extent, however, my turning speed is hardly that earth shattering in the first place so I'm not sure this is a good idea I do need to find some way of knocking at least 2 seconds off my average SB though which is going to be my focus for now.

---

Total solves so far: 300
Ao200: 21.71
Ao100: 21.59

Best Ao100: 21.59
Best Ao12: 20.16
Best Single: 13.53

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns, rows, back row (6/6)
U: back row (1/6)
L: best (1/6)
S: sune (1/6)
As: antisune (1/6)
Total: 22/42

 

[Foreright]

Not a great day yesterday - only managed 31 timed solves due to work interfering

I'm concentrating much more now on actually doing 1L CMLLs when I hit them but amusingly that doesn't really seem to be slowing my CMLL stage down that much. My second block is still absolutely terrible in terms of time. Move count is pretty bad but not the worst. As above, it's really going to be knocking that 2 seconds off my SB that gets me sub-20 regardless of anything I do with CMLL as my other stages are on target.

Trying to also use DFDB in the LSE stage a lot more as it's something I practiced extensively last time I tried this last year and I did a refresher on it last week. For the cases where I have both the LL edges on the bottom layer ready to insert I find it easy to do but for the other case where you insert with M U2 M / M' U2 M' I haven't worked out the trick to tell what to do yet. That would likely knock a bit off my LSE times as there's a noticable pause once those edge are inserted where I try to work out how to "MU" at the end.

Decided to ignore my own advice and start learning the next CMLL set. From my stats, it's actually the "S" set that's coming up more than any other for me right now (17% of my solves) so I'm learning that one. Refreshed my memory on the 3-sticker recognition in the "normal" sune position and have learnt 4 of the set and plan to drill them a bit tonight before learning the other 2 tomorrow. The algs themselves are pretty easy to remember but remembering which alg goes with which case (ie. same-opposite, opposite-adjacent etc.) is pretty tricky and will need a LOT of drilling.

At this point:

Total solves so far: 331
Ao200: 21.87
Ao100: 21.84

Best Ao100: 21.59
Best Ao12: 20.16
Best Single: 13.53

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns, rows, back row (6/6)
U: back row (1/6)
L: best (1/6)
S: sune (same-opposite) (1/6) learning: same-adjacent, opposite-adjacent, adjacent-same
As: antisune (1/6)
Total: 22/42

 

[Bingus cuber]

Sorry if I sound dumb, but what is CMLL and what does it solve?

 

[Cubinginatree]

Sorry if I sound dumb, but what is CMLL and what does it solve?

CMLL is an algorithm group that solves the corners in Roux. It happens after first and second block and before L6E (Last six edges).

 

[Bingus cuber]

Ok, thank you.

 

[Foreright]

Sorry if I sound dumb, but what is CMLL and what does it solve?

Yep, as above, it's an alg set that orientates and permutes the top corners, ignoring the edges (which are solved in the last step). There are 42 of them in total.

 

[OreKehStrah]

Not a great day yesterday - only managed 31 timed solves due to work interfering

View attachment 28657
I'm concentrating much more now on actually doing 1L CMLLs when I hit them but amusingly that doesn't really seem to be slowing my CMLL stage down that much. My second block is still absolutely terrible in terms of time. Move count is pretty bad but not the worst. As above, it's really going to be knocking that 2 seconds off my SB that gets me sub-20 regardless of anything I do with CMLL as my other stages are on target.

Trying to also use DFDB in the LSE stage a lot more as it's something I practiced extensively last time I tried this last year and I did a refresher on it last week. For the cases where I have both the LL edges on the bottom layer ready to insert I find it easy to do but for the other case where you insert with M U2 M / M' U2 M' I haven't worked out the trick to tell what to do yet. That would likely knock a bit off my LSE times as there's a noticable pause once those edge are inserted where I try to work out how to "MU" at the end.

Decided to ignore my own advice and start learning the next CMLL set. From my stats, it's actually the "S" set that's coming up more than any other for me right now (17% of my solves) so I'm learning that one. Refreshed my memory on the 3-sticker recognition in the "normal" sune position and have learnt 4 of the set and plan to drill them a bit tonight before learning the other 2 tomorrow. The algs themselves are pretty easy to remember but remembering which alg goes with which case (ie. same-opposite, opposite-adjacent etc.) is pretty tricky and will need a LOT of drilling.

At this point:

Total solves so far: 331
Ao200: 21.87
Ao100: 21.84

Best Ao100: 21.59
Best Ao12: 20.16
Best Single: 13.53

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns, rows, back row (6/6)
U: back row (1/6)
L: best (1/6)
S: sune (same-opposite) (1/6) learning: same-adjacent, opposite-adjacent, adjacent-same
As: antisune (1/6)
Total: 22/42

Your SB move count is fine. Top solvers aren’t that much more efficient.

 

[Foreright]

Your SB move count is fine. Top solvers aren’t that much more efficient.

Time however is not - I’m aiming for 6 seconds and am at over 8 on average and quite often 10+ for some solves.

 

[Foreright]

Turns out that the "S" set is not difficult to remember (for me at least!) and recognize and the algs are super easy to learn - I've learnt and drilled them all today and apart from one of the "T" set that I seem to be a bit slow remembering I think I have everything I know drilled and ready for a real solve now which is a positive. Just 3 sets to go now - am going to do As and L next and leave U for last as I seem to remember that was quite an easy one or maybe I'll just do U next - will decide in the morning

Looks like I'll probably be done with CMLL, at least in terms of learning the algs, well by the end of the month at this rate. I suspect the As set is going to be easy to learn also as they are just inverses of the S set - just need to learn the recognition for them. Hopefully I can retain them a bit better this time around!

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: left bar, right bar, front row, columns, rows, back row (6/6)
U: back row (1/6)
L: best (1/6)
S: all (6/6)
As: antisune (1/6)
Total: 27/42

 

[Foreright]

So, amusingly, I received a GAN i3 yesterday in the post. Bad news is that it's (obviously) set to use "standard" colours. I've been using the Japanese colour scheme (ie. blue and yellow swapped from standard) for over 30 years so as is it was pretty much unusable for me. No issue I think, I swap the colours on all my cubes (of every size!), but... GAN have been making harder and harder to do this with every release and I'm finally beaten. I simply cannot get the corners apart without either knowing exactly how they go together or potentially destroying one to see how it works. Annoying.

Regardless, I'm using this as an opportunity to try to swap to standard colour scheme (gulp!). I've swapped a couple of non-smart cubes (my WRM maglev, Tornado v3 and GAN X infinity) already so I have some around the house and in the home office to practice with. Holy crap, it's a proper mind-melting frustration

I did 50 (timed - lots of untimed...) solves yesterday for an average of about 35 IIRC which is quite humbling. First block is not too bad, nor is CMLL after a few solves getting used to the colour change but yikes, second block - it's REALLY difficult to remember that blue is now a side colour and *yellow* is the top/bottom and same for LSE. I suspect it's going to take a pretty long time to get used to this although my times are coming down steadily already.

I also did 50 solves today:

This is over 8 seconds above my usual time but I've pretty much already got FB and CMLL to the point where they are not that far off where I want to be. It's SB that's *really* suffering.

I'm going to give this a couple of days - maybe to the end of the week - and reassess whether I really want to do this as there are advantages - being able to buy new cubes without gambling whether or not you have to send it back 'cos the pieces won't come apart is a plus! Has anybody else tried something similar to this before? Ie. changing colour schemes? If so, how long did it take to be comfortable? I'll be interested to see if I ever get to the point where "standard" becomes natural to me whether I can still solve "japanese" too

As an aside, the GAN i3 is a MASSIVE improvement over the iCarry. It's obviously not as good as a modern flagship but it does actually feel like a speedcube to me and I was able to set it up fast and loose enough for my preference.

CMLLs Learnt:

O: all (2/2)
H: all (4/4)
Pi: all (6/6)
T: all (6/6)
U: back row (1/6)
L: best (1/6)
S: all (6/6)
As: antisune (1/6)
Total: 27/42

 

[Foreright]

Some progress yesterday - had a few solves where I got some kind of "flow" going followed by a few where it looked like I'd never solved a cube before Times are coming down pretty steadily though and more to the point, I'm starting to "see" the blue pieces after years of training to ignore them during the block stage.

50 timed solves again for an average of 24.48 which is over 3 seconds down on yesterday. Second block is still the slowest step relatively and the step where I'm still struggling the most to recognize that yellow goes on the top and not the side!

Positives for today - a 16.11 single (https://cbst.link/s/uVXSnBJhNWJF5qAwN6Kdq5) and my move count is (relatively) low for me too as I'm solving slower and am more efficient overall when I'm not confused by the colours.
Total progress so far since I restarted - you can see pretty clearly where I switched colour schemes Hopefully we'll see more of the green and blue and less of the orange and red over the next couple of days!

 

[Foreright]

Steady progress 54 solves today - some good, a lot bad but I am starting to get some faster solves where I'm seeing the blue pieces and am able to associate them with being on the side rather than the bottom - which is the main issue I have: thinking that yellow goes on the side rather than the bottom!

Down to sub-24 for the day. My Ao100 line is shallowing out a bit but progress is steady as I say so not worried yet.

I might start learning a new CMLL set this weekend since it's taking approx. 1 day to learn each - most likely As and leave L and U for last. I'm at 65% 1L CMLL in speed solves. Part of my newfound "slowness" is down to that tbh - I'm trying to do the 1L CMLL (assuming I know it obv.) for every speedsolve and it does slow the recognition for that step down a little as opposed to just spamming the 2L.

My second block is still a real issue though. 3.5 seconds over target for a 20s solve is pretty bad. A fair chunk of that *is* down to the colour scheme swap but still - that's a load of time to drop. I don't feel that my solves are horrifically inefficient - I mean, low-50s would be better than almost 60 but even at 60 STM, 3.0 tps+ would get me sub-20 and I don't feel that's unattainable if I can reduce the pauses a bit.

 

[Foreright]

Still improving - getting faster very steadily (ie. slowly!!) but I'm definitely feeling more comfortable solving with this colour scheme which is a plus and I'm going to stick with it going forward. Didn't manage to learn any additional CMLL over the weekend due to family commitments but I'll get on that during the week at some point.

58 timed solves today and my progress so far since I switch colour schemes looks like this over 350 solves:

Ao100 stands at 22.60 and dropping with nearly every solve. Ao200 stands at 23.16. My move count has dropped a bit overall but has plateaued a bit at around 56-57 as an average which is a little bit higher than I'd like but with 2/3 of CMLL learnt I guess it's probably about right and certainly within the bounds of being able to average 20 as that would mean a tps of just under 3. At the moment I am trying to slow my turning down a bit so I can be more efficient and look ahead a bit easier.

Highlight of today was a ridiculous scramble towards the end which resulted in a 31 stm solution with a CMLL skip. It was *mega* slow though at 18.27 (1.7 tps!!) and should have been more efficient than that as I had a couple of "R R"s in the solution. I also had a 39 stm solution with a CMLL skip which again was way slower than it should have been at 21.54. Two CMLL skips in one short session is pretty cool.

https://cbst.link/s/3rXKaLKRWCacozvREh4XPY
https://cbst.link/s/dyVhn4ZPKU3P7L3XiFn4hr

 

[Foreright]

Well, not a great week. Not been able to do that much solving but what I did was concentrate on efficiency and doing slow solves in front of the TV, when I'm idle etc. No learning CMLL or other techniques or anything like that, just trying to finally get used to this colour scheme and I believe I'm nearly at that point now except...

My times are utterly HORRIFIC. I'm struggling to break 25/26 on most solves. My average has tanked completely. This is all completely due to my utterly HORRIBLE second blocks. It's not just that they are inefficient (which they are...) but I spend FOREVER looking for pieces. Even if I totally memorize my FB solution and track the DR edge I then have a 2-8 second pause where I'm looking for anything to continue with on pretty much every solve now. No idea what's happened but I'm at the point of just giving up now as I have no desire to pick up a cube most of the time because of it let alone try to knock 4-6 seconds off my AVERAGE SB time to get to 20 seconds which now seems utterly impossible.

As I say, I'm quite capable of efficient solutions and have many solves which are in the 40-50 move range assuming I get a 1L CMLL. It's just that my SB solutions are horribly slow or filled with ginormous pauses which make me want to puke when I watch them back in Cubeast.

No idea how to proceed now. I shouldn't have to be 100% working out FB+DR solutions in inspection and implement full lookahead and 1L CMLL to get sub-20 according to conventional wisdom but that seems to be the case for me and that's not happening in the short term. Given I've been using Roux for over 6 years on and off I feel like I've hit the end of the road here as I'm not getting better.

 

[Foreright]

"Loving" the fact that someone "likes" the fact that I'm about to give up and am having a totally **** time of it (or not as the case may be).