Foreright’s road to sub-20 and learning CMLL

[Foreright]

Well, I see my last post in this thread was not very positive xD. I had a pretty long break after this - did the odd slow solve but nothing serious. Decided at the beginning of this month to make a proper push to sub-20 once and for all.

So, ironically, after all that time I’m pretty much back where I started this thread. Decided to do some timed solved and see where I was - result was average of 50 around 23.5 seconds. I remembered O, H, 4/6 Pi and 2/6 T CMLLs with the rest being 2 look.

After 400 or so solves my Ao100s are around 22 seconds. Similar to before, Cubeast shows that my FB is good (enough), CMLL is plenty fast enough, even 2L. LSE is reasonable but could be faster but SB is horrible.

It’s not uncommon for me to get a 24 second solve where the split is 3/12/4/5 or even 20-21 second ones where 10 seconds is spent on just SB.

To combat this I’m doing a lot of slow solving and really looking at my SB pair solutions and making sure they are vaguely optimal. It’s going to be a long job. At the same time I’m really making sure I absolutely plan full FB+Dr edge on almost every solve and that is bringing some results in that there’s significantly less “gap” between FB and SB. I’m still losing sight of SB pieces regularly though. I’m hoping that will get better with time.

Spent some of today drilling the missing CMLls in the sets I already know. I’m now pretty confident with O, H, Pi and T which is pretty great - that’s half of it done already. Going to start drilling U tomorrow if I get a chance.

I’m pretty confident I can get down to below 21 in the next week or so provided I spam enough solves. I’m finding SB pieces easier to find with every solve.

I said this before but I think that to get to sub-20 average I really need a lot better efficiency - partly that will come from more CMLLs. I’m sometimes able to get above 3 TPS which is going to be sufficient, even with 2L CMLL. Part of me thinks “just turn faster” but I’ve noticed that my efficiency REALLY suffers when I do this - like, I can push 3.5 tps but move count goes up into the 70s and above at that point.

 

[Foreright]

Spent a couple of hours whilst the wife was watching tv this evening drilling some more algs - namely U/Front Slash, U/Back Slash and U/X to add to U /Back Row which was my 2-look that I already knew.

Need to find a good U/Front Row alg though - going to struggle to remember R' U' R U' R' U2 R2 U R' U R U2 R' which is the inverse of my T alg but I seem to be really struggling to memorize!

Totals [25/42]

O [2/2]
H [4/4]
Pi [6/6]
T [6/6]
U [4/6] New: U/Front Slash, U/Back Slash, U/X
L [1/6]
S [1/6]
AS [1/6]

 

[Foreright]

Didn't get too much time for cubing today due to household chore and stuff but have managed to do some more drilling this evening and I'm starting to get the last couple of U algs integrated also.

One of the issues I'm finding, and I've probably mentioned this previously, is that I'm able to pretty quickly learn a bunch of algorithms and CMLL recognition of all non-S/AS cases is pretty much instantaneous at this point after solving with Roux for so long. I'm having trouble actually associating the algs and cases though, especially those that are fairly similar or are inverses / mirrors. I guess this will get better with more drilling and practice - well, at least, I hope it will!

Going to start learning S next - possibly tomorrow.

Edit: oh wow - thought I’d have a quick look a L and… it’s super easy for some reason! Got 4 down (not including the 2L one!) within 30 mins. Going to drill them properly at some point tomorrow but i wasn’t expecting that!

Totals [31/42]

O [2/2]
H [4/4]
Pi [6/6]
T [6/6]
U [6/6] New: U/Rows, U/Bottom Row
L [5/6] New: L/Front Comm, Back Comm, Pure, Diagonal
S [1/6]
AS [1/6]

 

[Foreright]

FWIW I did actually learn the last "L" case a couple of days ago which means I know everything (haha) except for Sune/Antisune cases. I know that last time I went through learning CMLL I left quite some gap between now and learning those as I found the recognition easy enough, but tricky to do quickly (and recall the correct algorithm) and it was quicker for me to just 2 look S/AS. I'll probably do similar here.

I have found that learning the L set has screwed up so of the other sets for me so I'm having to drill everything constantly. It's super annoying to forget one of the Pi cases that I learnt AGES ago and had in muscle memory but it is what it is. We've had a week off work with the kids and doing house chores / tidying up so drilling time has been reasonably limited and late in the evenings when I'm tired which isn't really helping. Planning to get back to actual solves mid next week when it's calmed down a bit and see how my new knowledge fares xD

Totals [32/42]

O [2/2]
H [4/4]
Pi [6/6]
T [6/6]
U [6/6]
L [6/6] New: L/Good
S [1/6]
AS [1/6]

 

[Foreright]

Started using my newfound knowledge in actual solves with pretty positive results actually - bit of a blip at the start of the session today which is to be expected but got into my stride with the new CMLLs pretty quickly and my efficiency has gone way up - average turn count has gone from mid-60s to mid-50s which is a step in the right direction, especially as I know my SB is far from efficient. I'm getting quite a few solves in the 48-52 range which is my eventual target.

In terms of times, I'm at 22.10s for the 50 solves I did today which I'm very pleased about. That's slightly down on my 2L CMLL times which given I am taking more time in general for CMLL recognition is a positive. SB is still the thing holding me back. Even with the bad SB, I had 11 sub-20 times and 4 sub-18 including a 15.99 today which is extremely positive.

On that note, anyone have any ideas as to HOW to practice second block? OnionHoney is really not cutting it as it's literally *being able to find stuff* in a speed solve that's the issue rather than SB solutions per-se. I can spam the SB solutions at easily enough pace for well sub-20 once I actually find the pieces! xD


That said, the last week learning CMLL has been very productive. Hopefully I can retain them. Going to give it a couple of weeks of solves and CMLL drills before attempting to learn the last two sets (S/As).

 

[Foreright]

Had family / home stuff going on for the last few days so although I've been able to do *some* slow solves and some CMLL drills I've not really been timing.

That said, I sat down and "just learnt" the AS set in its entirety. Actually, the algs themselves are quite easy to remember. Remembering which goes with which case is a different matter but an hour of drilling in front of the TV and I'm pleasantly surprised that I seem to be ok with it - for now. We'll see in the morning!

I spent a fair bit of time yesterday evening preparing for this - made the following in Word (which incidentally I note a mistake on - Left Bar is actually "S-O" and Sexy-sune-sledge is actually "S-A"...) which is 3-sticker recognition for AS which I've not been able to find anywhere else (!). This is pretty much the mirror of the Sune 3-sticker recognition I was planning to use and used previously. I'm also recognising in the position in the diagram rather than the more usual one as it's a left-right mirror of the Sune case recognition I use and I find it easier:

I actually find "(F') r U r' U2' r' F2 r" easier to remember for "Back Slash" / A-S and have changed that on my physical copy.

Excitingly that means I only have the 5 remaining S cases left - depending on how tomorrow goes with drilling and if I get any time to do so between calls at work I'm might get on with that tomorrow - certainly this week though.

Totals [37/42]

O [2/2]
H [4/4]
Pi [6/6]
T [6/6]
U [6/6]
L [6/6]
S [1/6]
AS [6/6] New: all!

 

[Foreright]

It's been a couple of weeks since I did finally learn the last few algs in S so CMLL is "complete" in that I have learnt them all. Getting them out quickly in a solve is another matter however! I'm making some mistakes in L/S/As sets (with recall more than recognition...) so more drilling is needed but I'm pleasantly surprised how smoothly it went. I have noticed that my CMLL times are more consistent for other alg sets - which is to be expected I guess as I'm not doing some cases as 2L anymore.

Of course, that now leaves the rest of the solve as an issue - notably second block which is my worst stage by a country mile. Been working on slow solving and trying to improve efficiency down from my current average move-count of 19.x to something nearer 15. I'm pretty easily able to get an average movecount over 20-30 solves for SB of 16, even without learning optimized algs for a few horrible cases, but those solves are in the 30-40 second range. Need to bridge the gap between that and solving at a better pace which is the difficult bit!

Totals [42/42]

O [2/2]
H [4/4]
Pi [6/6]
T [6/6]
U [6/6]
L [6/6]
S [6/6] New : All.
AS [6/6]

 

[abunickabhi]

What is your average drill time with each CMLL case?

 

[Foreright]

What is your average drill time with each CMLL case?

It varies depending on the case - average for last 100 solves is 3.92 for speedsolves. There's still one or two it takes a while to recognize / differentiate in the heat of the moment (mainly the ones in red below which are the slowest).

For sub-20 I'm aiming for under-4 for CMLL so I'm reasonably happy with progress so far albeit those cases that are 6 and above need work.
Right now I'm trying to concentrate on block building as it's by far the weakest part of my solve. It's not unusual at all for me to get a 21 second solve where 10-12 seconds of that is *just* the second block. I'm really struggling to improve SB. Either I solve the first block fast (sub-3 - probably 8 moves or so) and then my mind just goes completely blank and I spend 5 seconds trying to find SB pieces or I slow down the first block attempting to do lookahead into the SB and then the first block ends up taking 6 seconds instead with a 7-8s SB which is pretty much the same result.

I was finding last week that after 50 or so solves that my lookahead, such as it is, was improving but that resets overnight. I wonder if just spamming hundreds of solves is going to improve my SB but there's likely something more targetted but god knows what that is (!). I've spent a huge amount of time "slow solving" and whilst yes, if I solve at 1 tps, I can generally get through SB without pausing a vast majority of the time and in a reasonably efficient way (again, I spent a fair amount of time improving the *way* I solve SB so it's much more efficient that previous - average 17-18 stm), ultimately I need to turn faster without pauses and bridging that gap between a 1 tps second block solution and a 2.5-3 tps solution *with no pause* seems impossible. Don't get me wrong, I do have the odd solve where the SB pieces "just appear" before my eyes and I get a 3.5-4.0 tps second block which results in a very fast (for me!!) time, generally below 16s as long as I don't flub the CMLL/LSE but those are balanced out by the 25/26+ second solves where my mind goes completely blank.

Anyway, this is a ramble - if anyone has any suggestions as to how to improve my SB stage - literally to remove the big pauses then I'm all ears. Current average is a smidge over 21 and my stated aim has always been sub-20 so it is literally just second block that's stopping me getting there:

There's some scope to improve efficiency here in terms of STM but really, to be sub-20 I should be ok with low 50s as 53.27 requires a TPS of 2.7 to break 20 which is not insurmountable.

 

[Foreright]

Ok, so I haven't posted here about times for a little while. Been trying, as per last message, to work more on my second block than anything else and that's not helping with my times as quite often I'll get a *really* good FB+SB and then completely screw up the CMLL because I wasn't paying attention. Regardless, 28 times solves so far today and my last 100 solves look like this:

This is a slight improvement (0.3s or so...) in second block over the last week or so, pretty much all due to slightly reduced recognition time. Some of this is at the expense of slightly slowing down at the end of FB to make the transition a bit smoother so my overal efficiency / tps is not really improving right now.

I did have a couple of solves in a row where I had SB with TPS at around 4.5 which is extremely encouraging as I'd thought I'd never be able to do that in a million years - just need to do it on more solves now!

The above graph shows recognition time (note I'd restarted a new "session" so not many solves yet) - ie. the pauses *between* steps so the bold orange line is a moving average (of 50) for my FB to SB transition time. It's come down noticably in the last 50 or so solves where I've been concentrating a bit more on reducing specifically that. Hope to get it down to below 1s as a short(ish!) term goal and below 0.75s as a longer term one -- not really sure what's realistic here but we'll see.

Nice to see CMLL recognition times falling a bit also as I'm slowly getting better at recognizing all the cases I've learnt. Again short term goal is sub 1.5s with a longer term one of 1s -- not sure what's realistic here either.

All in all, some encouraging stuff - slow progress but hopefully steady!

 

[Foreright]

Again, it's been a couple of weeks - been setting up a garden room / outside office so although I have been solving a bit casually I've not been doing timed solves. It's only the fact that my eldest son (10) is showing some interest in cubing again after a little while off that I've remembered that I'm supposed to be going sub-20 xD We successfully learnt a LBL method over the weekend and he's super excited to be able to remember how to do a 3x3 without my notes and has taken his cube to holiday club.

I thought I'd have a quick go at some timed solves and after a bit of a warmup ended up with easily the best Ao12 I've had since learning full CMLL with the last 5 solves of that being the best Ao5 also. That's quite encouraging, especially as they are not amazingly efficient. My TPS is up from generally around 2.5 to almost 3 in some solves which is a big improvement and proves that I'm starting to do some "lookahead" or at least I'm seeing stuff faster as that's my biggest bottleneck right now:

Generated by AnalysisNew.xlsm
current ao12: 20.29

Time List:
#    Time    STM    TPS    Scramble
1.    17.21    50    2.91    U L' D2 R2 B D F U' R' F U' D' B2 U F2 R2 F2 B2 U R2 F2
2.    16.43    48    2.92    U2 F2 R' D F U2 L' F' R B R2 F2 L2 U2 F2 D L2 U2 D B2 L2 U2
3.    19.50    58    2.97    B' L U' L' D2 F U2 L2 B U R2 B2 U' L2 F2 L2 D2 F2 D L2 U2 B2
4.    16.72    50    2.99    U R2 U2 F2 R B R2 F2 U' L' F L2 D B2 U' L2 U2 D B2 L2 U2 B2
5.    20.49    43    2.10    R2 F2 D' R' B R2 L F' U' L U2 D2 L2 B2 D F2 D R2 B2 L2 U' L2
6.    22.50    55    2.44    L U B R2 L2 U2 R D2 F U' F2 U' R2 L2 B2 D' B2 L2 D2 L2 U F2
7.    21.93    60    2.74    F' R' U' D L' U' F2 U L2 F U B2 L2 F2 U R2 D B2 U2 D F2
8.    21.88    49    2.24    U F2 B L2 F L2 U' L2 B R F2 U F2 B2 R2 B2 R2 U' F2 B2 R2 U'
9.    21.34    61    2.86    U' B' U2 R U2 R' U2 B U' B2 L2 D2 F2 L2 U R2 D F2 U B2 D'
10.    22.65    52    2.30    F' D' F L D' B' D' F2 R F B2 L2 D R2 B2 D R2 U F2 R2 U' F2
11.    20.75    64    3.08    U' L2 U2 B R' B2 U2 D' L F2 L2 U' D2 L2 F2 B2 D F2 R2 F2 R2 D2
12.    20.59    49    2.38    F' B' U' R' D' R2 L2 F D2 B2 U' B2 L2 U2 B2 U R2 D' B2 R2

Generated by AnalysisNew.xlsm
current ao5: 17.81

Time List:
#    Time    STM    TPS    Scramble
1.    17.21    50    2.91    U L' D2 R2 B D F U' R' F U' D' B2 U F2 R2 F2 B2 U R2 F2
2.    16.43    48    2.92    U2 F2 R' D F U2 L' F' R B R2 F2 L2 U2 F2 D L2 U2 D B2 L2 U2
3.    19.50    58    2.97    B' L U' L' D2 F U2 L2 B U R2 B2 U' L2 F2 L2 D2 F2 D L2 U2 B2
4.    16.72    50    2.99    U R2 U2 F2 R B R2 F2 U' L' F L2 D B2 U' L2 U2 D B2 L2 U2 B2
5.    20.49    43    2.10    R2 F2 D' R' B R2 L F' U' L U2 D2 L2 B2 D F2 D R2 B2 L2 U' L2

At this point my Ao200 (21.50s), which I'm using as a global average, is about 0.8s above my best before I committed to learning full CMLL and is coming down steadily. I honestly think that I just need to spam 100s / 1000s of solves at this point to "force" my lookahead to get better as it is improving - my efficiency (52 STM average ) is good enough for sub-20 - at 3 TPS that would result in a 17.33s average which I'd be over the moon about.

 

[Foreright]

Been doing a little bit less solving recently due to house maintenance, moving into a new (garden) office etc. but trying to "keep my hand in" a bit with slow solves. Just decided to do a quick Ao5 to see where I was and got this:

19.89s Speedsolving average of 5 by BewareTheWereHamster on Cubeast

cbst.link

There's some positives here - move count is very good for me, still showing the same bottlenecks though - namely second block, my nemesis, and then the looooong pauses between steps albeit that does reduce after some solves as I get used to looking for pieces again. Then there was this as part of that average which is a big "ouch":

20.27s Speedsolving solve by BewareTheWereHamster on Cubeast

cbst.link

4.8s recognition / pause (!) and then 2.72s execution for SB is just... ugh...

So a bit of a mixed bag so far... global average is 21.11 and reducing.