So I guess it will be like a function to filter hands by frequency and only display hands with that frequency or within the range of selected frequencies

Thanks

I'm looking to install more ram in my computer. I know more ram allows you to solve larger trees and a faster processor will solve faster. How if at all does ram frequency effect things? I know it can be better for other things specifically gaming but will GTO+ gain any benefit or solving speed? Thanks

Why does GTO+ tell me that I have insufficient memory when I try to run the solver?

I'm trying to run multiple database solves at once:

1) UTG 3x, BB call (Range 1, 2)
2) MP 3x, BB call (Range 3, 4)
3) CO 3x, BB call (Range 5, 6)
4) BTN 3x, BB call (Range 7, 8)

Here are my steps:
1) Build the tree
2) Assign Range 1 & Range 2
3) Save file based on ranges 1-2
4) Repeat steps 2-3 for the other ranges; they have the same tree options, but different ranges for each player

5) Then, "Process all .gto files in a given directory"

The problem I'm having is that it seems like saving these files is not also saving the ranges along with them. So I just solved 4 different files with identical trees, intending to have different ranges for them, but now I realized it just used the last-used range & solved the exact same thing 4 times.

What am I missing? Can I run different hand ranges for different .gto files overnight? Or are these files only tree-specific, and there's no way to run multiple different hand ranges?

Thanks scylla!

https://www.gtoplus.com/memory/


That link above explains it fairly well. Just experiment with the tree building settings to see which reduce the memory the most. Doing things like going from 4 flop bets to 3 or 2 helps a lot. I'm personally also reducing down to 2 bet options for turns and rivers (maybe ~2/3 pot and 125% pot) for my large multi-flop runs, and only 1 possible raise/reraise sizes postflop (75%).

I'm not sure what the best/most optimal simplifications are, but so far I like:
1) turn off donkbetting if OOP player is not PFR
2) reduce flop bets to 2-3 options (so far i like 33, 70, check, as overbetting flop seems rare)
3) reduce turn/river initial bets to 2 options (66, 125)
4) set raise sizes on all streets to 1 option (i have 75 arbitrarily)
Note these numbers are just my initial best guesses, I'm still a beginner here
I'm also curious here, I have 3600mhz cl15 for gaming but only 8GB.

Regardless, if you install slower speed / different latency ram, you might run into instability issues. From my (limited) understanding, installing slower speed RAM will simply cause all of your RAM to reduce to that speed, not sure about stability. Also not sure about different CAS latencies but have heard it can have instability issues if we mix

I'd definitely be interested in installing additional RAM if it can speed up solves at all, interested to hear scylla's notes on this

I have 16gb right now at 2133mhz. I was looking to upgrade to 32gb as I want to solve heads up single raised pot spots which with wide ranges and multiple bet sizes I will need 20-30gb to run. Some ram kits its not much difference in $ between say 2400 and 3000mhz. Also I'm pretty sure having more ram wont help the solve speed but having higher frequency might, though it might be negligible anyway. But for these heads up spots even a 10% increase might help. I had some SBvBB single raised pot spots with multiple bet sizes across each street taking over a hour per flop and I have a decently fast processor, and heads up ranges are about twice that wide. I might even get 64 gb, ram is getting pretty cheap, I can get 64gb for like $200. Especially if wanting to maybe solve deeper than 100bb spots.

For a video on that go here: https://youtu.be/YuRsCQajRBA

For that use "Lock hand for hero" and "Play versus database".

For that, select "Play against tree".

Such statistics can be found within GTO+ in the aggregate reports.

The 9d7s4d flop in your database took 6 minutes to solve on my 8 core system. It's a fairly large tree, so that seems like a reasonable time. Is your solving time considerably longer?

For this, go to the graph for a decision and filter by frequencies.

In my experience the RAM speed has no influence.
For GTO solving, the speed of the CPU is the limiting factor.

The required RAM is displayed in the tree builder.
Your available RAM is displayed in the middle left part of the screen.
Please make sure that you have at least twice the required RAM available.



If you find that solves take too long, then please consider using just single bet sizes on the turn and river. This will considerably reduce the size of your tree, and make it solve faster, while not significanly affecting the quality of the solution. The turn/river lines are only needed to estimate the EVs for the followup lines.

I really don't think that there's much added value to building such huge trees. Even the simple ones already far exceed what the human mind can handle, so more complexity will, in the end, not lead to an increase in the quality of analysis.

You should reverse steps 1 and 2.
So first assign the ranges, and then build the tree with those ranges.

RAM only affects the largest size of tree that can be fit into memory.
It does not affect solving speed.

Thank you scylla, I love how quickly & in-depth you respond to questions. Hope you can maybe get some help answering questions in the future.

Gotcha. Thanks a lot. You are great.

I guess i will be upgrading to 32GB.

Other than the solving which is less than an issue, I do wonder why it takes like 3 seconds for the cards to completely show on turn and river. Like you know when the cards load on the "play against the solution", or simply just when clicking the tree (a timer of 0-100 shows up and it can take a lot of time. Laast night it took up to 8 seconds.



I'll have to check it out again and get back to you. But what would be the reason for Flopzilla and GTO+ to lag so much when working together? is that a CPU bottleneck or something else?

If I'm playing BtnvsBB in SRP, and assigning range for BB calling range that does not include all the hands I usually 3bet, do I need to also run BTNvsBB in 3bet pots separately with only those 3bet hands that I raise against button for value and bluff, or i can add them somehow to the original ranges of BtnvsBB and run them on the same trees?

Do you recommend running specific ranges against each specific flop as a separate save file, or just running the database for 50+ unique flops is recommended?

And what does weight mean?

Thanks Scylla. Appreciate your contribution.

After assigning the new ranges you have to go to 'Build tree' and build a new tree, then you can go to 'edit tree' and accept changes which will be with your customized tree as long as you haven't imported the new tree.

Thanks, tried this overnight and it worked. Not the first time I'll screw up a 24-hour-long analysis I'm sure

Would it be possible to get a 'hand navigator' for the starting hand distribution? After inputting the turn and river I could look at the flop and see what lines it distributes each starting hand into? Would be quite helpful to look and see when it considers a hand a mix or pure strategy.

Ok, I will consider it for future releases.

If the ranges are different, then this is a separate situation.

Whether you want to create separate savefiles or a database is not something where I can offer a recommendation. This is up to the user.

When solving, you have used the option "Basic storage". This means that the turn data is not stored, and recalculated when requested by the user. The advantage of this approach is that the savefiles are much smaller. The disadvantage is that data sometimes needs to be recalculated live. If you prefer not to recalc data, then use the option "Extensive" storage.

Why does the turn report not match closely to the sim once its run? Example here, the 8 of clubs says bet 4.36 42% of the time in the turn report but when I select the 8 of clubs and solve, its 33%.

If you have used "Basic" storage, then the turn report will have been based on the regular solve. However, the data that is shown when going to the actual line in the tree is based on a recalc. So the data comes from two different solves. When using "Basic" storage, it can happen for there to be differences between the regular calc and the recalc. This will become less and less frequent when solving to a lower dEV. If you want to avoid this altogether, then use "Extensive" storage.

but let's say if I just want to view hands at a decision that bet 75%< only, are we able to do that on the matrix and in the popup?

Hello,
How can I activate the "absolute mode" in flopzilla pro?

Hi I am trying to get the tree built so that otf villain min raises my cbet
No matter number I put into the "IP settings" ----> "OOP has bet" category, the amount of 70 always shows up in the tree for villains flop raise size vs cbet