The solved trees themselves will require RAM as well in order to be stored in memory. With "Basic" storage the required RAM per tree will be negligible, but with "Extensive" storage, having roughly 100 trees in memory adds up surprisingly quickly. So as more trees in the database are solved, the available RAM will drop.

To get around this, it may be preferable to store the trees as separate files, solve those files, and only after solving merge them together again into a database.

So basically you'll need to:
1) Export the database into separate files
2) Solve those files
3) Merge the files together again into a database

For instructions on how to this, go here: https://www.gtoplus.com/processingdatabase/

Hi, would it be possible to add a rounding feature? For example, after a GTO solution is produced could there be an option to round the frequencies to a certain percentage, like 5%, 10%, 25%. And then toggle between the true GTO and rounded solutions.

I know that rounded solution wouldn't be a pure GTO, but humans can't really play true GTO strategies across all trees anyways and ability to round up would really help with studying.

I've been using GTO+ for quite some time now and it really is my favourite postflop solver, I do agree with some of the people who said adding a option to input a geometric sizings would be great, especially for low SPR scenarios, for example 1.5 SPR and we would like to use 25% and geometric (50% in this case) using the "get the money in smoothly" option the smaller sizings get stripped from the tree.
Also would like to request a small change/new feature in the "Play against solution": setup a threshold for an action to be considered correct and warn me when playing in the training mode, cause I don't want to drill a sim and have to open the tree and see that my action was taken 0.05% or the time which could be very likely just an non-fully convergeable combo or whatever. So a 5% or just options as a threshold would be extremely useful for those who use Play against solution a lot. Thank you in advance.

Eike

The problem with rounding is that we would be moving from one solution that’s difficult for the human mind to interpret, to a different solution that’s still too difficult to interpret. The only difference is that the latter solution is much less accurate. When looking at solutions, I would recommend looking at hand values (top pair, set, flushdraw, etc) and see if you can find reasons why certain strategies are chosen. Understanding the motivations behind certain strategies is far easier to translate into something that can be used in live play.

Geometric sizing has been added for v134.
Some sort of threshold feature is being considered.
It's not necessarily technically difficult, but the challenge here is exactly which approach to use.

Scylla when come out v134?

I don't think using a threshold to say a mix was an error is the correct approach. When it comes to mixing frequencies errors we really need to be looking at it over a large sample of hands and comparing our frequencies to overall GTO frequency for each hand category. If after I played 1000 hands the software would tell me I tend to bet my top pairs 80% of the time but GTO is 70% then I would have much more useful feedback.

In general the game is more about getting the broad frequency strokes correct rather than getting in the weeds about what exact combo is preferred in any one individual hand.

Yeah, but some combos have a very low frequency like 0.3% in the solution that could be a pure action if it was solved to a lower dev, this can cause you to be taking actions with a high frequency of the time with that combo and the solver doesn't advert you about this very low frequency that could be due to dev and consider your play as correct even if you take that action 100% of the time, I known on GTO mixed actions have the same ev, but I talking about extremes that could be caused because of the Nash distance of the solution, like 3% or less idk, a treeshold for this situations would be good

Enviado de meu motorola one hyper usando o Tapatalk

Basically, in summary, it's not entirely straightforward what to do here. At the moment we offer the option to press CTRL and quickview the solution prior to making your own selection. Some people have requested the ability to set a distribution with a slider bar. To me this seems very cumbersome to work with, as well as being overly focussed on single combos, as opposed to the entire range (if you want to use this approach, then it seems a lot easier to just visualize what you're looking to do, and checking your assumptions by pressing CTRL). A threshold has some merit, but has some disadvantages as well. We're still trying to figure out if we're adding more functionality, and exactly what this would be.

I would prefer not to provide release schedules.
There's often delays, and I would rather not set expectations, and then fail to meet them.

Hello everybody!
a general question, I play MTT, and for parsing the hands I purchased flopzillaPro and GTO +.
The question arose what data should I take from the distribution and transfer it to the SIit & GO tables in order to get an accurate calculation?



what value of the stack should I put in the chip or in the big blind, and what value should I set for the prizes if this is not the final table !?

For accurate ICM conversions the entire player pool (not just the active players, but all the players that are in the tournament) is needed, as well as all the prizes. So for an MTT this could be a massive amount of data to enter. This is one of the reasons why we don't offer MTT's; it's just not practical to enter this much data. If you're far away from the money, then it should be a resonable approximation to use "Cash game" mode. And for the final play the Sit&Go tab will be sufficient. We may add more extensive MTT options for later releases, but right at this moment we only offer "Cash game" and "Sit&Go".

I created a tree with some type of pot and stacks. I started developing the tree via edit tree. Now I want to resize the pot and stacks. How can I do this?
I tried to do it by rebuild, but I get a message: Pot had to be re-adjusted in "Stack" window.

Do I need to change the size bet throughout the tree? What does it depend on? If I give a percentage bet, will the problem also appear?

Hi Scylla,

Are you aware of anything that might cause crashes, particularly on the latest W10 updates?
I have 2 Ryzen systems, both experiencing crashes when running databases in "Process files in temp folder" mode, always same thing happens, cpu fan revs up to max for a couple of secs, then "no display" error and I have to restart.
I've tried to narrow down any driver issues, but everything is up to date and one of the systems is on a fresh W10 install, just updates and chipset/gpu drivers. Cpu/board temps are fine. If it helps, running on AMD's own balanced power plan. When I run on 14/16 threads it seems to crash less often but still happens, sometimes it lasts 30min, sometimes several hours but keeps crashing.
I'm on the latest gto+ version.

Thank you.

I'm not aware of anything, but I'll check if I can reproduce the issue.
At the very least set an exception in your virus scanner, just in case it's interfering with the software.

If either the pot or stacks have changed, then it's not possible to build the exact same tree.
This is because players would be all-in at a different part of the hand.

Hi scylla

What a cool software man, nice job !
Is there a way to make the 13x13 grid bigger ?

Resizing of individual elements is currently not available.
We can only make the grid bigger if something else is made smaller.
So it's a bit of a challenge.

I created a tree and started adding more flops. The first three flops did not resolve with a "not enough memory" message and the next one started to resolve, why?

Thank you! got it! I have such a question, I will give you preflop solvers, like Simple preflop?

Can you set the program to save the data and shut down the computer after solving?

We're not really that interested in preflop solvers, given that those require huge resources, which are out of reach of most of our users. We prefer to focus our efforts on features that can be used by everyone.

Can you check how much RAM is available on your system?
This is displayed in the lower left of the screen.



Then, load the tree that you want to solve, go to the editor and click on "IMPORT TREE".
This will show how much RAM is required for this tree.

It's possible to save after every solved tree.
For this, place the files that you want to solve into a directory.
Then go to the "Folder" icon and process all files in the directory.
The savefile will be updated after every solved tree.

We can however not offer a feature that shuts down your computer.

I looked it up. It turned out that the difference for the AQJ raindow flop and two monotone for example is 16gb. This surprised me!

edit: There are very large differences in ram requirements between some flops. Is there any list of which are the most demanding?