Open Side Menu Go to the Top
Register
MonkerSolver MonkerSolver

08-15-2018 , 07:21 AM
Quote:
Originally Posted by brasil270
Thanks, so i cannot isolate spots...So for this scenario how would you go about comparing the ev's?

I can do the full sim of bb capping and cbetting and another w/ him checking. Then I can do bb flat all to sb 3bet and the ev when sb checks and when it cbets 100% ---> the problem is when bb flats then the pot size stays smaller...can I still just look at bb/100 here to see what is best?
You can compare the EVs of playing the two pure strategies here (100% flat vs 100% cap). The resulting EVs will be representative of the two options. You can't exclude a live player in a multi-way game tree if you want to know the EV for a multi-way game.
MonkerSolver Quote
08-15-2018 , 11:22 AM
for 2/3 blind structure, if i input 20,30 for blinds in lhe, everything seems fine, but then the raise pf is the sb amount of 2 instead of 3 units, how do i alter this to make the raise 3?
MonkerSolver Quote
08-16-2018 , 04:43 PM
More on limit trees. When you make say a limit tree with maximum number of bets 3 I have trouble building even the tree. There are committal percentages given as default. If you don't change them at some point on turn or river it doesn't allow you to have cap of 3 bets but instead one or two if the pot was large when reaching turn or river.

I tried to tackle this problem by increasing the stack sizes (remember it's limit) so the committal zone is not reached and also another way was to change the committal percentages. Monker got stuck in both of these cases and it should not be a memory issue, the trees require one tenth of memory available.
MonkerSolver Quote
08-16-2018 , 08:22 PM
How is it for plo heads up?
MonkerSolver Quote
08-16-2018 , 10:06 PM
I’m curious what kind of CPU I would need to build to get Monker’s input on simple but massively multiway postflop PLO spots, the kind that come up commonly in live poker?

Example: In a six-way single-raised pot, the flop is Q72 all hearts and the SB donks for pot. It folds to me on the button and my continue range is…? How much RAM would I need to get a question like this answered in, I don’t know, less than a day?
MonkerSolver Quote
08-17-2018 , 03:46 AM
Quote:
Originally Posted by brasil270
for 2/3 blind structure, if i input 20,30 for blinds in lhe, everything seems fine, but then the raise pf is the sb amount of 2 instead of 3 units, how do i alter this to make the raise 3?
It seems there isn't a way to tell the software to adjust to these blinds automatically. For blinds with this structure in limit, you're probably going to have to build a custom pre and postflop tree to get the bets that you want.

Quote:
Originally Posted by PlumpSquirrel
More on limit trees. When you make say a limit tree with maximum number of bets 3 I have trouble building even the tree. There are committal percentages given as default. If you don't change them at some point on turn or river it doesn't allow you to have cap of 3 bets but instead one or two if the pot was large when reaching turn or river.

I tried to tackle this problem by increasing the stack sizes (remember it's limit) so the committal zone is not reached and also another way was to change the committal percentages. Monker got stuck in both of these cases and it should not be a memory issue, the trees require one tenth of memory available.
What do you mean by got stuck? How big of a tree are you trying to build? Also, if you can't build the tree, how do you know how much RAM it takes?

Quote:
Originally Posted by doctorflush
How is it for plo heads up?
Works well for HUPLO.

Quote:
Originally Posted by Leslieweizen
I’m curious what kind of CPU I would need to build to get Monker’s input on simple but massively multiway postflop PLO spots, the kind that come up commonly in live poker?

Example: In a six-way single-raised pot, the flop is Q72 all hearts and the SB donks for pot. It folds to me on the button and my continue range is…? How much RAM would I need to get a question like this answered in, I don’t know, less than a day?
RAM is not the primary bottleneck in terms of how fast you can solve a game tree. Your CPU setup will have a more significant impact. Stack depth is quite important here in terms of how long something like this might take to solve. I estimate that for say an SPR of 4+, this would take a lot longer than a day, and require a significant amount of RAM (500g +).
MonkerSolver Quote
08-17-2018 , 04:07 PM
Quote:
Originally Posted by IsaacAsimov
It seems there isn't a way to tell the software to adjust to these blinds automatically. For blinds with this structure in limit, you're probably going to have to build a custom pre and postflop tree to get the bets that you want.


thanks, preflop is easy by doing custom blinds (20,30) but for postflop it just raises in incremements of 2 instead of 3. is there a way to build the tree w/ that amount built in per street like all flop bets are 3, all turn bets are 6? Or do you have to build each individual branch?
MonkerSolver Quote
08-17-2018 , 04:46 PM
Quote:
Originally Posted by IsaacAsimov
What do you mean by got stuck? How big of a tree are you trying to build? Also, if you can't build the tree, how do you know how much RAM it takes?
It takes forever to try to create the tree, task manager says memory usage goes up little by little with no upper boundary it seems and eventually the software becomes unresponsive. The trees were at first in 10-50GB range and worked just fine, when you make stack size larger even by 10% the tree can no longer be built and interpolating from previous runs the tree size seems to increase roughly linearly by stack size.
MonkerSolver Quote
08-17-2018 , 09:06 PM
Quote:
Originally Posted by brasil270
thanks, preflop is easy by doing custom blinds (20,30) but for postflop it just raises in incremements of 2 instead of 3. is there a way to build the tree w/ that amount built in per street like all flop bets are 3, all turn bets are 6? Or do you have to build each individual branch?
Preflop the raise is also by 2. Are you seeing a preflop raise by 3 when you set custom blinds to 20, 30?

Quote:
Originally Posted by PlumpSquirrel
It takes forever to try to create the tree, task manager says memory usage goes up little by little with no upper boundary it seems and eventually the software becomes unresponsive. The trees were at first in 10-50GB range and worked just fine, when you make stack size larger even by 10% the tree can no longer be built and interpolating from previous runs the tree size seems to increase roughly linearly by stack size.
What I meant by how big was the tree was for how many players are you trying to build a tree?
MonkerSolver Quote
08-17-2018 , 10:36 PM
approx how much ram do i need to run 3 or 4 handed postflop NL situations? 100bb single-raised pot, 3b pots, etc?
MonkerSolver Quote
08-18-2018 , 12:10 AM
Quote:
Originally Posted by RickOSU
approx how much ram do i need to run 3 or 4 handed postflop NL situations? 100bb single-raised pot, 3b pots, etc?
For fairly simplistic trees, 128GB of RAM should be enough to cover all spots.

Quote:
Originally Posted by PlumpSquirrel
More on limit trees. When you make say a limit tree with maximum number of bets 3 I have trouble building even the tree. There are committal percentages given as default. If you don't change them at some point on turn or river it doesn't allow you to have cap of 3 bets but instead one or two if the pot was large when reaching turn or river.

I tried to tackle this problem by increasing the stack sizes (remember it's limit) so the committal zone is not reached and also another way was to change the committal percentages. Monker got stuck in both of these cases and it should not be a memory issue, the trees require one tenth of memory available.
Post a screenshot of your 'Settings' from the 'Tree' tab.
MonkerSolver Quote
08-18-2018 , 12:35 AM
Just bought the 2990wx. Tested 62 threads vs a friend using 6 threads. We were solving the same tree at roughly the same speed. I'm getting around 40% utilization vs 100% using PIO or GTO+. Is the scaling really that bad or am I doing something wrong?
MonkerSolver Quote
08-18-2018 , 08:55 AM
Is it possible to run Explo preflop simulation for MTTs, with custom defined defending ranges?
MonkerSolver Quote
08-18-2018 , 12:17 PM
Hey, I very recently purchased Monkersolver and started renting my own server in order to run it there, but I have 0 clue how the program works.
Would there be any experienced Monkersolver user in this thread willing to offer help on how to use the program? (for a reasonable hourly rate)
Would be happy about any recommendations

(Or also happy to join a study group in case sth like that exists)
MonkerSolver Quote
08-18-2018 , 06:39 PM
Quote:
Originally Posted by IsaacAsimov
Preflop the raise is also by 2. Are you seeing a preflop raise by 3 when you set custom blinds to 20, 30?



What I meant by how big was the tree was for how many players are you trying to build a tree?
no i am seeing it to "2", all i need is for each raise to be 3 and it will be completely set, but I am unable to do that!
MonkerSolver Quote
08-18-2018 , 06:53 PM
i have 64gb of ram, would that not be enough to run 100bb single raised pots 3 ways, 4 ways? how about 3bet and 4bet pots?
MonkerSolver Quote
08-19-2018 , 07:07 AM
Quote:
Originally Posted by EricDarland
Just bought the 2990wx. Tested 62 threads vs a friend using 6 threads. We were solving the same tree at roughly the same speed. I'm getting around 40% utilization vs 100% using PIO or GTO+. Is the scaling really that bad or am I doing something wrong?
You should be able to see more than 40% utilization imo. At least on any server I use with a high thread count, that's what I see. All of my servers run on Linux distributions (Debian 9 or Ubuntu 18.04) and my utilization is 100% with various Xeons. It would be interesting to see if you see the same issue with Linux. You could loosely follow this guide to setup Monker with Linux.

Otherwise I'm not quite sure what the issue is at a glance, it could be some other OS limitation in Windows for example. You can add me on Skype if you like (check PM), and I'll see if I can give you some more useful input.

Quote:
Originally Posted by RollinBoy
Is it possible to run Explo preflop simulation for MTTs, with custom defined defending ranges?
Yes.

Quote:
Originally Posted by brasil270
no i am seeing it to "2", all i need is for each raise to be 3 and it will be completely set, but I am unable to do that!
Yes, I took a look at it, and it doesn't seem that a raise by 3 can be used in this case. It should be a raise by 3 for a limit tree with a BB of 3, but the solver doesn't seem to adjust properly adjust to the custom blinds for limit trees.

When the solver was initially released, there were no custom blinds, so this was probably just an oversight to correctly expand the feature to LHE, as this isn't a case that would have likely come up at the time.

Quote:
Originally Posted by RickOSU
i have 64gb of ram, would that not be enough to run 100bb single raised pots 3 ways, 4 ways? how about 3bet and 4bet pots?
You didn't mention for which format. For NLHE I would say that some simplified trees could be made yes. 3 and 4-bet pots shouldn't be a problem.

For experimenting with different hardware configurations, I would recommend using a test server with Google for example. Check this guide out to see how to do that.

Quote:
Originally Posted by OMG_IM_SEXY
Hey, I very recently purchased Monkersolver and started renting my own server in order to run it there, but I have 0 clue how the program works.
Would there be any experienced Monkersolver user in this thread willing to offer help on how to use the program? (for a reasonable hourly rate)
Would be happy about any recommendations

(Or also happy to join a study group in case sth like that exists)
I can add you to the Skype group, I've sent you a PM with my Skype.
MonkerSolver Quote
08-19-2018 , 03:58 PM
How many iterations per node should I wait for solving 3way 3bet pot (7gb). Is the iteration per node the indicator that I should look for or another one? Thanks.
MonkerSolver Quote
08-19-2018 , 04:31 PM
trying to buy the software but it has declined two cards that have been previously working perfectly for me.. anyone else having payment problems? are there any other ways to pay than the "store"?
MonkerSolver Quote
08-19-2018 , 10:06 PM


for whatever reason monker is showing that I only have 0.44gb of ram available, but my task manager is showing 28gb. What do I need to do to resolve this? I have my .ini file set to -xmx16g
MonkerSolver Quote
08-20-2018 , 12:21 AM
Quote:
Originally Posted by RickOSU


for whatever reason monker is showing that I only have 0.44gb of ram available, but my task manager is showing 28gb. What do I need to do to resolve this? I have my .ini file set to -xmx16g
fixed it, had to reset my java settings in system settings.
MonkerSolver Quote
08-20-2018 , 10:03 AM
Quote:
Originally Posted by reXne
How many iterations per node should I wait for solving 3way 3bet pot (7gb). Is the iteration per node the indicator that I should look for or another one? Thanks.
Basic accuracy starts at 10 I/N. Don't forget to reset the average. For postflop sims, I think you will see good results at 15-20 I/N total, and reset at around 10 I/N.

Quote:
Originally Posted by RickOSU
trying to buy the software but it has declined two cards that have been previously working perfectly for me.. anyone else having payment problems? are there any other ways to pay than the "store"?
Try contacting your bank. It's likely they are declining the transaction for security reasons.
MonkerSolver Quote
08-20-2018 , 12:52 PM
Quote:
Originally Posted by IsaacAsimov
Basic accuracy starts at 10 I/N. Don't forget to reset the average. For postflop sims, I think you will see good results at 15-20 I/N total, and reset at around 10 I/N.
Thanks for the answer! What exactly does reset average do?
MonkerSolver Quote
08-25-2018 , 06:01 AM
Quote:
Originally Posted by reXne
Thanks for the answer! What exactly does reset average do?
At any given time, the strategy displayed to you by the solver is an average of all past iterations of the solve up until your last reset. It is recommended to reset once around half way through your intended solve target (in I/N) to remove early results that may skew the average, because those early strategies are near random.
MonkerSolver Quote
08-27-2018 , 06:39 AM
did anybody buy ranges on solveroptimized.com? their support suggest me to buy one hour ''how to use monker'' instead of answer how to open that ranges in monker lol
btw how to exclude some combos in monker ranges?
MonkerSolver Quote

      
m