Do the preflop solutions for 0 flops selected and 20 flops selected look very different for 150bb stacks?

If so, how much RAM/cpu power do you need to generate practical preflop solutions in Edge for cash games?

It's very hard to comment on anything without looking at actual tree. It's likely the numbers you look at means something different than you thin they mean (maybe they are in chips, not bb) but it's very hard to even start guessing without knowing what you are referring to. The best way is to:

1)Attach the config:
-use "copy to clipboard" button in the treebuilding and calculation tab
-paste the text:
-either to pastebin.com and link here
-here but enclose it in [ code ] [ /code ] tabs (without inside spaces)

2)Make a screenshot of the place you are looking at

This way I can have the same tree you have and look at the same place you are looking at

Yes, pure preflop solutions look very different than the ones with real postflop play.

64GB RAM for most practical cases (you may want to have more for big trees with multiple bet sizes). Fast modern CPU, hard to say what's a minimum but you want at least a decent hexa core.

Hi guys,

I'm new to PIO, should the area that I circled in blue not be filled with %
combos ?



It works fine with suited hands, but can't see anything when I hover offsuit hands.

It should be.
The reason it isn't is that there isn't enough space to display minimum size font. The way to fix it is to make the PioViewer window bigger by either dragging on the corners of the window or making it fullscreen

Much better indeed, thanks for your quick answer!

in the future, will it be possible to use the exported ranges from monkersolver (they're in ppt form) and import them to plo calc?

Yeah, it's under development. Monker can export Pio ranges, though:

I assume PLO Calc can read those? If so, I will buy it today.

We've done it for one of our testers and it's in dev right now.
I have one question though, the files we got from our testers look like this:

while Jeff W pasted this:

Which one is Monker's format?

Both are monker format options. JhJs3s2s@26 is PPT format. JcTdKs6s: 0.15425469110627946, is "Pio" format.

Yeah, that's the format we used for Holdem because in Holdem you only have 1326 hands in range so it makes sense to do a format which is both human readable and easy to parse/generate. In PLO there are 270725 hands so human readable concerns are out of the window and writing those combos everytime to file is quite painful (the files are bigger than 2mb per range even with 2 digit weights).

Anyway, we will accept both. Last question is in what form do you get those ranges?
Do you always save them to files or can you copy them to clipboard?

We implemented a converter which just converts the whole Monker's directory of ranges to our format but if you can copy them at given point without saving to file we will take care so you can paste them in Pio as well.

Im going to solve a bunch of trees with full flop node-locking.
What is more right thing to do: 1) build tree - solve - node lock - resolve and see result 2) build tree - node lock - solve and see result ?
Thanks.

The second option. It will be faster and more precise.

There is an option to copy them to clipboard.

Hey, sorry if that's already been asked before, but I'm buying a new processor and wonder what's the best (or close to) processor to make the calculations insanely fast?

Depends on your budget and expectations. The best deals right now are:

1)For a standard powerful build - Ryzen 1700/1800 CPUs
2)For a home work station - AMD Thread Ripper

2nd option is about 2x as fast as the first one.
One of our users (look at 2 last pages of this thread) overclocked the Thread Ripper to 3.9Ghz and got sub 2second Pio benchmark. This means it's 50% faster than for example dual Xeon 16 core server I am using. You can beat those times but not by much and it would require huge investment.

You can run Pio benchmark on your setup:
https://www.youtube.com/watch?v=5Ozjfnatwms (question number 2)

and see what times you are getting (preferably run it a few times, there is some variance in the results). The benchmark is very good approximation of speed so 2x faster benchmark means 2x faster solving times. If for example you get 10 seconds on your current computer you can hope to be 4x faster with good Thread Ripper build (without overclocking).

I did a small research about this feature but I couldn't find anything
I'm not sure if I understand it, so here are my questions;

How should I fill this field if I play... ?
1) NL 400
2) NL600
3) NL1K


What happend if I set 0 as it is shown in the picture ?

This is just so the solver can calculate bb/100 for displaying accuracy. Accuracy is in chips so it only affects what's displayed at the bottom (where chips are translated to bb/100) and when the solver stops if you put a condition in bb/100.
It will display nonsense in "exploitable for" message but will still work normally.

Ok, Thank you very much, I thought my simulation could be wrong If I set an incorrect number

The only danger is it stops too fast or goes too long if you give incorrect bb value and set the desired exploitability in bb/100 but you can always stop it manually at any point anyway so it's not a big problem.

how did u assign weights for flop textures in the notepad files we use for preflop calculations?

I am not sure what you are asking. Why would you want to assign weights to those flop? They are already weighted.

I am sorry, I misread the question (as "how do you...").

I've written a bit about the process here:
https://www.piosolver.com/blogs/news...the-whole-game

Since then we improved on it a bit and are now recommending using those subsets:
https://pastebin.com/AtyU84qC

out of the ones we ship.

If I start and Preflop Simulation, Can I press stop, "Save Full Tree" close PioSolver, load the tree, and press "Go" to resume the tree calculations?

I'm nearly sure that the answer is Yes, but I don't want to take the risk of loosing one night of simulations

It will work but it's not recommended because:

1)the trees are very big (as big as in RAM so that can be 50GB+ on your hard driver)
2)they are very slow to save/load
3)the preflop solver doesn't remember the state exactly when stopped so stopping/restarting won't be as precise as just running it in one go

I promise this is going to be my last question for a while
Which number do you recommend nowadays? (Preflop Solver, 74 spots)


Because I saw one video where you have 0.3 but in my simulation of a preflop spot of 74 boards its now very slow and the strategy its not changing too much