Yes, this is correct.

Yes, they don't know how the opponent is going to play within constraints of the tree (the ones you marked in green). While the algorithm itself uses current approximations to get close to the equilibrium the players still don't "know" in a sense that the resulting strategy is good against anything the opponent may choose. By "good" I mean that it guarantees the EV will be no worse if the opponent chooses a different strategy.

That is within limits of "exploitable for" number (which means that perfect adjustment against current approximation would win as much per hand). For example in your screenshot if a perfect all-knowing opponent would play against a current solution they would win 0.148 chips per hand which assuming this is 5/10 spot is 1.4bb/100. Of course this maximum exploitive strategy would be very exploitable in the first place and you shouldn't be afraid anyone is going to use it (because that requires knowing exactly how you mix your actions at every point in the tree) but it is theoretically possible to construct it.

Hi !

I want to run 2-3 small preflop trees in a row, during the night

I tryed this way

1) I saved preflop trees as scripts, using this button in preflop tab
https://gyazo.com/6b89c2436a4174a5871ec197f0b65dea

2) Then I made aggregator script
https://gyazo.com/c245f9f808702e6c6c2553b39a57b58b

3) But I am definitely doing something wrong, cause
https://gyazo.com/c27b2d66bed80b7e4f48a45b9b006205

a) It doesn't save results (very strange, cause I do same as here https://cdn.shopify.com/s/files/1/07...05195795588396
- "save" command goes right after "go" command)
b) Doesn't run second tree

What is the right way to do my question ?

Thanks in advance !

Unfortunately it looks like I made a mistake editing the sample script and as it's a very old example I haven't noticed that, thanks for pointing this out.

What you need is "wait_for_solver" command after go. The best way to see how the scripts are constructed is to use the "generate script" button in the postflop tree building and calculation tab. It goes like this:

(those are sample file paths from my computer, nothing to model your script on)

You don't need skip_if_done line. Set_accuracy should go before go. You can put it at the top of the script as it's needed only once (unless you want to change it between trees) as it's a global solver setting.
Please keep in mind that set_accuracy accepts values in chips not % of the pot. I recommend testing those things with very small/quick trees to not waste too much time. Once you see it works you can change it to solve serious trees.

Let me know if you need more help with it.

I run PIObasic today for the first time and i saw it only used 60% of the cpu something else called "service host: local system" was taking an extra 12%. (don't know if this is related to piosolver)

my question is, can i give more cpu power to piosolver so it will be faster?

This is not related to PioSOLVER. It's either some OS service like file indexing or some "online" protection thing which scans files/programs when they are running. In the latter case it's highly recommend to disable it as those things are detrimental to performance.

Basic version uses max of 6 threads so on a quad core i7 CPU it will use max 66%. That's only about 5%-8% slower than using 100% though (due to nature of hyperthreading, in short first 50% is almost all of it and another 50% is sometimes beneficial - it depends on the program).

punter11235, Thanks a lot - your comprehensive instructions did help !

Hi,

I like to buy a product, but I am not sure what is best for me.

I'm between the basic and the pro.

I'm only use it at one computer so the difference is the "not scriptable"

I don't understand this, what mean this?

The second question is: is the free version and the basic version using up to 4 threads and only the pro version is using up to 12 CPU cores (12 hardware threads)?
I am very happy whit the speed of the free version

Thank u very much

It means you can't automate solving of many flops for when you are away from your computer. There is no functional difference when it comes to features between basic and pro versions.

Basic version is actually compiled to use 6 threads these days so for quad cores CPUs it makes very little difference (it's the same for quad i5 and very minimally slower for quad i7).

Hi,

Thank u very much!!

Is a payment whit skrill possible?

Punter, I again would want to stress the efficiency/user-friendliness of a feature which highlights within an action region the lowest equity bluffs that are most incentivized to bluff vs fold because these are the most counterintuitive nuggets of info.

I.e. http://imgur.com/a/Wa4Qf

A4hh/A4cc and some non spade/diamond T8/98 are the best candidate to fill out are range besides all the obvious high equity bluff randomizers.

This would be very helpful if it would stand out within this -all yellow- grid landscape.

TY in advance!

The store accepts any debit/credit card (just don't choose PayPal) although our payment processor (BrainTree) refused some Skrill cards lately so it's hit or miss of it's going to work. Other than that check pm.

I am not 100% sure what you are doing but you can just go to raise 6 and then hit Equity OOP which in combination with "square size proporational to weight" checkbox will show you raising range with equities instead of only yellow squares. It seems to me that this is exactly what you want. Please correct me if I am getting it wrong.

Hello, do u know what is the safety %% for cpu and ram using? When my pio works - it takes 50% cpu and 60% of ram, is it safety for my laptop?

Hi punter,
Usually all my trees contains multiple betsizes/raises, is there any chance that, in the next update, the displayed bets/raises when browsing the tree to be shown as % of the pot (just like in the creation of the tree) and not like actual chips, because in big trees it gets difficult to figure out which size is which and i believe nobody should care for the actual size in chips - ie I play 200nl and use simulations on 1000nl like you suggested.

How much RAM it takes depends on how big a tree is. CPU usage dictated by number of threads used. The default number of hardware threads your CPU support but max 6 (in case of basic version). Neither has anything to do with safety although if your cooling system is not fully functional or you are using a laptop prone to overheating then not using 100% CPU is a good idea.

It won't change at least for now but as the total pot is visible all the time it's simple matter of division.

I disagree with you about sizings. I think everybody should care about sizing in chips/blinds/whatever units. Again, maybe this is not the most convenient as of now if you are used to thinking in terms of % at every point but you can see the pot here:

https://gyazo.com/abc085eb096855f0ee698a8b96915c9b

(in this case it's 206 with OOP having invested 36, IP 118 and 55 being starting pot) so even in big trees it shouldn't be too confusing what actual bet sizes are.

Hi,

I am studying some previously solved trees and am attempting to lock a strategy at a given node. However i am getting an error message saying i have an incomplete tree. I have simply loaded a previously solved tree and they are complete. Any idea?

***edit***

i have found that small saves do not comprise the whole tree - just flop and turn.

also i cannot re-solve from the turn onwards by clicking go in the build tree window for the same reason i presume.

maybe a feature that allows the tree config to remain present in the small save so at least we can make a tree complete again as it would be nice to be able to explore scripted solutions without killing a hard drive with full saves?

I agree with the user above. It would be nice to have %'s especially for people who play with increasing blind structures. Even if working out which sizing is which is doable mentally, a simple number in the box would make things so much easier and more efficient.

Hi,

I am trying free version on i7-6950x and piosolver process only uses about 35-40% of CPU power. Are there any performance limitations on a free version, or what else do I miss here?

Yes, it uses max 6 threads.

Hi,

I made several preflop solves and solving them with a script. Unfortunately there was one error, and some other scripts solve until some value for "Exploitable for:" and than remains around that value.

#1 https://gyazo.com/bdc8f3f1bf1c7b0132f340b6bc24054d
The error in this script came at the start of the second preflop solve. There where a total of 6 solves in that script, all with other ranges. But all with the same #threads and same accuracy.


#2 This one was in a other script. I had earlier also some problems with this. The script solves a solve until some value and than remains solving around that value. https://i.gyazo.com/6f87f92ea847e076...b344ec13d0.png

Accuracy for the solves in this script where 13.
Maybe the problem is that i had 13 instead of 13.0 here?:
https://gyazo.com/fc1c79f89b360947101c22f8e97b43a9
13/2625 = 0,004952380952381 = 0.05%. Fine accuracy right?

Ty

do piosolver close to opening icloud feature?
and how much will it cost??

It's very difficult to comment on anything without seeing the script itself.

It looks like you inserted an incorrect command or you were using a very bad text editor which broke the line when you were copy-pasting.

I am not able to tell what the error is looking at the screen. Everything looks correct.

Accuracy is given in chips and having 13 there is ok.
0.05% is very good, I wouldn't recommend solving to that accuracy as, especially for bigger trees, the solver will take a very long time to get there.

It's very unlikely we are going to provide any cloud related feature ourselves. You can however rent a dedicated server yourself and use Pio there as many people are already doing.
You can also buy preflop solutions calculated for weeks on many servers from our friend at PioCloud, here:

http://piocloud.******.com/

HI,

I have a question about the scripts. I would like to run a script with some boards, thats cool. Is it possible the solver choose the most recent betsize at the flop and run again with use only that size?

Thanks

I am not sure what you mean by most recent betsize so I am assuming you mean the most frequently chosen. This is not possible automatically, you would need to do some programming to generate a new script and run it.

Hi Punter,
thanks for answering the question regarding number of cores. I just wanted to make sure I can use all cores from a rented server.

I have got a couple of other questions.

First regarding rake for lower stakes, e.g. NL50 I used 5% rake and 60chip cap with 1000 chips. Is this correct to use because you can`t have 0.5 chips as SB or BB.

And when I run a preflop sim and then save the small/very small tree i can`t use them for nodelocking etc later on? But I will still be able to browse the strategy, EV and ranges right?
What would you recommend when someone wants to be able to use the runned simulations again? Should I just rerun them from the very small saves or still save the full tree.(e.g. when you use smaller preflop trees you can go down to 32-64gb of harddisk space, I think for postflop simulations you could just rerun it, because it doesn`t take like 10 hours to run them)

Basicly I am not sure, weither or not to download the big trees from my dedicated server.

Yes, 60 chip cap is 6bb in 5/10 and in 50NL 3$ is 6bb as well (0.5$ * 6) so that's the correct way.

Yes, as small saves don't have all river solutions (just cached EVs for flop/turn) any change in strategy would mean that EVs can no longer be shown. That means that things like rounding or node-locking won't work.

Yes, the solver recalculates missing parts of the tree automatically. If you have turns saved then the process is usually very fast (200ms more or less is expected) but if those are very small saves then ra-calculating turns may induce significant delay.

If you want to browse it - small saves are good.
If you want to node lock then you need to build the tree again, node lock, solve. It's not a problem though as node-locking on already saved tree wouldn't be faster than on a fresh one and sometimes the solver have big problems coming back from precise solution once you mess up with it by node-locking. I recommend nodelocking on fresh trees or on ones solved for just a little time.

I honestly can't see one reason to download full tree saves. Remember that to load a full tree save you need as much RAM as the original tree took (so 32-64GB in your case) while to load a small save you need barely any RAM.