oh and I just had this idea of doing something like freehand selection in range explorer, instead of using the check boxes to tick the type of hands we want to analyse why not have a button that will display a simple range selector window, so we manually select the combos to show in the range explorer for example if I wanted to see pocket pairs between mid and botom card in K95 it can be messi to create the cathegory in range explorer but if I could just manually select the exact combos of 66-77-88 that'd make the range explorer way more powerful, and this way users can combine the stuff anyway they want if I want to display something like overpairs, trips and nut flush draw it's very easy to do, the only thing is that in the range selector the hands that are not in the range will be grayed out in the same way they are in the preflop charts so that we can not choose to display hands that are not in the ranges at that point.

hope that makes sense, if it sounds too messy I can try to clarify :-P

tx

This is already available.
There is "Selection" clickable range selector in the Range Explorer. You can choose the hands to display manually there. You can also first use the checkboxes and then use "Selection" to remove/add hands. Try it

hi, I think you missed this post :-P

wow awesome!! Jesus I've been using this software for ever I can't belive I missed that, it works as I expected haha just one thing, can you make the hands that are not in the range be grayed out so they can not be selected just like in the preflop charts

tx again!

Hi, what "Save Script (with currently selected set of flops)" button in Preflop tab does exactly? Can I make a lot of different pre flop sims, save to script and run it like post flop one?

It just saves the script to run a preflop tree to file. You could theoretically chain the scripts (see the Skype group FAQ here: http://piosolver.myshopify.com/pages/faq-videos) to run many preflop sims in a row but this is not recommended (it's way faster to kill the solver between executions than dispose of the tree and build a new one).

We are currently experiencing some problems with the licensing server. Apparently it affects users located in Australia. As we don't control the licensing server it may take a while to resolve. I am trying to get someone to help me debug it on Skype and get offline activations working. I am sorry for those problems as I realize they are annoying. Thanks for your patience!

(The person who asked this questions meant 16 core Xeon machine clocked at 3.1Ghz).

Well, there are trees and trees. "A tree" doesn't really say much.

Here are some guidelines you may find useful:

-PioSOLVER scales more or less linearly with number of cores, there is just a small performance penalty for multi core systems. For example 16 cores give about 15.2x speed-up over 1 thread on 1 core and 24 cores give about 22.5x speed-up. I haven't tested on 32 cores but I would expect a factor similar to 24 cores, maybe a bit worse.

-the newer generation of CPUs is better although there isn't that much difference between the newest ones

-you can use number of cores x frequency to approximate performance (then subtract a bit if it's really a lot of cores). For example 5820k which is a 6 core CPU clocked at 3.3Ghz gets 6 * 3.3 = 19.8 while 16 core system clocked at 2.6Ghz gives 41.6.

Notice that "32 vCores" mean 16 cores (marketting mambo jambo, it really is 16). That means that this system gets 16*3.1 = 49.6 or about 2.5x faster than 5820k i7 (it would be closer to 2.2/2.3x in practice because of not exactly linear scaling).

Also notice that you need an edge version to get maximum out of those big dual/quad CPU machines.

EDIT: it's really hard to beat overclocked 5820k significantly and that's what I recommend people buy for home use. I think it's more efficient to rent if you want a dual CPU setup unless you are a hardware enthusiast who is good at fishing for cheap parts on e-bay. Xeons are in general very overpriced in comparison to i7s and their only advantage (for the solver) is that you can put 2 of them on one mother board. It's tough to find a dedicated server more than 2x faster than overclocked 5820k (unless you are ready to pony up like 500$/month). i7s are also better for home use than Xeons (higher core speed means better performance for things which aren't multithreaded).

hi punter, a small feature request, can you add the browser, I know that I could just mentally rest all bets from the starting stack but it gets trickier in river spots when deep stacked after many bets raises and calls...


https://gyazo.com/bbc177cdda2c549d9304fd1b6a51ac1f

tx

To be honest I have no idea what you are suggesting here

Does Piosolver work on a Mac?

It doesn't work on MacOS, you need a way to run Windows. The most popular solution among our customers who use Mac to run Pio is either Parallels or Bootcamp.

I get "is there a way to lock the whole tree" question a lot so let me explain that there is no need for that.
For example if you want to achieve the above follow the steps:

1)solve the tree to decent accuracy
2)stop the solver and click "calculate results"
3)change your strategy (you can do it using node-locking interface but there is no need to lock anything)
4)click "calculate results" again

Compare the EVs/MESes from step 2) and 4) to get the result you want. Notice that you don't need to "resolve" anything. It would be counterproductive with one player not being able to change their strategy

Request for 1.9:

Please add another all-in threshold option for % of CURRENT stack. In fact this might be able to replace the current all-in threshold (for % of initial stack) altogether.

To give an example of why I think % current stack is better, suppose OOP bets turn large with a very polarized range then checks river. If IP bets, he will often want to choose a small sizing. But with the (% of initial stack) all-in threshold set, he may often only be able to check or bet all-in.

You might say, "Why not just disable the all-in threshold to solve this problem?" Well, the problem is that it's often useful for it to be implemented in certain parts of the tree but not in others. So I think a % of current stack all-in threshold is a better option all around.

There was discussion about it before. I think the current option is better. The problem with:

I think it's just not something real players use (even if they think they do). For example if you have already 900$ invested out of initial 975$ and now they bet 35$ (less than 50% of the current stack) they are very unlikley to fold for the last 40$.

On the other hand if you bet 500$ out of your 975$ and got raised all in (let's say the starting pot was 500$ here) you may consider folding now.

The "% of the initial stack" expresses exactly how people react: if the percentage is big the pot is already very big and there is little money behind (so that little money is likely to be added to the last made bet).

Okay, but your reply doesn't address the problem of bet sizes not being used in certain parts of the tree because the all-in threshold has been reached (sizings which would be used with significant or non-negligible frequencies otherwise). That to me is really what I don't like about it.

Having an option to disable the all-in threshold at certain nodes is another idea.

I think I've addressed it. It just seems unlikely that such spots exist and are impossible to take care of by raising the all-in threshold. What about posting a problematic config so maybe I could see the problem?

There are 200+ lines in the simplest of trees with one bet size, once you go into something a bit more complicated it's 1000+ of lines. Anything node specific is just a bad idea as it would take forever to insert every line.
In the next version there will be tools which make every tree possible from a simple config though.

"Hotness" doesn't work in the basic version, does it?

It does. Make sure you have version 1.8.2.3 of the viewer though (About in the top menu).

I have this version but when I click "Runouts EV comparison (hotness)" than absolutely nothing happens. I don't even know what should happen (report/new window or popup) but as long as nothing happens...

Edit: I didn't how it works. #gofigureit: I did it. Sorry

Hi,

I have not used PIO in a while. After I tried today I wasn't able to connect to the solver. I tried updating to the newest version but no luck.
Any suggestions what I can do to make it work again?

Yes. Email us with the information about what exactly happens (that includes exact error message you are getting) when you try to connect as well as your registration key. The most common reasons for the license stopping to work is Windows 10 upgrade or hardware changes.

Some questions I've got about hardware today:

Laptop and fast don't go together and if you try to make it work the price will be heavier/clunkier/uglier laptop and still not that much performance in comparison to your average modern desktop.

This is in general true. Additionally many (most?) laptops will have heating problems at some point so you will not avoid buying a cooling pad if you want to run it for many hours straight (or decreasing performance for it to not heat that much).

There is always option of laptop + rented server/cloud instance. In such case you can buy a shiny ultrabook and just connect to your server if you need to run Pio. It can be quite cheap if you are not after big dual CPU builds.

I am not a person who knows most about hardware. I am programming the solver and I test it in nice environment: a few years old i7 desktop and a rented 16 core server. Many problems with laptops just don't apply to me (I have an ultrabook as well but I never run the solver on it directly).

I expressed my view about laptops many times: they are slow and expensive. You can find big machines out there but they will still be slower than what you could get for like 50% of the money if you go with a desktop. My only recommendation is to go with real quad core CPU (which already cuts significant part of the market, for example all ultrabooks) and 16+GB of RAM.

You can easily test it using "estimate tree size button". Some examples:

-100bb, BTN vs BB, 40% and 70% bet sizes everywhere (75% and 150% on the river), one raise size (52%) is 7.8GB

-same tree but 135% bet size added on flop/turn: 12GB
-same tree but allin added everywhere if it's not bigger than 3x pot: 13.4GB
-same but in addition to all the mentioned options you also add a pot size raise everywhere: 19GB
-same, but this time ranges are 100%: 36GB

As you can see you really have to go compltely wild on the tree to get it bigger than 32GB after recent memory optimizations.

RAM is useful though. Maybe in the future there will be an option for the solver to use more RAM but make the calculations faster and more precise.

From benchmark I've seen likely it's going to be a huge difference, closer to 2x than to 1.5x.

5820k starts with 3.3Ghz, doesn't overheat if you spend 15$ on a decent fan and you can overclock it as well. Even without overclocking it's 3.3Ghz * 6cores = 19.8 speed units while 2.6 * 4 = 10.4. I get conflicting benchmarks so far so I am not sure if Skylake is significantly faster than Haswell per core at the same frequency but the difference is surely not that big.

Yes, for like half the price you can get 5820k + necessary things. You can consider renting a server as well. You can find a cheap server for 30-50E/month. For example 4 core Xeon @ 3.5Ghz, 32GB of RAM. It's going to be 500E/year and you don't need any new computers. It will always be more convenient as you can connect to it from your ultrabook and you can run it 24h/day without worrying about heat/electricity or maintenance.

I am putting my answers here (even though the question was asked by email) because I get questions about laptops a lot so it's convenient to put my thoughts on it in one place. Another good place to ask is our Skype group as people there have varieties of hardware and collectively way more experience than me.

Hi, I´ve bought the basic version of PIO SOLVER, but I think I want to have the PRO version. Is there a way to 'upgrade' or to return the basic and just pay the difference for the PRO? I bought yesterday, so it´s not like I already used it and want to change or something.
And btw, is there a big difference between the two versions?
Thanks in advance!

Just send us an email describing what you want. Include your key.

-pro version has 2 activations
-pro version works with up to 12 threads (it makes a difference on hexa core CPUs)
-pro version is scriptable (you can run/generate scripts to automate work)