Gotcha i understand that you don't support CREV was just wondering why i kept getting different frequencies as i thought in general they should all be relatively the same solutions.




---

The accuracy in both is set to 0.5% dEV, in regards to your first point i don't have a 4! range in the CREV model but i do have a 4! range in the pio model would that effect the tree/frequencies that much by like 4%?

Also do you mind explaining the last point, so in general ur saying that piosolver will be more accurate then CREV in regards to certain hands and is a more accurate solver in general?

It looks like you don't understand afterall. We don't support CREV, we don't know why it gives results it gives, we have never used it our lives nor we plan to.
There is nothing I can tell you about results given by CREV so there is 0 reason to paste them in our support thread. Go ahead and paste them in CREV support thread and ask them why results are the way they are.

As I've never used CREV I have no idea how accurate their solutions are.

Hi punter, I've a problem.

Board Kd 8s 2s

Range OOP:


I've locked bet all combo flop:


Turn Jc
If I want to lock turn betting all combo again, Pio automatically moves the combo with Jh in the check range.
Instead, if Turn is Jh Pio automatically moves the combo with Jc in the check range.




I tried everything, even to make the check range lock at 0%, but the result is always the same: combo with Jh moves in check range
Where am I wrong?

This is a known bug in node-locking tool and isomorphisms. In short it doesn't work on some cards correctly. In your example it would work on Jh turn instead of Jc one. The solver only keeps one of the branches in memory because they are strategically the same. I wrote a bit more about it including the workaround, here:

https://forumserver.twoplustwo.com/s...postcount=2413

(last part of the post)

im having this error when trying to run aggregate analysis report on a specific script folder results:



reports on other solutions are fine

PLOcalc- how many licenses for one purchase? (on how many pc's i can run it simultaneously)

Usual reasons:

1)You have some other save/script in that folder (make sure there isn't any)
2)One of the saves is malformed (the solver fails to load it), you can see if it always stop on the same flop
3)You are running older version than 1.9.2.5 (go to About in the top menu and check), if that's the case please run the updater with the "force update" option and if that doesn't help please email us at support@piosolver.com including your key and/or registered email

2 computers.

Hi, I know I'm not the first to ask, but I still don't understand :

What is the difference between the basic and the pro ? Is it possible to explain it clearly for someone who is really bad with computers ?

More specifically, under 'limitations', I see 'uses up to 16CPU cores (16 hardware threads)' What does that mean ? Does that mean my computer may not be strong enough ?

Under the basic version, I read 'uses up to 4 threads' again same question, or does that mean there is a limitation of usage ? and I also read 'not scriptable' and I wonder wtf does that mean and is it possible to explain it ? I guess it means ... I don't want to guess...

Thanks!

HU preflop 75bb spot. Locked whole OOP strategy (third image) to overfold 8% compared to GTO and kept the 3B and calling ranges as they were to see how the IP open raise range would change (first image). The obvious guess is that it would increase its open raise frequency, but for some reason it decreased. Compared the EVs of opening vs folding with the cutoff hands, and hands for which open raising clearly has higher EV are mostly folding (second image). It's clearly not a matter of balance here, so I'm having trouble coming up with an explanation other than that the solver is in some way faulty.

With the pro version you get:

1)an option to run it on 2 computers (instead of only 1)
2)an option to run scripts (those are useful to automate tasks like solving many trees overnight when you are sleeping)
3)support for bigger CPUs - this doesn't matter on laptops or low-mid level desktops but matters if you have top of the line CPU (with 6 physical cores pro version is about 15% faster, with 8 and more the difference becomes very big)

No. It only means that if you have really powerful computer then the basic version may not be able to use all its computing power when solving.

It's impossible to comment without seeing the whole tree (we need a config) and exact ranges you locked (we need those ranges). Preferably send them to support@piosolver.com or paste to pastebin.com and link here (alternatively paste here directly useing [ code ] [ /code ] tags (without spaces inside). Once we have those we can try re-running the tree ourselves and see what's happening.

A quick question is if you use rake. If yes then it's natural that the solutions are different and may seem randomly difference as once the rake is included the game is not 0 sum anymore and have many equilibriums. This means the solver may converge to one of them or the other depending on starting point. This is not a limitation of the solver but just how the math is.

If you locked the hands after solving it may be impossible for the preflop solver to "come back". It's the best idea to build a new tree, lock the ranges and only then solve from the starting point. The reason for this behavior is that the solver remembers how far it is into solving process and locking doesn't reset it. It will show correct exploitability of course but the difference between calling and folding borderline hands is very small so it won't influence it much.

In fact that is what I did. I first solved the situation with no locking, opened a new Pio window, entered the same situation, then locked the OOP strategy.

Hi, I don't know if this has been asked before - but I want to buy a high end pc to run piosolver sims (including preflop).
Have a question regarding CPU and graphic card:
1. Does pio use graphic cards for processing (i think the technology is called "cuda")? If so is this signifcant for performance?
2. Processor: what's recommended between the following (or do you have a different recommendation):
Intel Core i9-7900X
INTEL® CORE™ i9-7960X
INTEL® CORE™ i9-7980XE EXTREME
AMD R9 1950

Computer going to be used for running pre/postflop sims, mostly for heads up, including deepstacked sims.

It doesn't.

Speed is more or less proportional to number of physical cores * frequency they run on. AMD cores are mayb 5%-10% slower.
Some of our users have problems with memory stability with AMD CPUs so it's recommended not to overclock them and to make sure you get fitting components.

Config:
https://pastebin.com/ghSFtpgn

OOP locked ranges:

3B:
https://pastebin.com/v6XFRgDr

Call:
https://pastebin.com/8uj28jpe

Fold:
https://pastebin.com/nCA1axmq

Hi, I purchased the Edge version today and when I click on 'connect' and the option 'PioSOLVER-edge19AVX' like it shows on the video, it says : 'Failed to connect'.

What should I do ? Please note I know almost nothing aboutt computers.

I also want to say that it works with the option edge 1.9.2... And the error code with 19AVX does not say the key is incorrect and does not say a code of error like in the installation video.

So because I have access to the versions edge 1.9.2 but not AVX, I am wondering what is the real difference between the 2, in the installation video they seem to say AVX is better. I also wonder how could I make it work ?

Thank you!

Also I tried to build a tree and it says memory needed : 12944MB. Does this mean I must buy another computer ? I tried in on a windows Lenovo but I also have a Mac which should be very good computer. I have the Macbook pro. Maybe I should try it on the Mac with Bootcamp ?

Thanks, it will take a while to run it and take a look.

It looks like either your computer or your operating system doesn't support AVX instructions. Those are advanced instructions of new CPUs. Older CPUs or operating systems don't support them.

AVX version is about 15% faster on computers which support AVX instructions.
You need both a modern CPU and an up to date version of Windows.

It's always possible to construct a bigger tree so no matter what computer you have there is always a tree you won't be able to build. The bigger the tree the more RAM it needs. Reasonable postflop trees (even with many bet sizes) are usually below 16GB unless you really go wild with options. If you want to build bigger trees you need a computer with more RAM but also probably a faster CPU as they will take a long time to solve (the bigger the tree the longer it takes to solve).

I wrote a bit about hardware requirement in the FAQ, here:
https://www.piosolver.com/pages/faq#hardware

You see, this is a problem. It seems (from questions here but also your email which I am going to respond to in 10 minutes) you expect everything to be clear to you even though you don't know anything about computers. Sadly it's difficult because you are asking specific question which I am responding to and then you are surprised by something entirely different and frustrated that my answer didn't include information about it.

It's both my duty to try to guess what you meant/need and your duty to try to be as clear as possible. You will always be frustrated when getting answers about the topic you (by your own admission) not know anything about. It's not the person's who answer the question fault. There are two things you can do about it: either learn a bit about computers or accept that communication about those topics is going to be difficult/frustrating for you even though everyone is doing their best.

I've run your trees on 74 flops both locked and without a lock and the results are as expected:

1)opening range without a lock:
https://gyazo.com/7c16ea0e34f418cf852ffba9afc03e06

2)when locking your ranges:
https://gyazo.com/e99ce9e2e45cd99e80beff138a355ce4

Please notice that I am running a dev version which handles multiple bet sizes postflop in preflop trees way better than the current one. You included a lot of overbets and 2 sizes on the flop. This is likely the source of the problem. The current preflop solver doesn't really handle those well because of memory saving which is necessary to make the trees smaller.

Cool. So then this will be what Pio 2.0 will be capable of?

Hi, I tried all day to download myself the windows on my Mac. I still couldn't do it. Maybe I was stupid try to do it from a download. Tomorrow I plan to go apple store to buy it & hire the technician. But if I do this, do you think my Mac will be good enough to solve enough ? I have 16 go memory, and 209,1 go available.

You wrote before the pre flop solver was very demanding.

Because if my mac is not good enough, I will just buy another Pc instead of hiring the technician to put windows in my mac.

But if I must buy another pc, what specification should I ask? I would just like to ask the tree a few bet sizes per street with donks & raise. And I still did not have the chance to see what the pre flop solver looks like.

There will be 1.10 version which is going to be a free update. It's getting delayed because I have problems with implementation of some features.

It's likely won't be enough for trees you want unless you go "run Windows on Mac instead of MacOS) route then it will be borderline for a tree you tried (around 13GB).


You need at least 64GB of RAM to run the preflop solver, fast modern CPU is recommended as well because preflop trees take a long time to solve (it's recommended to get 6+ core CPU, it depends on the budget and expectations though).

It depends on budget and expectations. If you want to run the preflop solver then 64GB of RAM is recommended, some of our users went with 128GB but those are very expensive machines.
With CPUs: the more you pay, the faster it will be although at some point it's diminishing returns. Popular choices are Ryzen 1700x-1800x, ThreadRipper or top of the line Intel CPUs. With new AMD CPUs you get lower prices but you need someone knowledgeable to assemble it as those systems are often unstable if RAM/CPU are not matched correctly (or if they are overclocked).

In general it's hard to say what is "recommended" as there is always a bigger tree to build and you can always solve it faster. My personal view is that 16GB is plenty for trees even with multiple bet sizes but then it seems people want 3-4-5 bet sizes per street, deep stacks and it's not enough anymore.

Hi,

I'd like to study spots (6max, 100bb deep) where PioSolver leads the turn a decent % of his range after check-calling the Flop.

I can't find many spots. I haven't found any in fact.

Any idea which kind of spots (in terms of board / turn card type / preflop ranges type maybe...) I should look for?

Basically, do you know in which circumstances Piosolver leads the Turn OOP after XCalling the flop?


Thank you !

In general it's safe to always check after check-calling flop. Even if that's sometimes chosen in optimal solutions there is little to none EV difference. It' similar to leading the flop after calling preflop - you basically never do that.

That being said there are some spots where Pio likes to lead turn. The circumstances I am aware of combine two points:

-the card which "reverses" equities in favor of OOP players. Those are usually cards which pair the board with high/middle card. For example something like JT2 flop and T turn
-small bet sizes available to OOP

Thank you!
What about the river after Xcalling the turn? is it the same?