There are no more memory restrictions.

If you have 8GB it's usually possible to buy another 8 (just make sure to buy a compatible one). If you already use 8GB of RAM that means you have 64bit system

Hello,

I am happy to announce a small PioSOLVER update. This is mainly a bugfix release with some nice features as well. More details: http://piosolver.com/blogs/news/33449540-piosolver-1-3

Highlights:

-preflop equity
-more features to make analyzing strategies easy (like copy-pasting as CSV)
-automatic script generator for easier scheduling

While we are already way ahead of the rest of the market in performance and features department we are still far away from the final goal. I hope to make another announcement next week bringing another step in that direction. Small spoiler is that it's about huge performance and memory efficiency improvements.

Stay tuned!

... punter building up the suspense again ... so bloody curious now !!!

If i buy piosolver pro, is it possible to later update to piosolver edge for the remaining cost? As in $1099-$475 (given no price changes).

Yes.
We don't plan changing the prices for foresable future. Of course such decisions are always taken taking into consideration what is going on on the market/tax regulations for specific countries/what gives us reasonable pay but the way things look right now the prices are likely to stay where they are which includes the upgrades.

I am happy to announce another PioSOLVER and PioViewer update. This one is small but quite exciting. We are announcing two updates:

1)Experimental PioSOLVER 1.4
It brings big performance/memory efficiency improvements.
The improvements are as follows (as measured on a not too fancy 3 years old i7 quad):

100% ranges: 1.2x speed, 1.2x memory
75% ranges: 1.42x speed, 1.55 memory
50% ranges: 1.8x speed, 2.2x memory
25% ranges: 3x speed, 3.9x memory
15% ranges: 4.5x speed, 5.7x memory
8% ranges: 7.2x speed, 8.5x memory

This means, some of the practical trees are now solvable well below 3 minutes and sometimes below 1 minute. For example: (all the measurements on i7-3770 quad)

a)Your typical 3bet pot with HU ranges 100bb: below two minutes to 0.4%

b)Your typical 100bb single raised pot with autocheck OOP on the flop: 2.5 minutes to 0.32%

c)CO/MP vs BB 3bet pot 100bb - 40 seconds to 0.4%

It goes without saying that exploitability numbers are for the tree you can browse instantly without any delays nor recalculation.
This version is available today for our edge users and will be available once it's feature-complete with stable interface for pro and basic version users in about 5-6 weeks.


2)PioSOLVER/PioViewer 1.3.2

This is mainly a bugfix release with some nice features as well.
Highlights:

-equity realization in the Range Explorer, GIF

-more Range Explorer features like ability to see strategy by category as well as quick comparison of range for RAISE/BET/CALL/FOLD

-ability to run scripts from PioViewer to make it even more convenient for scheduling calculations on many flops

The schedule is usual: edge version users have it already shipped. Pro and basic version users will get the 1.3.2 version tomorrow/in 2 days. Free version is likely to be updated in near future as well.

More details on our blog:
http://piosolver.com/blogs/news/3396...s-and-a-bugfix

Have fun!

Let me hype some features in approaching PioSOLVER 1.4. For today we have two:

1)Vertical browsing

Recent discussion about space/time trade-offs gave birth to this feature. In PioSOLVER you have access to the whole tree all the time which opens many possibilities like aggregation reports (one showing best cards to 2nd barrel in SBvsBB spot here: http://imgur.com/xOLADGj this particular one confirming common wisdom that overcards are the best cards to 2nd barrel while cards pairing the board the worst)

Having the whole tree available it is also possible to jump branches "vertically" so changing say river card without going back and forth in the tree. Using this fact we developed vertical browsing: you can just use arrows on your keyboard to change rank/suit of current turn/river card and see how the strategies/ranges change.

Strategies:
http://i.imgur.com/71slZOr.gif

Equities:
http://i.imgur.com/dBsZGDM.gif

This is another feature to make it easier to understand how GTO frequencies, ranges and EVs change depending on exact runout.
It works very smoothly on full trees but even with small saves (turns+rivers) the delay is well below 500 miliseconds.

2)Speed

I posted some benchmarks in other thread and I will post more once 1.4 is officially announced. Today however I would like to go back to the discussion from NVG Claudico thread. To remind you: Claudico had 32 cores to recalculate rivers and was often taking 30 seconds or more to make a decision. It was explained by the authors with the fact that they need many bet sizes and were forced to use lossy abstraction anyway (hence Claudico didn't understand blockers). My claims that this is ridiculously slow were dismissed.
To set the record straight, here is PioSOLVER solving deep (7x pot) river situation with 6 bet sizes and 4 raise sizes. This includes long chain of minbet/minraise/minraise etc. and doesn't use any abstraction what-so-over:

http://i.imgur.com/rdGLRdW.png

(This is with change_step 0.5 settings which you can use in 1.3.2 or free version as well, defaults are tuned for flop cases).

As you can see PioSOLVER needs 4 seconds to go below 0.1% exploitability while it uses only one CPU core (we don't do multithreading on rivers because for bigger trees it's more convenient to split threads earlier and it wasn't worth it to solve rivers faster as they are already very fast).
Claudico had 32 cores and needed 30 seconds to reach way less accurate solutions (and thus it kept blundering because of lack of card removal effect).

This means we are at least 2 (or more likely 3) orders of magnitude faster. While it won't be easy to beat Pittsburgh super computer with a Macbook pro we are getting closer by the day.

That's all for today, more coming soon!

+1

We offer import from (by copy-pasting). As to import to... well once more than 5 weights are supported with many decimal places then maybe

Hey punter11235

How to start tree in 3b pot when OOP should bet first.

Solver shows check like 90% times with and without flag on "Start the tree with OOP decision"

As we know, in 3bet OOP we should bet at least 50%+, how can i setup piosolver for oop bet?

Solution for this spot PioSOLVER shows to OOP Check 90% and fold 65% times when villian bets, in my opinion this is even not close to unexploitable strategy

Or i'm wrong? And check 90% is optimal?

I think I understand a little, on the other boards pio shows normally all actions.

in the "old style tree building" format, selecting "start the tree with oop decision" is how you do it, as you've done.
if you don't have it selected it should show a check 100%.

this just isn't true.


a quick look at the ranges you've set up on this board, looks like villain just has much higher concentration of good Qx and draws while hero only has QQ+, a small % of AQs, and some weak queens and then a lot of things that miss this board.

OOP is always first to act.
If you are using old style tree-building you need to check the checkbox but I think it's better to use the new one which is way more powerful/customizable.

IP range is way stronger on this board (65% vs 35% equity) so no wonder OOP almost never bets.
I solved that case (well I tried to copy ranges from the picture) to 0.1% (which btw takes about 90 seconds in new version) and it bets only 3.8% of the time. It's a bit surprising it's not 0% as usually weaker range OOP just checks everything.

The solver doesn't care about human sentiments like initiative preflop and it displays how close the strategy is to unexploitable so an opinion of a human about the matter is of little value
It's more constructive to focus on "why".

You can start with building a tree and then going to the browser and showing equity:
http://i.imgur.com/psfOUh9.gif

You can then see that this is a special board where IP has huge range advantage. On something more "normal" like K73r it's OOP who holds the advantage (not only in equity but also in number of nutty hands) and that's why it wants to bet more.

Thanks for the explanation, mates!

New discoveries are waiting for me in understanding the game

I'm a big fan programs such as Piosolver, crev, simple nash, etc. But pio have really new level of game improving and i'm really happy.

Hello,

I am happy to announce PioSOLVER 1.4 release. This is a release I am most excited about since the beginning of the project as it brings huge performance and memory efficiency improvements. It's on average 3x-5x faster for 6max cases and about 2x for HU ones. The memory requirements are way lower as well. More details here:

http://piosolver.com/blogs/news/3539...er-1-4-is-here

The highlights of this release:
1)Huge performance improvements which allows for solving typical 6max cases with one bet size in 1-2 minutes.
2)Vertical browsing which makes it possible to quickly and visually compare ranges on various run-outs in the same line, like here: SEE GIF
3)Some often requested features like more options for tree-building, strategy rounding as well as Cepheus ranges included in default Ranges.

PioSOLVER is now so fast it becomes practical to analyze tens of hands after the session. It's faster to solve a tree to very good accuracy than it would be to set exact filters in HM or input conditions in CREV. You can create few hundred flop library overnight (well, as long as you stick to trees with one bet size) and then use our multi-flop aggregation plugin to derive GTO strats like cbet% or check-raise % over many flops.

Memory efficiency improvements mean that 8GB of RAM is no longer a big constraint for 6max cases (it's still a bit painful for HU ones). For 6max ranges it's usually 2x or more improvement.

I also encourage you to check out the vertical browsing feature. It never was as easy to instantly see the difference for say calling range on various turn/river cards or how 2nd barrel range changes across various turn card. You can never hope to achieve that with standard decision tree interface.

There are more details to come in near future.
I can't believe myself how much PioSOLVER is capable of in this version. Have fun!

A small wish.

Can you add in "Select Range" option/button Invert Matrix?
I'm working on 25BB GTO and constructing preflop ranges first.
GTO implies the presence of each hand in the range in balance/weight.
Example:
We construct shove range with weights against limp, with new button(invert matrix) it would be possible to save checkback range and start calculate situation limp range vs checkback range.
Manually build each range is very difficult and it takes a long time.
I hope you'll add this button.

Yeah, it's on the to-do since forever but keeps getting pushed down by urgent things.
I really like this feature so I am pushing it close to the top right now

The Update is really really good. Speed is incredible. Good job

Today start solving with new Piosolver pro 1.4. Speed is amazing.
I solved 31% vs 38% faster than 3 times.

I second these.

I have only 8gb of RAM so the trees I could build before was fairly limited, but now I can put in multiple sizings on all streets and it's much faster.

Great software, wholeheartedly recommend it!

This software is amazing. Support have also been very helpful. I play 100/200Zoom, I have had this software for about 1 month, I have probably gained the cost of it back in EV at the tables already, (have the $ 475 pro version)

New version: confirmed awesome. Even for very wide HU ranges in SRPs, speed improvement is well noticable; for HU 3b pots even more so.

(Can't personally speak for 6-max cases, but have no doubt that speed improvements with small ranges have to be way beyond this).

I have the basic version.. get an error when running after installing from updater when i try to open pioviewer.exe

"Failed to connect: Pioselver is malfunctioning. It didn't reply to is_ready."


EDIT: Problem resolved when I run pioviewer as admin.

Yeah, this is unfortunately a common problem. It's caused by the Windows warning about downloaded .exe files. We will try to fix it in the future, for now here are the steps to solve it:

1)close PioViewer
2)open PioSOLVER-version.exe (double click on it)
3)if there is any Windows pop-up uncheck "ask every time at start" or check "don't ask again"
4)once the solver console with a welcome string appears, close it
5)restart PioViewer, it should work smoothly from now on

If that doesn't solve the problem, try disabling anti-virus/firewall and see if it works without them. If it does, look for filters/exceptions in those programs. If it doesn't, contact the support

Yeah, that may work as well. Thanks for pointing this out. In general though neither the solver nor the viewer require admin rights to run.

I was trying to run some turns but am probably not understanding this correctly. For OOP I am entering a kind of linear 3 betting range and for IP kind of a wide 50% defending range on a AsKdTh2d turn. When I build the tree and press "go" under the tree building and calculation tab it has oop checking 100%. When I build the tree under the old style tree building tab and check start the tree with OOP decision, it has oop betting 82% but then has IP calling 99%. What am I doing wrong? Thanks.