I will pass it to PioViewer dev.

It's a good suggestion but I don't think we will be doing improvements to aggregation reports in the near future. We would like to focus on some more fundamental functionality in the near future instead of doing small features/improvements.

I will pm you Skype contact so you can report it. To be honest I've never used the color settings myself

is 'PioSOLVER-edge.exe' the avx version for 1.10? just crashes instantly for me. i have avx instruction set on my machine. 'PioSOLVER-edge-oldCPU.exe' works fine however.

Yes, there are some more instructions though which are available in IvyBridge or later generations.
Please drop us an email with your exact CPU name, I will see if I can make a compile for you which uses AVX instructions but not the newer ones and if there is any performance gain from it.

hi punter,
you said earlier this year that the new version will have the cfr files at least 2x smaller. Is this the case? because it seems not.

I am careful not to promise things, can you point me to the place I wrote that?
It is possible to make saves smaller it's true. It's not a priority thing though as it's a niche usecase. As it is it's already painful enough to keep old saves backward compatible so I am reluctant making small improvements like that.

In runout aggregated frequency reports how is the hand order for hands that aren't in the pre flop range chosen? Doesn't it make more sense to leave out hands not in the pre flop range? And the EV's of hands are affected a lot by weight right? 53s realizes 110% of its equity vs SB 3bet with a 5% weight, but will realize a lot less with a higher weight? Will the weights of hands skew the hand order?

This one also doesn't work properly for me in new update. All EV numbers get the color for lowest EV regardless of the EV number and player.

It does, it's a bug if that happens. Is it on postflop saves? Are you using 1.10 version? (There was a bug in 1.9 which showed hands for IP incorrectly).

EV of a hand is independent of its weight. It's true that the opponent could adjust if you play differently but as long as they don't the EV is independent.

It's hard to say what happens once you change ranges. That's why solving is difficult and solving preflop is more difficult - it's not that easy to predict what the optimal play with different ranges is going to be.

Confirmed, thanks for reporting.. This was already mentioned by someone else as well. There will be a bugfix tonight with many issues fixed, hopefully this one as well.

"It does, it's a bug if that happens. Is it on postflop saves? Are you using 1.10 version? (There was a bug in 1.9 which showed hands for IP incorrectly)."

I updated and it's the same. I think maybe you misunderstood what i meant. I ran a "runout aggregated frequency report" across multiple flops. In the preflop order text file all hands are listed, even those that aren't in the selected postflop range. The hands that aren't in the range are all listed after the ones that are so it isn't really a problem. Was just wondering how the order of hands that aren't in the range is created.

They are filled in according to the default preflop order. This is done so the slider doesn't suddenly stop on incomplete preflop orders.

https://forumserver.twoplustwo.com/s...postcount=2303

Yeah, everything checks out.
About smaller saves: I actually thought about implementing it but keeping the saves backward compatible is already a lot of work (it delayed this release quite a bit) and this request would only be useful for very small group of users so I decided against it. We try to change our approach to devote more work to fundamental functionality instead of adding tons of small features/improvements so it's very unlikely to make a cut anytime soon.

When it comes to RAM itself: it's now possible to run trees on less precision with 50% of RAM in all edge/pro/basic versions.
This is already available in the solver but not in the graphical interface. Once all the urgent problems in 1.10 are fixed I will post instructions how to use it. The solver is slower with those settings but quite a lot more reliable than preflop solver was in the previous version (still not as precise as the one on the standard settings though).

Hello, @punter11235, I have two questions:

1. I am going to rent a server to run PioSOLVER, there are TWO Intel Xeon E5 2630v4 (10 core x 2.20 GHz), so totally 20 cores, and 256 GB REG ECC RAM, I have PioSOLVER Pro version, so my question is do I need configure my pio to fit this server? now the pro version support up to 16 core CPU now?

2. The second question is I am going to develop some plugin for pio, but I am not familiar with C# language, do you recommend some tutorial to learn how to develop plugin?

Thank you very much!

Hello, thank you for the update. I have some questions. Thanks in advance for your time.

1. Here you say that preflop solver is less precise but in skype group you've said that it's more accurate. Am i missing something?


2. Now in the node-locking menu all buttons are at ''fixed'' mode by default, previously they were at ''propotional''. Is that a deliberate change? It seems to me previous way was more comfortable to lock one action quickly.

3. Why was "Cut selected line" feature removed?

4. And the last one is probably a bit dumb. Could you please explain me the meaning of preflop orders and all of connected features? I work with PIO a lot but I really dont understand why its so important to select a range in appropriate order. I feel like Im missing something useful.

My pio is not updating, when I go to pio updater and try to update it says I am up to date, but I have version 1.9.2.0?

there is a new updater, was posted earlier in this thread

No need to configure anything although pro version won't be able to take full advantage of this hardware. You may try running two at the same time though to get some more efficiency.

Would you like to do something that works with PioViewer or just something that works with the solver directly (like your own graphical tool). Our plugin functionality is very limited right now as it wasn't very popular.

Running on half precision was always less precise than on full precision. It's better in 1.10 version than it was before though. It's more accurate than the previous preflop solver but still less accurate than the postflop one. Btw, you can now run preflop trees on full precision as well (they will take 2x RAM). This is for now undocumented feature.

I don't know, I will ask PioViewer dev

1)We want to cache the whole tree structure in the future so it's important the tree structure doesn't change once it's built.
2)It's better to build a new tree without that line (you can now add/remove arbitrary lines in tree-building form instead of removing it after the tree is already built.

There is no usecase for cut_line we intend to support.

It's probably not that important, it's very nice to have a slider following an order for various situations though. Maybe it's also useful when deciding which hands should be always played according to solutions and which should be adjusted according to reads (so you do so called "soft adjustment" choosing only borderline hands so it's not too obvious).
All in all it's just something we were frequently asked for so we added that.

You need a new installer/updater here:
https://www.piosolver.com/blogs/news/piosolver-1-10

This is because we've changed the whole distribution system (personal links are no longer needed).

Okay I have updated from the link, but when I click on pioSOLVER-pro I get this error:
https://gyazo.com/0a6e5ac0bc94be51908e8af3ef954eab

If I click on the old pioviewer I get the same error, but then pio opens and its still the old version 1.9.2.0, when I try to run a solve it tells me to connect to piosolver ?

It looks like updating Turboactivate.dll file file. Please remove that file manually and then run the updater gain (with force update checked).

[QUOTE=ticoarm;53230649] Thank you very much.

About plug-in, I will do something with PioViewer, actually, my goal is I want to generate my own format aggregation report for multiple cfr files, I only need some information of the report, not all the information, so in order to reduce the report generate time, I want to develop my own plug-in, now my PioViewer support me to do this?

I am sorry but plug-ins to PioViewer itself are not supported anymore. You can still write your own program to communicate with the solver and get any information you want but that's much bigger task.

Is there a way to copy the bet sizing/ranges in the browser into tree building? When I open a flop from a script I made the info in tree building doesn't match the browser. It shows a slightly different configuration I was using around the time the script was made

It's not possible.
Saves created by 1.10 version contain treeconfig though so that problem won't exist anymore.

Hello,
how much ram memory would you say is needed to solve a preflop situation 25bb deep ? (is 32gb enough ?)

I have a bunch of node-locked trees and now want to just change EffectiveStacks and resolve.

Is there another way to do this, except building new trees and manually copying locked ranges?