Hi scylla, I'm using a fairly old computer with an Intel Q6600 CPU and 4gb RAM, would you expect much of a performance increase with 8gb RAM, or is the software more reliant on the CPU?

The speed of the solver is almost exclusively decided by the speed of your processor (as well as its number of physical cores). The amount of memory you have only affects the largest possible tree size that fits in memory. If you can run all tree sizes that you like, then additional memory will not be necessary. If on the other hand the software crashes for large trees during either EV runs or runs of the solver then an upgrade in memory would be desirable. That being said, an update was just released that features between 2x and 5x memory compression for the solver, so from this point on it should be rather unlikely for it to run out of memory. It may still happen for EV runs or the Max Exploit tool though; memory reduction there will be addressed in later versions.

Awesome update, the memory compression is really nice. Now we just need a better tree builder to take advantage of all that extra memory.

I once tried to build a multiple bet size tree by hand but it took forever, there's just so many nodes to go through.

BTW, it did crash with the speedup ticked. Any idea why?

Do you by any chance have a really old processor?

I have an FX6300. It's kinda old but it does have all modern instruction sets.

From what I can tell it should work. For now though, please just leave the speedup option turned off until it becomes a bit more clear how prevalent this issue is. Also, do the example files in /equilibrium_sample_files by any chance work for you? It's of course possible that you're doing something that hasn't been tested for. In that case please send a savefile to support so that we can try to reproduce the issue.

Thank you for the feedback,

Scylla

They seem to be working, also other simple trees work as well. Seems it only crashes when I use it on more complex trees with multiple river bet sizes.

fwiw, solving the 4bet tree, before update - 381 seconds, after update with "speedup" selected - 277 seconds, using Q6600, 4gb RAM.

+1 for that. Trying to build trees with multiple betsizes and raisesizes is really hard right now.

What sort of speeds are users getting on modern computers for 4bet trees, and what is the best type of processor for this software?

Is it possible that in those cases the trees are so huge that you've run out of memory?

Yes, you already reported earlier that the solver slowed down considerably with larger trees, right? It sounds like you may be swapping to disk when running out of memory. In that case a memory upgrade would probably be helpful. Other than that, does the program by any chance speed up with "Compress memory" selected? The memory compression should actually slow the program down by ~25%, but in your case it may work a bit differently.

On my desktop the example file /equilibrium_sample_files/flop/flop_4bet.stx takes 85 seconds. On my laptop about 100 seconds. Both systems have 4 cores and are running on 8 threads. The most important factor when it comes to processors at the moment appears to be the CPU speed and number of cores.

PS:
And 290 seconds on my slowest laptop, with 1.7Ghz and 2 cores.

Doesn't seem to be any difference with memory compress enabled.

Well, memory compress should slow the program down by about 25%. So if there's no difference, then you're actually getting a 25% speedup from using less memory. I can't really say what would cause your system to run slower when using more memory, but there's a pretty decent chance that upgrading to something like 8GB would make everything run more smoothly. It could always be related to some other part of your hardware though, so unfortunately it's hard to give guarantees here.

Hi,

Is there a possibility to output the weighted excel ranges with a "," between each hand as a single line in CREV?
Because then it would be possible to import these weighted ranges into Equilab.
Thank you for your response!

Are you referring to the Alt+W toggle?
Because there you just need to check "show as single line" in "grouped" output.

Yes, I am referring to this toggle. The problem is (as you can see it on the picture too that you posted) that there is no comma (",") sign between the individual hands.
For example: ...[/8.35][8.22]4d4h[/8.22][8.14]...
Unfortunately because of this Power Equilab can only partially import the range. Basically it can only import those hands which have the same weight and are grouped together and separated by a comma sign.
So it would be nice to have a "," between every individual hand like that:
...[/8.35],[8.22]4d4h[/8.22],[8.14]...
That would work well. I have already tested it.
Thank you Scylla for your great work by the way!

So I mainly play HUsngs and I'm trying to construct preflop ranges based on what I think opponent's ranges are. Mostly I'm trying to figure out which hands I should flat/3-bet/fold oop at different stack depths and which hands to limp/call (fold) vs an iso at shorter stacks. But I'm not sure how practical this CREV solution is because it's based on a check down whenever a raise is called, which we all know isn't going to happen most of the time so I'm not going to 3-bet small pairs every time like the solver suggests because I won't get to realize that equity so easily. How do you guys adjust your preflop ranges to account for postflop play?

64 Bit Download Link of new Version 3.3.1 doesn't work for me -

starts with 0 bytes / second and never ends.

32 Bit Download works fine .

Any Issues wiith the link or can you maybe give another Download Link ?

Thanks

Worked now

The preflop solver in CardRunnersEV is a basic solver, however, when adding a checkdown it is possible to set custom checkdown percentages for individual hands. For that, click the 'C' button in the "Add action" menu.



Go here for a basic rundown of the custom checkdown feature: http://www.cardrunnersev.com/customequities.html

For example, if you enter "77-22: 80%" then all equities for these hands will be handicapped by 20%. So if the equity is 45%, then it will be turned into 80%*45%=36%. In this manner you can strengthen/handicap certain hands. The solver has been adapted to work with this feature, so it will take this data into account.

Ok, I have made this adaptation in the output. There's probably a release in the next few days with minor changes and fixes, so with any luck you'll have it shortly.

Ok, great! It sounds like a server was down when you were originally trying to download.

That's a nice little feature. But i'm still kind of shooting in the dark when trying to estimate a checkdown % if anyone can give me a rule of thumb. i don't use any kind of tracking software. I'd have to guess that vs an aggressive opponent, the hand is checked down maybe 20%?