Computer Configuration for Trading

[Imaginary Friend]

Can you increase maximum JVM memory limit to 32 GB? I had a message about insufficient memory a few days ago with a JVM memory limit set to 16 GB.

[Svyatoslav]

Not via UI, but if you open "C:\Bookmap\Config\launcher-options.conf" on Windows you can edit xmx parameter there

[Imaginary Friend]

Thanks, that's what I was hoping for.

[willy_wihardja]

Hi, if I'm running DXFeed for 20 tickers with HIRO tracker for most, does that fall under the heavy use recommended specs? Will run IBKR TWS alongside as well as 1-3 browsers for TradingView charts.

Trying to decide if I should get the 5600X+32GB RAM or 5900X+64GB RAM. The 5800X doesn't make sense value-wise.

Would Multibook be a factor for increasing my core count needs?

[Andry API support]

Hi,
5600X is a bit on the lower end. 20 tickers is definitely a heavy use case. So 5900X sounds better.

Would Multibook be a factor for increasing my core count needs?

Currently not but may be in future.
 

[willy_wihardja]

Tku Andrey!

[r_lizza]

I'm actually just about to build a new PC and was thinking of going with a 5600x until I saw this thread. Would the 5800x or even possibly an intel i7 11700k or an i9 10850k work better with using 20 tickers? 

[Svyatoslav]

Hi,
There is, presumably, a microcode bug, affecting running Java on 10th and 9th gen Intel CPUs, so you might want to avoid those: https://bugs.openjdk.java.net/browse/JDK-8263710 (so maybe don't get 10850k; for those affected - switching to SerialGC in bookmap settings might help, also see below)
I have no data on 11th gen to say for sure, but it probably works there.
There also is a partial mitigation which we'll attempt deploying soon (Java 17 should have it, and it just came out).

As for the rest of your question - while bookmap itself is not that well multithreaded, you will typically run other software alongside with it, so something like 5800x will give you more headroom. If you just run bookmap and nothing else - you likely won't see a difference in current versions (you might in the future).

[isolantbob]

hi,
could you please help me upgrading my setup? Im hesitating between these different computers to run bookmap + some charts on tradingview :

1. Dell - Inspiron 3891 - Nvidia GeForce GTX 1650 Super 4go - Intel Core i7-10700F - RAM 8 Go - SSD 512 Go - 
   
2. HP pavillon gaming TG01-1869nf- NVIDIA GeForce GTX1650 Super 4go - Ryzen 5 4600u - Ram 8go -SSD 512Go
   
3. SKILLKORP SK16-R51650S - Nvidia GeForce GTX 1650 4Go - Ryzen 5 3600X - Ram 8 Go - SSD 512 Go
   
4. MILLENIUM MM1 MINI RAMMUS - Nvidia GeForce RTX 3060 - Ryzen 5 3600 - Ram 16 Go - Hdd 1 To + SSD 480 Go
   
5. SKILLKORP SK45-R73060 - NVIDIA GeForce RTX3060 12 Go - Ryzen 7 5800X 3,8 GHz - Ram 16 Go - SSD 1 To
   

1,2 and 3 are about the same price, 4 and 5 are a bit more expensive.
 
Thanks.



 

[Svyatoslav]

Hi,

First - please define what kind of use case do you have in mind for Bookmap? Requirements will differ wildly depending on if you want to watch a single relatively calm instrument (almost any modern hardware will do) or if you want multiple active instruments with MBO data and a bunch of addons (in the latter case I would at least throw in another 16GiB of RAM to option 5, though, depending on exact expectations, that might not be enough).

However, let me still comment on the above... I would avoid building a new system with 8GiB of RAM. I'd say 16 is the minimum you want. Try to have some free slots for an upgrade to 32GiB in the future, if possible, cause that might be a good thing to have. Also if you can go for a desktop - that's better than a laptop due to most desktop components being faster at the same price point (I believe your options 3/4/5 are desktops)

Re1: I would avoid because of https://bugs.openjdk.java.net/browse/JDK-8263710

So based on the above considerations - in your list options 4/5 look like the best two with option 5 being the best due to a better CPU (that's 5800X, right?).

If you can get a lower end GPU - might save you some cost without sacrificing too much. I'd consider diverting some of the budget from the GPU to RAM.

However, I didn't look too deeply, so above is a very approximate advice, might contain mistakes.