Development PC - Minimum Spec
Moderators: hgm, Rebel, chrisw
Development PC - Minimum Spec
I am considering getting a new PC for making changes to open source chess programs and running test matches among them. I'm not interested in being competitive so I don't need a top-of-the-line PC, just one that is adequate (and cheap!). I don't want to be throwing away money on something that will not be sufficient. Does anyone have any suggestions regarding CPU, RAM type and amount, and any other hardware considerations I should be aware of?
-
- Posts: 778
- Joined: Sat Jul 01, 2006 7:11 am
Re: Development PC - Minimum Spec
If you are not planning to create egtbs, virtually any machine should do. Check your development tools for minimum or recommended configuration.
Compiling chess programs is a pretty light task.
Compiling chess programs is a pretty light task.
-
- Posts: 173
- Joined: Sun May 11, 2008 7:43 am
Re: Development PC - Minimum Spec
I would get a Dell.com.
Studio XPS 8000 model $699
Customize it to a i7 processor and you'll be fine.
Don't forget to add your monitor.
Apply online for financing. $30/mo.
Studio XPS 8000 model $699
Customize it to a i7 processor and you'll be fine.
Don't forget to add your monitor.
Apply online for financing. $30/mo.
-
- Posts: 1243
- Joined: Sat Dec 13, 2008 7:00 pm
Re: Development PC - Minimum Spec
Correct, just get any cheap quadcore machine. Make sure you get a good monitor and a 64 bit OS. You should be able to get everything for <500 USD.jwes wrote:If you are not planning to create egtbs, virtually any machine should do.
4 cores are useful for:
- testing the SMP algorithms
- running multiple testmatches simultaneously
-
- Posts: 317
- Joined: Mon Jun 26, 2006 9:44 am
Re: Development PC - Minimum Spec
Compiling chess programs may be a light task, but testing chess programs is not. For chess programming, testing is a bottleneck, where faster is better.jwes wrote:If you are not planning to create egtbs, virtually any machine should do. Check your development tools for minimum or recommended configuration.
Compiling chess programs is a pretty light task.
-
- Posts: 1154
- Joined: Fri Jun 23, 2006 5:18 am
Re: Development PC - Minimum Spec
Agreed, and for this purpose, # cores is more important than speed.rjgibert wrote:Compiling chess programs may be a light task, but testing chess programs is not. For chess programming, testing is a bottleneck, where faster is better.jwes wrote:If you are not planning to create egtbs, virtually any machine should do. Check your development tools for minimum or recommended configuration.
Compiling chess programs is a pretty light task.
-Sam
Re: Development PC - Minimum Spec
Thank you all for the replies. A further question:
Is the amount and type of cache a problem? For example, if there is not enough cache, when I am running a test match, could it happen that one program will reside more in the cache giving it a speed advantage?
Is the amount and type of cache a problem? For example, if there is not enough cache, when I am running a test match, could it happen that one program will reside more in the cache giving it a speed advantage?
-
- Posts: 243
- Joined: Sat Mar 11, 2006 8:31 am
- Location: Malmö, Sweden
- Full name: Bo Persson
Re: Development PC - Minimum Spec
No, and yes.pnorthover wrote:Thank you all for the replies. A further question:
Is the amount and type of cache a problem? For example, if there is not enough cache, when I am running a test match, could it happen that one program will reside more in the cache giving it a speed advantage?
The program that is best tuned for your environment will have an advantage.
But that is only fair.
-
- Posts: 1154
- Joined: Fri Jun 23, 2006 5:18 am
Re: Development PC - Minimum Spec
I wouldn't worry about it personally. You will probably be running test games with fast time controls and small hash tables. I am sure there could be some odd issues related to cache...but that's just life. Its not a big enough deal in my mind to factor into your purchase decision. The fact is, for your first chess program the engine strength gains you will get when adding features are large enough that you really don't need to have a perfect testing environment, just something where you can run lots of games quickly. For example, when if you set up a test of a programming change, are you going to want to patiently wait for it to finish? Probably not...you will probably want to start coding new changes while you wait. You might even want to compile your changes, or run a new version to make sure your change worked or...anyway the main point is if this is your only machine you are not going to want to set up tests perfectly and leave it alone, so you just have to accept your tests are a bit imperfect. You will, however, want lots of cores (4 is a good price point I think but more is always better) so you can do all this testing and coding and web surfing and whatever at the same time.pnorthover wrote:Thank you all for the replies. A further question:
Is the amount and type of cache a problem? For example, if there is not enough cache, when I am running a test match, could it happen that one program will reside more in the cache giving it a speed advantage?
-Sam