Confused_by_Theory wrote on 06/09/19 at 19:59:02:


All of the Ryzens have very good price/performance for chess, so it's up to your budget. Also: do you want to keep some of that budget for an RTX card to run Leela on?

Quote:


The problem will be the mainboard. Most of the previous generation mainboards will support the Ryzen 3xxxx series, but not all of them and not all features. It would be sad to build a new machine now and to have to get an old mainboard.

Quote:


For performance, yes, aside from the GPU if you want to run Leela. For stability, it is worthwhile to buy a good PSU, mainboard and RAM as well. Nothing is more annoying than a machine that hangs or crashes if you analyze longer than 15 minutes!

A nice big SSD is also a worthwhile investment, so you can throw your databases, opening books and Syzygy endgame databases on it.

Hello.

So there is a Ryzen 9 on the way. That does sounds mildly enticing. Also if not that then Ryzen 7 3700x is also good(?)

But then... What about fixing the other parts of the computer and upgrading the CPU to Ryzen 9 in a month. Guess maybe another more or less random CPU will be needed as stop-gap in this case; although it should be possible to find something cheap then perhaps

Also. Am I correct in assuming other components are less important compared to the CPU?

Have a nice day.

Maybe wait till the third series of Ryzen chips comes out in a month. The 3700x looks like a steal for price / performance.

https://www.tomshardware.com/news/amd-third-gen-ryzen-7nm-launch-intel-cpu,39449...

daniel7472 wrote on 10/16/17 at 06:46:09:


The issue with the i5 is the expensive motherboard. You can use a B350 with the Ryzen - I recommend the Asus Prime Plus, which you can overclock in the OS with the Zenstates utility.

Bondefanger wrote on 09/27/17 at 15:49:00:


Let's just admit that for some of us (like me) it's simply fun to know you have a strong engine--just as it would be fun to have a car that went faster than you could ever drive it. Irrational, but true.

proustiskeen wrote on 09/27/17 at 14:50:17:


Choosing the right chipset has nothing to do with these things you mention here. I could do these things just as well on a 10 year old Intel Celeron laptop.

Of the things you mention, only the part about having "the best player" could be affected by the chipset. And the importance of that, if you are primarily an OTB player, is certainly up for debate.

My present desktop computer is at least 5 years old, and optimized for silence rather than speed, and I do not feel the least bit at a disadvantage because of that that.

If the opposite became true, I would propably also quit chess.

barnaby wrote on 09/25/17 at 20:07:05:


Thank god it isn't.

Back in 1998 I build a computer far stronger than anyone else I knew had. Optimized for chess calculations, it would calculate away on various positions when I wasn't at home. Either at critical positions or playing engine tournaments in my favorite dubious opening lines.

What good did it do for my chess? None at all.

When owning the proper chipset is an important tool in playing chess is when I know the game has passed me by and its time for other things a life..

So long and thanks for all the fish.

It will be hard to use Houdini on Amazon EC2 as there is copy protection involved in the install.

I've used Komodo and Stockfish on EC2 under Windows installs. Very easy. Just drag your Chessbase 12/13/14 install file over, don't register it, and set up your engine in the Engine Cloud on the EC2 instance. Very efficient way to get huge iron if / when you need it.

That said, I haven't done this since I got the Ryzen7 1700.

It is possible to run a chess engine on Amazon EC2 instances in the cloud. Naturally, you would have to pay for this. But it does make the choice of buying computers much simpler, and paying per hour is much cheaper overall than paying for a beefy computer!

If there's enough interest I can elaborate on how it's done. It's not too complicated.

Brief overview:

1) On Mac and Linux (not sure about Windows) UCI engines talk to GUIs via standard streams. This means that to spoof a remote engine to look like a local one, all you need to do is write a script that connects and launches the remote engine. Hand this script to your chess GUI, and it will be none the wiser that the engine is actually running elsewhere.

2) Cloud based CPUs don't necessarily all live on the same hardware. This extra latency can hurt chess programs (although I haven't investigated to what degree). You can provision all your cores to be on the same hardware on AWS, but not in most other cloud providers. This costs a bit extra.

TN wrote on 08/21/17 at 07:49:49:


For chess engines, Ryzen blows Intel out of the water. Having more cores per $ spent is important, but just as important is the fact that Ryzen's version of SMT (Infinity Fabric) is more efficient than Intel's hyperthreading, making full use of all threads a net plus with most engines.

Thanks for the tip. Regarding the search, I did that but couldn't find anything relevant. I guess I'll stick with R5 1600 since I still don't understand the benefit of making the effort. Sure, R7 will be better but I couldn't find any strong argument in favor of R7, so I can make the effort. If the difference in analyzing a position is just a couple of minutes ... well its not worth it, for me. If the difference would be I get different evaluation ... then yes, that would worth it.

I am about to buy a new PC and I am torn between Ryzen 1600 and Ryzen 1700. Except chess I work with Autocad (MEP, Electrical, Revit) do some casual gaming (company of heroes) and light programming. What will you suggest, considering cash is important. If the difference in performance justify the money, I will do the effort, otherwise not.
The rest of the system:
16GB DDR4 3200mhz
256gb SSD
Radeon R9 380