Strength of NCM

Hi! I am planning to purchase NCM pro but chessify looks stronger to me. And how strong are the NCM engines compared to chessify? I am planning to create an opening course for a big company so i am planning to use this platform for analyzing, finding novelties and etc… and is NCM enough for this?

And by the way can the admin send me any results or tests using the same hardware used for pro members?

Greetings! I haven’t tested NCM’s performance directly against Chessify, and I’m not exactly sure what hardware they’re using, so I can’t really say one way or another. I know they charge $1188 per year and we charge $19 per year, so I wouldn’t be surprised if they outperform us given that kind of budget :slight_smile:

Also my understanding is that on Chessify, you’re not limited to 30 seconds per calculation, so there’s surely some advantages there.

I’m hoping to add an optional pay-for-what-you-use server rental feature on top of our existing $19/year pricing model, so at some point you’ll be able to rent, for instance, a c5.metal server (AWS EC2) with a 96-thread Xeon Platinum 8000 series processor for something like $0.10 - $0.15 per minute, be able to able to run all of NCM’s engines on it, have a persistent hash table between requests, and not be limited by a 30 second maximum think time. I don’t have a timeline for this yet, but we’ve been steadily reshaping the internals of NCM with this in mind for the past couple of years.

If you’d like to evaluate NCM’s hardware, sign up for the trial and I’d be glad to upgrade it to full access. Just shoot me an email at

Hello! Thanks for your reply! Is your existing hardware good for creating an opening course with decent quality? Because if it is, then i would buy the pro package

Ps: its 12 midnight now so if you answer i will have to wait until tomorrow. Thanks in advanced :smiley:

It might be :slight_smile: I suppose it all depends on exactly what you need. With the pro version, you can analyze anywhere between 100ms and 30 seconds per move on servers equipped with Dual E5-2680v2 CPUs, or in the case of LcZero, servers equipped with RTX 2080 GPUs. All calculations get 100% of the server’s resources, i.e., each server handles no more than one calculation at a time.

I’m just not sure if that meets, exceeds, or falls short of what you need for the purposes of creating an opening course.

I can tell you that at the starting position, Stockfish 13 on our 20-CPU core servers seems to hover around 17000knps, and reaches a depth of 40 after 30 seconds. Here’s the last info line in the UCI output:

info depth 40 seldepth 53 multipv 1 score cp 25 nodes 499356957 nps 16644122 hashfull 998 tbhits 0 time 30002 pv e2e4 e7e5 g1f3 b8c6 f1b5 g8f6 e1g1 f6e4 f1e1 e4d6 f3e5 f8e7 b5f1 c6e5 e1e5 e8g8 d2d4 e7f6 e5e1 f8e8 c1f4 e8e1 d1e1 d6e8 c2c3 d7d5 a2a4 a7a5 f1d3 g7g6 b1d2 e8g7 d2f3 c8f5 d3f5 g7f5 h2h3 f5g7 e1d2 g7e6 f4h6 f6g7 a1e1 d8d7 f3e5 g7h6 d2h6 d7a4

(You can click the engine name after calculating to show the entire exchange between NCM and the engine including all of the UCI output.)

Anyways, if have any specific positions you’d like me to analyze with any hardware/engine combination, let me know, but I encourage you to try it out yourself!

Okay this seems pretty good enough. Are your stockfish dev builds NNUE? And if i am not mistaken they have nets right? And may i know the nets if they exist? (Sorry just new here)

They do! Stockfish 13 and subsequent dev builds are all built with the NNUE network embedded in the binary. You can verify that by clicking the engine name to see the logs. For instance, Stockfish 13 shows:

info string NNUE evaluation using nn-62ef826d1a6d.nnue enabled

While the latest dev build shows the NNUE network which was just committed yesterday:

info string NNUE evaluation using nn-8a08400ed089.nnue enabled

I used chessify 15 days pro ,its very costly, and played ncm latest sf dev build 20 cpu against chessify 500,000kN/s server, result all game ended draw, only chessify beat lc0 t40 net ,chessify works the following way. Once you buy a package, you receive
a certain number of coins. In case of a pro package, you’ll receive 2500
coins. However, regardless of the package, you will have access to all
their servers and engines. It is completely up to you to decide on which
server you want to spend your coins. For example, if you have the
package pro (2500 coins), you can analyze on the 500,000kN/s server for
31 minutes. Please also be informed that after buying a package, you will receive
Free access to their 30,000kN/s server but i used 30,000kns all game chessify lose from ncm I asked about this chessify they replied its possible and send me their test result its showing 50 cpu play with 1 cpu win 31draw 19 50 cpu against 4 cpu win 23 darw 27 and they said Stockfish working on 50 CPUs will analyze around 50 times more positions and will still reach higher depth much faster than Stockfish working on one or four CPUs.So the problem was that our cloud Stockfish engine, which worked on a faster server, reached the same and sometimes even lower depth than the local one. This was taken as a problem on our part,

Thanks for your replies Chendry! I will probably buy the pro package once i will start creating the opening course😄

Edit: BTW if you could test NCM with chessify that would be amazing

Hi puttutathy! Thank you for your inputs! Do you mean the NCM Stockfish strength is actually the same with chessify Stockfish? And are the NCM Stockfish servers faster?

According to to strength not huge difference between ncm and chessify ,ncm good for$19 for 1 year if you want huge strength intel duel xeon platinum 82602,2 3,7GHz 48 core(96 threads) 58000kns available in other site but very costly

Thanks again for your useful inputs! Could you measure NCM and chessify on a scale of 1-10? Eg. Chessify is 8 And NCM IS 7

I don’t think I’d be able to provide any kind of meaningful comparison without knowing exactly what kind of hardware Chessify is using, engine settings, etc. I can’t find anywhere where they advertise that. Based on @puttutathy’s comment above, I’m wondering if they do use something like a c5.metal instance at AWS? IIRC, those are Xeon Platinum / 48 core / 96 thread servers.

So that would be a faster CPU with more cores, and faster cores. But I can’t really quantify the difference without playing a whole lot of games between the two.

We are planning on adding support c5.metal servers or similar at NCM, and I’ve done some rough benchmarks in the past. From my records, for Stockfish 13, compiled with make -j build ARCH=x86-64-vnni512 COMP=clang the results for 30 second calculations from startpos are:

using 48 threads and a 100GB hash:

info depth 40 seldepth 55 multipv 1 score cp 16 nodes 1655283435 nps 55174275 hashfull 109 tbhits 0 time 30001 pv e2e4 e7e5 g1f3 b8c6 f1b5 g8f6 e1g1 f6e4 f1e1 e4d6 f3e5 f8e7 b5f1 c6e5 e1e5 e8g8 d2d4 e7f6 e5e1 f8e8 c2c3 e8e1 d1e1 d6f5 c1f4 d7d6 h2h3 f5h4 b1d2 h4g6 f4g3 c8d7 a2a4 d7c6 d2c4 d8e7 e1e7 g6e7 a4a5 a7a6 c4e3 a8e8 a1e1 f6g5

(55,174 knps from startpos)

For reference, here’s the result of a 30 second calculation I just did with NCM:

info depth 38 seldepth 51 multipv 1 score cp 29 nodes 498417353 nps 16613357 hashfull 999 tbhits 0 time 30001 pv e2e4 e7e5 g1f3 b8c6 f1b5 g8f6 e1g1 f6e4 f1e1 e4d6 f3e5 f8e7 b5f1 c6e5 e1e5 e8g8 d2d4 e7f6 e5e1 f8e8 c1f4 e8e1 d1e1 d6e8 c2c3 d7d5 a2a4 a7a5 f1d3 g7g6 b1d2 e8g7 d2f3 c8f5 d3f5 g7f5 g2g3 c7c6 g1g2 d8d7 e1d2 a8e8 h2h4 e8e4 h4h5 g6h5 a1h1 f5g7 f4e5 f6e5 d4e5

(16,613 knps from startpos)

So, assuming Chessify is using a c5.metal server as their 500,000 knps server, which is a big assumption, I would expect their 500,000 knps server to be noticeably stronger that our 2x Xeon e5-2680 v2 servers. But it’s also a lot more expensive.

Have you considered running the engine locally on your computer?

Another route you could take if you want a ton of computing power would be to rent a c5.metal instance yourself from Amazon – I’m seeing that spot instances are as low as $0.92 per hour:

Then you can run the engine remotely on that. If you happen to be running on Linux or a Mac, it probably wouldn’t be too difficult to connect a local GUI like Area or SCID to an engine running remotely over SSH. I’m sure it’s possible with Windows too but I’m just not as familiar. Let me know if you’d like to explore that route and I’d be glad to help!


Thanks again for the detailed info @chendry!

$0.92 per hour still seems quite expensive + it would be a hassle to do the set-up. I have a few very important upcoming chess tournaments. After those i will consider buying a pro subscription here since the equipment seems more or less fine and the customer service is great. Btw it would be great if there are tiered subscriptions, ex. Gold tier gets the best hardware. And it would be great to add stronger and better hardware, i wouldn’t mind paying a few more bucks for those.

Stay safe
Regards, Fletch

Your proposal on strengthening the server here with paying more money. I am with this idea with satisfaction. We are interested in working here to increase the strength of the server.

1 Like

please read what i said ?(i said you want huge strength intel duel xeon platinum 82602,2 3,7GHz 48 core(96 threads) 58000kns available in other site but very costly
] means not chessify using chessify only saying 500,000kN/s server not mention any hardware other site lot of site available for example

Doh! Sorry, @puttutathy. I misread.

That actually looks like decent pricing - $12 USD per day for that server on the bottom? I just did some quick searching and I found a “Dual Processor Intel Xeon Platinum 8160” (not 8260) for $1099 per month / roughly $37 per day.

All are very costly ,If you provide , a c5.metal server (AWS EC2) with a 96-thread Xeon Platinum 8000 series processor for something like $0.10 - $0.15 per minute ,its very good for us or increasing ncm thinking time with more prize

Yeah I agree, I wouldn’t mind spending a few more bucks

I tried to use the new nn-8a08400ed089.nnue on my local machine yesterday but the engine crashed without starting. Then I noticed that this latest network is 40 MB, where past networks have all been 20 MB. Did you have to do anything special with the engine settings to get it to run this double-sized network? Thanks!