After the break for the holidays. I'm getting back to this, although I've already checked that it works via cli, the visual layer is missing, there's a bottleneck in which gtp and I are stuck. The order it uses is: 1. tlcs (server.ini configured with the correct paths, we'll delve into this later) 2. wireguard configured correctly 3. Run tlcs. 4. Start the tournament from the framework (the tlcs console receives the tournament data, in the case of winboard it's a debug file that adds the data) 5. From the vps, run node-tlcv and then we already have tlcs and node connected, then the web viewer retransmits the tournament in real time. In our framework, we've taken this idea, but we've created the app from the chessgui framework (ccrl.live retransmits with this framework and the same system) and it's not a.debug file but a txt. The issue is that the tournaments do work, but we can't get the tlcs client to change the board, although the.txt file does record the moves and node-tlcv from the vps connects as with winboard, to tlcs. As although it's the plus version, it's still gpt at the biggest plagiarism ever created, it makes mistakes and can leave a space, add an incorrect character in the code and we have to keep debugging and debugging, but without managing to retransmit as we did with winboard. Why I didn't use winboard, because it only allows you to launch a game between two chess engines, not a tournament. In any case, I take this opportunity to wish everyone a happy new year again and we'll move on to programming issues when the holidays are over. Hugs, comrades.

@jordiqui Here's the logo for the dark theme: it will look better than the official one: https://hardlimit.com/logo_tema_oscuro.png

And I'm going to try to push the English forum a bit.

@cobito said in Cluster 3 nodes of Hp proliant dl360p gen8:

@jordiqui You mean the MPi topic that you commented on here, right?

It's noted. It will be the first change in the next iteration, but I don't know when. I have many changes noted from all the pages and I need to organize the work to optimize time.

To be honest, it would be very interesting (and an uncommon feature).

Let's see how 2024 comes in...

That's @cobito: But with everything you do with the forum, etc. when you can because I imagine integrating MPi must have its own challenges. A hug and a Merry Christmas in advance

@jordiqui Set up all recovery measures and etcs, alternative rescue email, an alternative number (your wife's own) and whatever you can.
In the terminal it would also be good to set up something anti-theft, to be able to lock it, locate it, and more things.

Biometrics is now finally starting to be a standard for identification, that and the mobile through the ISP's infrastructure are the number one unofficial means of identification. Even banks are abandoning confirmation sms to do it also by biometric recognition.
So consider the mobile line and the mobile itself as if they were a master key.

Salu2.

Honestly, "I have" three PCs.
Two at work and mine at home (what else )
The ones at work are one, a mini barebone that runs W XP because it's the one that controls the programs that are launched to the CNC and either the Siemens card or the sinumeric program are not compatible with more modern operating systems.
The second PC is an amd3+ that is not compatible with W11 and has W10 pro
And mine has W11 and honestly.
I don't notice any improvement in W11 compared to the previous ones.
If I notice something, it's because of the hardware that is duplicated from one to another, not because of the software.
Regards.
P.d it is very possible that when I remove the raid 0 from my PC I will return to W10.
Regards 2

@sylver said in Raspberri pi 3 b:

Buenas @jordiqui

Sometimes it happens, read errors and data corruption in Raspberrys are more or less frequent.

What SD card do you have? From what I have read, it must be a Class 10 original (not cloned) and experts recommend that it be SanDisk Extreme pro or Samsung for greater reliability.

Regards

SAndisk vpro I think he has.

@cobito In my case, I only have a digital anthropology project, so I just correct the errors that the Google console alerts us about, as you mentioned, and duplicate content. In this case, I always try to add the canonical URL that I want to be the relevant one. But of course, I publish the content myself and I am the one who decides which word from the academic texts I want to rank. In the case of a forum, you have already commented on this. Then there is the issue of being linked to websites considered spam, but all of this can be checked from webmaster tools, now search console. The part about complying with the standard should be easier for those of you who know how to program. But going back to what you mentioned, I had to leave this job, I was self-employed, precisely because of what you have quoted. It is insane and very enslaving. At least if you do it alone without a work group like agencies do.

Hello colleagues, following the trail of the server world, etc. I realized that the watts of the power supply that each of the U servers I have is 250w. So far so good. Now comes where in the PC world they wouldn't have sold it to me, but here they have sold it to me and by a landslide. To begin with, these servers work looking for redundancy and security, not power. That is, if one fails, the other continues. Well, again. So the hp proliant dls360p gen 8 gen9, etc. come prepared for you to have two power supplies. If one fails, the other continues, that is, redundancy. But if they sell you a machine, even if it's second-hand, it has to work well. That is, without failures. The ones I have come with a single PSU of 250 w. Many servers of these, although they admit two CPUs, only have one, etc. The fact is that in the ones I have, only the two CPUs at full load consume 95-120 watts each. Let's imagine that they consume 120 watts with a voltage of 1.2. They are used and I don't think they consume the same as new ones. 120 by two are 240 watts. If the power supply is 250 watts what do we do with the memory, pci cards connected, sound, hard drives, vga, etc. Well, in the past, I returned a PC in a famous store of those that became fashionable for this issue. The power supply didn't give enough for the hardware that was mounted. How much does it cost to do things well? I think not much. Well, again, I don't think the server world differs much in this sense from the PC world. It's math. And I'm from the humanities. Anyway, I have ordered two 750 w power supplies, to cover my back, because if the demand for watts exceeds what the power supply supports or it works badly or in some cases, it doesn't even turn on. That being said, we continue to pay attention to your publications. Goodbye.

Add to the list very capable cooling, in both cases.

As for DDR4/5, we will see when DDR5 modules with higher frequency are affordable, it is to be expected that there will be a performance gap in that sense, but we will see.

@cobito Thanks!!! I'm going to do it through WSL, which I had discarded, but I have already found out why it was failing, I needed to have Hyper-V activated. Let's go for it!

@defaultuser In this specific case, the texelcl.exe file is created so that UCI access engines can use the cluster, that is, all the nodes or CPUs. But by removing texel, you can set up a network server, and if the first one fails, the second one will continue with the tasks, and so on. I didn't want to get into it more because it's a whole world. But I've reduced it as much as possible. A big hug.

@cobito Bought as new from a second-hand store, plug and play. Plug in, put cables and we already have everything up and running. By the way it's from that brand, it didn't reach twelve euros metallic with five 1000mbs ports.

Hello colleagues. Looking at these photographs, now from a distance, I get goosebumps because these gadgets are designed to fit into a closet. One wrong contact and a short circuit at the edge, then goodbye, who knows, the board, etc. Today I am excited because this is coming to me that you can see in the link:
Rack Cabinet
with its two corresponding trays for the two U. We will leave quite a bit of space between them, because the final goal is to mount four nodes (now I have four CPUs unused, two e5 2660v2 and 2 2651v2). As soon as I have everything mounted I will share the photographs with you. Since I am at it, I will ask you this question (I have already seen information, but I like to know your opinions) before with the simple swift I had enough, but they recommend a good one and last generation network cables. Any suggestions? in second hand, physical store, they have had one with quite a few ports for a while and I understand that it must be 100-1000gbs. You will tell me.

@jordiqui
Look at the server and look at the wardrobe. My joy in a well.....

Although I'm late in sharing, I bought a gaming laptop and eventually had to change it to a desktop. Because, regardless of what people say, gaming is better on a desktop. It heats up less and has better performance from my point of view, of course.

@_Neptunno_
Hello dear friend, yes it would be from scratch. The Corsair tower and DDR5 latency issues because I don't know when and they will surely be very expensive, so 4 (at least 32gb). I want to make it clear that it's because they have released these processors for that socket, otherwise I would have gone straight for a Ryzen 9. But anyway, I see that it's going to cost a lot of money. Anyway, soon we will see ourselves sharing and validating tests. To get the most out of the cores of the graphics card and, if possible, overclocking it, what recommendations do you have?