Posts by walli

21) Message boards : Science : New experiment PADLS for distributed computing (Message 9179) Posted 14 Jul 2022 by walli Post: I'm going to close the gaps between #170 and #180.
22) Message boards : Science : New experiment PADLS for distributed computing (Message 9173) Posted 13 Jul 2022 by walli Post: I just started to process #173 and #175. Let's see if my results overlap with yours (which should happen then because I process #173 not just up to 11.350.633 but 12.000.000, so probably most if not all results from #174 should be found in my #173 result file). And yes, I chose my values big enough so that I don't have to fiddle with partitioning of larger intervals (1.000.000 x 500 gives me some headroom). And using smaller values like 50.000 or 100.000 on my 24+ core CPUs also doesn't make much sense, this will cause too much overhead. Maybe 500.000 will bring a better balance (so that I won't waste nearly one full 1-million-cycle at worst), but I haven't tested that yet... I'll do that after the PrimeGrid challenge. Time to get some sleep now, good night :).
23) Message boards : Science : New experiment PADLS for distributed computing (Message 9169) Posted 13 Jul 2022 by walli Post: @Demis Thank you for reviewing my results and telling me that I didn't make any mistakes :). I added a few more test cases before I've read the latest comments on this forum, see the #174 subdirectory of my cloud storage. I understand that the larger the overhang of the interval (which is to be processed) is, the more results (incl. ricochet solutions) will be found (see e.g. result_2000000_001_16.txt). @Natalia I left out a few intervals (<= 10 millions) for your turtle, but no problem, I'm going to process them as well :).
24) Message boards : Science : New experiment PADLS for distributed computing (Message 9137) Posted 11 Jul 2022 by walli Post: I go up to #250, then I'll take a break because I need my computers for PrimeGrid.
25) Message boards : Science : New experiment PADLS for distributed computing (Message 9134) Posted 11 Jul 2022 by walli Post: @Demis I've just sent you a private message with the download link to my results in case you need them for review/verification. If you want me to double check intervals that have already been processed by someone, or if you need further information about my setup, just let me know. :)
26) Message boards : Science : New experiment PADLS for distributed computing (Message 9117) Posted 9 Jul 2022 by walli Post: С другой стороны; говорят, что существует бесплатная аренда компьютера (так называемые облачные вычисления). As I mentioned earlier, these cloud computers (virtual servers, VPS, ... whatever poeple call it. "cloud" is only a fancy buzzword for everything that's not operated directly by yourself and your own infrastructure (you outsourced it)) are not really free. Providers/Hosters usually grant you free access for a period of time only, like a few days or weeks for test purposes or as a promotion offer. As far as I know only big companies like Google, Oracle, Amazon, Microsoft etc. give you 1 "cloud computer" for free for up to a year (or even forever). But these are small machines with only 1 or maybe 2 CPU cores and other limitations (like small amounts of RAM and disk space and no Windows OS) so we are just talking about a second turtle :(. It seems to be the same with the Russian providers you linked above. But hey, congratulations that you got access to another computer now :).
27) Message boards : Science : New experiment PADLS for distributed computing (Message 9111) Posted 8 Jul 2022 by walli Post: If we are talking about servers in terms of "hardware", then servers are computers just like desktops, workstations or laptops. The only difference is that most servers are not connected to a monitor and/or keyboard/mouse (they are operated in headless-mode/remotely). So basically you can buy a computer, configure it as a "server" (which means that you install software that "serves" all kind of services to users or other computers on the network) and let it do its job from the corner of your room :). If we talk about quality, yes, good servers are more expensive (e.g. from HP, Dell, Fujitsu, ...), but they are made of better components and can have special features you need in special environments. If we talk about servers in terms of "software", we talk about the operating system (like Windows Server, Linux etc.) and the applications who offer services to the public like a webserver (which hosts websites) or a mailserver (that sends and receives emails) or a boinc-server or... And most servers are installed without a graphical user interface, they are configured and controlled via command line. Because I cannot give you access to my private computers at home, I could only rent a computer/server somewhere and let you use it, and that's not just the electricity fees :). У нас в России подорожало не только электричество, подорожало всё. It's the same here in Germany, everything became extremely expensive.
28) Message boards : Science : New experiment PADLS for distributed computing (Message 9108) Posted 8 Jul 2022 by walli Post: Давайте проще: вы можете дать мне компьютер в удалённое управление? Unfortunately no, because I would then have to pay for this server myself. I'm already curious to see what your second test of #174 will bring up. By the way, I'm going to expand my range to #240.
29) Message boards : Science : New experiment PADLS for distributed computing (Message 9104) Posted 8 Jul 2022 by walli Post: @Cloud Computing: There is 3 kind of servers: virtual servers, virtual dedicated servers and real (bare metal) servers. Sometimes they have different names like VPS, VServer, ... Bare metal/real servers are the most expensive ones, they are computers somewhere in a data center and you can do whatever you want with them, they're all yours (like your home computer). Virtual servers are the cheapest ones. They are basically virtual machines running on a real server, but depending on the available resources, a hoster/provider can run many of them at the same time and you share the resources with other customers, especially the CPU(s). Virtual dedicated servers are in between, but they are like virtual servers where you have dedicated CPU cores/threads on the virtual server so you don't share the them with other customers. Hoster usualley offer you free tiers/trials, like 7 days, 30 days or some even up to 12 months. I know only a few big (US) players who give you virtual servers permanently for free, like Oracle, but they offer you only low performance machines with 1 shared CPU core and not much RAM or disk space... And they generally run Linux, *BSD or similar, but not Windows (because of license fees). So yes, we are at a dead end here :(. @#174: I processed this interval with different configurations, but the results were always the same (you will find them in the download location in a subfolder). The first value in the filename is the setting from config.txt, the second value is the number of loops in the .bat file (should be irrelevant) and the last one is the number of threads I used.
30) Message boards : Science : New experiment PADLS for distributed computing (Message 9099) Posted 7 Jul 2022 by walli Post: Hm, ok, that's possible. What I already observed was that when I used different thread configurations, the squares in the result files were in different order, but that looks reasonable. At the moment my config looks like: - "15 1000000" in "config.txt" - "MaxThreads" is set to 22/30 on a 24/32 core machine - I increased the loop in "PADLS_rule15.bat" to "... (1,1,500)..." so I don't have to worry about large ranges/intervals. I could decrease the value in "config.txt" and increase the loop and test again.
31) Message boards : Science : New experiment PADLS for distributed computing (Message 9095) Posted 7 Jul 2022 by walli Post: У меня нет результатов каждого интервала, обсчитанного мной, отдельно. Зато есть общий файл с моими результатами. Я добавила к моим результатам ваши результаты из интервала №174 и проверила объединённый файл на дубликаты. Ok, understood. Один квадрат в ваших результатах оказался оригинальным, и это странно. Hmm, I have no idea what could have happened, but I also don't know how the programs work internally. Do you know if there are any bugs in the software? Я добавила этот квадрат в новую порцию результатов. Но это не очень большая проблема. Ok. Я не поняла, о чём вы здесь говорите. I just wanted to say that processing the ranges/intervals is a manual job and therefore it's easy to make mistakes (and it's also possible that our computers are broken and return wrong results too). Because of this many projects run double checks (e.g. they process the same ranges/intervals more than once and on different computers) to make sure that the results are valid and nothing is missing (like in our case). I'm aware that we only have very limited resources here on this project and so we should not waste them too much :). I don't want to process all ranges/intervals again, I was just curious if we already run into problems because of missing results. And we did, but i don't have a clue on "how important" the missing results are...? And of course I don't want you to process ranges on your own computer in vain, I'm going to leave out all ranges <= 10M again. The question is if it really makes sense to use your computer to process ranges or if you have more important things to do where you need your computer for and don't get it blocked by processing these ranges? Unfortunately, your computer is pretty slow by today's standards, but we are unable to get you something better :(. walli
32) Message boards : Science : New experiment PADLS for distributed computing (Message 9093) Posted 7 Jul 2022 by walli Post: I know that there are a few intervals that you already processed yourself. The reason is that I wanted to ask if you got the same results (I couldn't do it sooner because the forum/server was down for about a day). And as it shows, it's not the case: "К тому же, у вас какие-то странные результаты в этом интервале, они не совпадают с моими результатами." What is the problem, did you get more or less results than I did? Or completely different results (e.g. because someone made a mistake and took the wrong values for start.txt and what.txt)? ... The manual process is very error prone and there is no double checking. By the way (and just to give you some numbers): Depending on the computers I use for you experiment, it takes about 24-37 minutes to check a 1 million range :).
33) Message boards : Science : New experiment PADLS for distributed computing (Message 9082) Posted 4 Jul 2022 by walli Post: 180-200 should be finished by tomorrow evening, I'll expand my range to #220.
34) Message boards : Science : New experiment PADLS for distributed computing (Message 9070) Posted 2 Jul 2022 by walli Post: Ok, understood. :) I updated my cloud software yesterday so it's possible that it had a little hiccup when you checked for files.
35) Message boards : Science : New experiment PADLS for distributed computing (Message 9065) Posted 2 Jul 2022 by walli Post: I don't know what you mean with: "Я не нашла по ссылке на ваше Облако файла с мощностями диапазонов.". Could you please describe this in other words for my translator? :) Is there a file missing that you need?
36) Message boards : Science : New experiment PADLS for distributed computing (Message 9058) Posted 2 Jul 2022 by walli Post: Ok, I think I'm going to close the gaps between #180 and #200.
37) Message boards : Science : New experiment PADLS for distributed computing (Message 9057) Posted 1 Jul 2022 by walli Post: Hi Natalia, I've been very busy in the last months, but I can help out a little bit again. Since I've lost track of the ranges that have already been processed, just tell me a few intervals you are interested in. walli
38) Message boards : Science : New experiment PADLS for distributed computing (Message 8363) Posted 28 Mar 2022 by walli Post: Hi Natalia, @Excel: It seems that you have Excel 2003 installed. I'm sure that it can also handle CSV files, but it might be tricky to load/import them (Drag&Drop doesn't seem to work, maybe Copy&Paste from an editor like Notepad?). @Interval information: Yeah, I stopped at 39.999. Perhaps I could go a little bit farther, but we've already seen that at ~42.500 the gaps/deserts are getting really big. And sorry, I haven't had the time yet for all the other things you asked me... walli
39) Message boards : Science : New experiment PADLS for distributed computing (Message 8306) Posted 23 Mar 2022 by walli Post: 174 800.604 161 1.818.370 199 2.359.253 170 2.864.711 183 4.638.123 194 5.143.952 167 7.379.517 182 8.202.049 158 9.043.027 169 9.706.012 Correct :). I don't know what this Excel message says. I will continue to calculate interval infos up to 35.000 (or maybe 40.000) and create sorted lists/tables when finished.
40) Message boards : Science : New experiment PADLS for distributed computing (Message 8290) Posted 21 Mar 2022 by walli Post: Мы с черепашкой посчитали мощности следующих интервалов №141 - 35728717 КФ СН ДЛК №142 - 10624322 №143 - 26432018 №144 - 44821070 №145 - 9653474 №146 - 18293216 №147 - 12029364 №148 - 62391676 №149 - 119513384 №150 - 37600678 Берём из этого диапазона в обработку интервалы №№ 142, 145. Остальные интервалы свободны. I changed the source code a little bit and wrote some scripts to automate the calculation (on Linux), so expect some info on the intervals within the next few hours/days. Yes, I know that you already asked to implement snapshot support (for all the big intervals above 40.000) in one of your earlier posts, but I haven't looked at the source code in detail yet.

Previous 20 · Next 20