New OpenCL application for AMD GPUs
Message boards :
News :
New OpenCL application for AMD GPUs
Message board moderation
Author | Message |
---|---|
Send message Joined: 23 Apr 21 Posts: 87 Credit: 118,916,217 RAC: 143,941 |
Any plans for NVIDIA? must be thinking of Milkyway. Asteroids has always had CUDA apps for Nvidia. but the OpenCL apps for AMD are what's new here. |
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
BTW I get a 404 error on that link you sent meThat's because he used Bing, I've never found anything on Bing but adverts, it's made by Microsoft, and nothing by Microsoft ever works. Try Google, or even better, Duckduckgo (then you get to tell people to duck it!) |
Send message Joined: 22 Nov 17 Posts: 159 Credit: 13,180,518 RAC: 0 |
BTW I get a 404 error on that link you sent meThat's because he used Bing, I've never found anything on Bing but adverts, it's made by Microsoft, and nothing by Microsoft ever works. Try Google, or even better, Duckduckgo (then you get to tell people to duck it!) LOL “The good thing about science is that it's true whether or not you believe in it.” ― Neil deGrasse Tyson |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
Now it's a week after the parallel work progressed on my (AMD integrated graphics card) octcore, i get like 700 points (the task uses 0,1 cpu) during a night session (of eleven hours) whilst Einstein@home (binary pulsars, using 0,5 cpu) would add 15'000 . The Task jump in about 3 percent every 100 seconds. I read in older posts in other forums that the parallel computing for asteroids ("rocky cookies") wasn't showing very significant performance increase, some few percent in some way. So i keep the computer trying to get the parallel computing to swing, since the performance could become far more competitive compared what guys at einstein@home or folding@home provide, what they make gpu's capable of.
|
Send message Joined: 1 Jan 14 Posts: 302 Credit: 32,671,868 RAC: 0 |
Last modified: 23 Sep 2023, 14:42:33 UTC Now it's a week after the parallel work progressed on my (AMD integrated graphics card) octcore, i get like 700 points (the task uses 0,1 cpu) during a night session (of eleven hours) whilst Einstein@home (binary pulsars, using 0,5 cpu) would add 15'000 . The Task jump in about 3 percent every 100 seconds. I read in older posts in other forums that the parallel computing for asteroids ("rocky cookies") wasn't showing very significant performance increase, some few percent in some way. So i keep the computer trying to get the parallel computing to swing, since the performance could become far more competitive compared what guys at einstein@home or folding@home provide, what they make gpu's capable of. It's unfair to compare "points", most people call them credits, between projects as each project has it's own basis for awarding them. Gving more credit per task often means more users coming to your project which can push a project to the very brink of failure or cause them to spend money to upgrade the hardware that they would rather use for other things. It's best to pick the project you crunch for because you like what they are doing and think it's useful to what you think is a good thing and stop worrying about the credits or "points" each project gives out.. As a person with over 1 Trillion total Boinc credits or "points" I can tell you that all those "points" and $10US will get me a cup of coffee at a Starbucks coffee shop but without the $10US I am not getting anything!! |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
@mikey : I placed the credit amount to show the assume-able output of the graphics card potential, how much improval could be achieved in the upcoming months. Participators «gathering as many points as possible» arent deeply interested in science progress or stabilisation of a project, we would agree on that. So what a credits comparisation would you have? about 1 : 20 like my octcore? @other aster(dr)oids reading aswell: If the recently available parallel programm (open CL) were also in use for (integrated AMD gpus for) windows 7 ("on own risk") this would be nice, why are too many people concerned about so-called safety issues about boinc, like «win 7 does not receive any support anymore, we guess that we have to abandon it», whilst many people have complications with windows 10 in common uses. Cybercrime should be punished heavier, because it's making harmed people distrust, inflicting damages, to technology that already isnt explained detailed enough. I wonder how the float point unit (fpu) correlates with the gpu, but i am far away from being an expert understanding how computers calculate in most aspects. I have like gremlins lurking in my pc's without having much in there that would attract things like malware. I try too hard. Furtherly i would like too see which WU/thread works on which asteroid, if the asteroid number were showed. I would apprechiate if i get some idea what the calculations are about, because with «corona vaccine in rosetta@home» almost two winters ago i felt like having contributed to a project that was rushed in impatiently, just taking away computational potential, (artificial) mRNA doesnt pay off, just worsening people, i would say. So i let my AMDs be AMaDeus in the manner i can. |
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
Last modified: 24 Sep 2023, 15:48:47 UTC As a person with over 1 Trillion total Boinc credits or "points" I can tell you that all those "points" and $10US will get me a cup of coffee at a Starbucks coffee shop but without the $10US I am not getting anything!!Ten dollars for 1 coffee?!? Better making it at home. Probably 10 cents. I think 10 dollars here in McDonalds gets you a burger, fries, and the coffee, and some change. |
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
@mikey : I placed the credit amount to show the assume-able output of the graphics card potential, how much improval could be achieved in the upcoming months. Participators «gathering as many points as possible» arent deeply interested in science progress or stabilisation of a project, we would agree on that. So what a credits comparisation would you have? about 1 : 20 like my octcore?Best to compare your CPU to your GPU here, then compare your CPU to your GPU at Einstein. Unless the GPU Asteroids program gets a lot faster, it's very slow, and better running on the CPUs. why are too many people concerned about so-called safety issues about boinc, like «win 7 does not receive any support anymore, we guess that we have to abandon it», whilst many people have complications with windows 10 in common uses.I've never known Windows 7 to catch fire. What is this "risk"? I don't open weird attachments in emails. If I download something dodgy on purpose from something like Piratebay I scan it with an AV or send it to virustotal. Cybercrime should be punished heavier, because it's making harmed people distrust, inflicting damages, to technology that already isnt explained detailed enough.Some people are fighting it - https://www.youtube.com/@ScammerPayback/videos and I reported this website to every organisation under the sun, they used to sell non existant eggs and birds: https://parrotsaviaryfarm.co.uk/ - looks like it worked, the site has gone. I went after him from multiple angles, including informing his bank, hopefully they got some money back from him and returned it to unsuspecting customers. I have like gremlins lurking in my pc's without having much in there that would attract things like malware. I try too hard.What gremlins do you have? Furtherly i would like too see which WU/thread works on which asteroid, if the asteroid number were showed. I would appreciate if i get some idea what the calculations are about,Agreed, perhaps there is or could be a list of asteroids, and we could use the task name in boinc to find it. because with «corona vaccine in rosetta@home» almost two winters ago i felt like having contributed to a project that was rushed in impatiently, just taking away computational potential, (artificial) mRNA doesnt pay off, just worsening people, i would say.Their skycovione is meant to be way better than the others. But they're still using the others aswell as far as I know. Maybe there isn't enough being made? I'm not sure, because I never got vaccinated, I don't worry about a virus. |
Send message Joined: 14 Feb 23 Posts: 1 Credit: 4,132 RAC: 0 |
|
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
|
Send message Joined: 22 Nov 17 Posts: 159 Credit: 13,180,518 RAC: 0 |
Any support for Mac OS X ?? Many of us have AMD cards ... Soon. But only for the CPU tasks. For now. Sorry, only computers can run it. Why do people buy those incompatible overpriced things? LOL Don't listen to Peter (Lamberto). He is well known for his British humor. “The good thing about science is that it's true whether or not you believe in it.” ― Neil deGrasse Tyson |
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
Nah I just call a spade a spade. Why should programmers have to write everything twice just because Apple want to be different? And it's spelt humour :-PSorry, only computers can run it. Why do people buy those incompatible overpriced things?LOL I have installed the pro driver on one of my 280X machines, we shall see.... |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
@(Peter) Umberto : The gremlins in my computer systems arent too abbreviate from the classic 90ies definition (mostly like https://www.lemonamiga.com/games/list.php?list_publisher=Gremlin%20Graphics in some way), or what types of those electric creatures would you know or describe, like glitches? Neither i download any doubtable stuff from the internet. Neither i would get any corona vaccination, the newer vaccinations are far less good than older ones, not just due to mRNA «technology» (many two legged freaks want too early to be proposed by governments, i›d say), coronavirus would be as unimportant as Adenoviruses. @other participators aswell: But the problem issue would be all the dirt we get into our bodies we cant get used to in a fast manner, not just bad genetic strains (that nature would attempt to recycle, also in "us"). I want to find out what benefit a networking card (next to some gpu card on the mainboard/motherborad) would have, if this would rearrange the processing for boinc in some ways. So am i the only person here that rather uses night current, when the weight on the electric grid is (or would be) less heavy? And does the contribution to A@H also work in the Edgeworth torus (the «Kuiper Belt») to thinnen the region where a planet nine could be (found), or learning the shape of the comet discs out there? Or is asteroids@home mostly about the asteroid torus («asteroid belt) or perhaps some Jupiter (greeks and) trojans? or about centaurs? because it is not (re)named comets@home? I would prefer to donate my pc potential to «asteroids» outside of Jupiter, anybody having the same preference about this, maybe even the guys from NASA? Guys like Josef Durech, Ranko (unknown surname) or Georgi Vidinski could inform us why the chose certain asteroid groups (rocky cookies?) so we can learn why we support those asteroid details. GAIA would have found many new objects outside of Neptunes trace? Or are we supposed to accept the work believeing this must be enough, fitting the entire process? would we let calculate rather efficient if we just trust in the guys (hard work) behind A@H? |
Send message Joined: 22 Nov 17 Posts: 159 Credit: 13,180,518 RAC: 0 |
Simon Jäkle, It seams to me you haven't red any of our informative materials at all, stating out loud such ignorance on the project goals. But it's OK, I'll help you with that right away. If you haven't seen this article, may be now is the time to take a look on it: Unfortunately it is still in Chech only, but you can use bowser's tool to translate it: If you dig a bit more you will find the DAMIT - Database of Asteroid Models from Inversion Techniques (English). There is huge amount of information (including scientific publications) on the subject. And last but not least if you still have that feeling and you're not satisficed, you always are welcome to join one of the so many other projects out there. “The good thing about science is that it's true whether or not you believe in it.” ― Neil deGrasse Tyson |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
Calculation evidence continuing on (search for) planet nine, in a weird-seeming retrograde orbit(!), this paper (updated september 25th) https://astrobiology.nasa.gov/news/caltech-researchers-find-evidence-of-a-real-ninth-planet/ shows (perhaps see Mike Browns ted talk https://www.youtube.com/watch?v=2UZKME9WP9M ), i would kind of love to know if i am contributing to a grid working on that (without seeming like in a cherry picker herd). Thank you Georgi Vidinski for at least not ignoring my post, i do not want to thinnen what your work would be, i struggle with computers and/or parts of the internet distorting me, denying stuff and pages that could make me look like a bad or sleazy handycrafter. I just have a lot of stuff i could not improve in the latter. |
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
@Simon Jäkle - I don't have a vaccine either. I didn't bother because I don't worry about small risks, I don't wear a seatbelt either. But my sister's family of four have caught covid three times after being vaccinated fully. Yet I've caught it zero times after no vaccine. We've both met the same number of people we could have caught it from. Then there's the guy who runs Universe@Home who caught it before and after the vaccine, with the same symptoms, yet the 2nd one was supposed to be the mild virus. Makes you wonder if the vaccine does more harm than good. The first instance I heard of covid was a cruise ship where everyone caught it, but 80% of people didn't even get symptoms. @ Georgi Vidinski - Did I read something once about the possibility of mining asteroids on the way to moon/mars missions? And we were studying the composition of them? I skimmed through the article but am too busy to read it thoroughly. |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
@Lamberto Vitaly: Many people contesting won't try to get a big picture of topics, a kind of connected, interconnected, they blame that work on "too many changes" happening. How cyclopsic one-eyed must a human be (or act?) to "follow virusmania" as they did in the winters of 2020/2021 and 2021/2022 ? how can they wage for their individual situation how to leave the world outside? I would care about the big picture even more if the progresses were rather stable, so i still wonder if A@H builds up space knowledge in contiuitive or consecutive way, by being constructive enough with the current used for it. Instead of being an economic threat by sucking away current produceing little new data, i don't know if A@H is efficient in sense of building along without provoqueing weak spots or hazards later on. For as long as i dont have an equivalent or equation like "A@H calculates the trajectories of three asteroids a day using 30kilowatts of energy (1 Newton would be 1 Joule)" and this is not too much for the environment, i am letting my computers run "having to be faithfully of", i dont want to get used to such an "intermediate state", somebody confirm that A@H is not letting occur a hole in the future, because then all our ambitions about A@H would become obsolete. |
Send message Joined: 27 Sep 23 Posts: 1 Credit: 4,814 RAC: 0 |
Last modified: 4 Oct 2023, 9:58:40 UTC Tested with a AMD ROCm on my laptop with Radeon RX 6600M https://asteroidsathome.net/boinc/result.php?resultid=405107770 it looks like mesa rusticl is not capable yet to support boinc? Did somebody managed to make it work with rusticl? Wed 04 Oct 2023 12:34:19 PM EEST | | OpenCL: AMD/ATI GPU 0: AMD Radeon RX 6600M (driver version 3590.0 (HSA1.1,LC), device version OpenCL 2.0, 8176MB, 8176MB available, 9748 GFLOPS peak) Wed 04 Oct 2023 12:34:19 PM EEST | | OpenCL: AMD/ATI GPU 1 (ignored by config): AMD Radeon RX 6600M (navi23, LLVM 15.0.7, DRM 3.49, 6.2.0-33-gen (driver version 23.2.1 - kisak-mesa PPA, device version OpenCL 1.1 Mesa 23.2.1 - kisak-mesa PPA, 8192MB, 8192MB available, 6093 GFLOPS peak) second string looks like a clover? Right now with ROCm I see this power consumption |
Send message Joined: 11 Sep 12 Posts: 12 Credit: 5,798,909 RAC: 2 |
Since Monday 2nd of october i don't have any work units for asteroids@home no more, neighter for the gpu, even einstein@home "refused" to give me alternative work units. Since thursday 5th of october i have at least some einstein work units, but why no ones from A@H ? This lack doesn't have something to do with the project boinc central? Are there too many systems with varieties so the work flow is getting restricted , "safety issues" being assumed?
|
Send message Joined: 14 Jun 23 Posts: 85 Credit: 5,914 RAC: 0 |
|
Message boards :
News :
New OpenCL application for AMD GPUs