Posts by Sunny129
|
1)
Message boards :
News :
Milkyway/Bitcoin Utopia Update
(Message 62218)
Posted 26 Aug 2014 by Sunny129 Post: Sunny...they are two separate campaigns so please don't confuse the two. Their %$20,000 goal is separate from the efforts here. If MW fundraising via cash donations reaches the $40,000 goal on it's own, then the BU fundraising will go to secure the project for the future. my mistake...thanks for clearing up the confusion. i just now realized that BU is an actual independent DC project that just so happens to be helping the MW@H project out by having its own fundraiser. |
|
2)
Message boards :
News :
Milkyway/Bitcoin Utopia Update
(Message 62211)
Posted 26 Aug 2014 by Sunny129 Post: i guess the people at BU didn't know that ~$19K had already been raised by 7/26/14 (the starting point of their graph), and that the total was up to $23,481.92 by 8/4/14...so that red line needs to be shifted up accordingly... |
|
3)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61382)
Posted 11 Mar 2014 by Sunny129 Post: At that point my rac should break the million per day mark, at least until I have to move to another project. which Einstein tasks are taking 30+ hours to crunch on your 7970? the BRP4 Perseus Arm Survey tasks only take a few hours each on my 7970, and that's running them 3 at a time. the BRP4G Arecibo GPU tasks only take ~1 hour each on my 7970, and they also run 3 at a time. the new FGRP3 GPU tasks are quite immature, but they only take ~3 hours each on my 7970 (as opposed to 11 hours for the CPU versions of the same tasks), and i run those 6 at a time. |
|
4)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61342)
Posted 7 Mar 2014 by Sunny129 Post: I was going to post a time for my GTX 260 c216, but it only seems to be crunching MW v1.02 WUs. perhaps that host is using either an app_info.xml file or an app_config.xml file that specifies the use of (has an entry for) the Modified Fit application. in other words, perhaps that user is forcing his GTX 260 to crunch the Modified Fit tasks (and thus forcing it to use the OpenCL-coded application binary, since there is no non-OpenCL-coded version of the Modified Fit application). perhaps you should PM him and see who he is able to receive Modified Fit tasks. this also makes me wonder if what i told Arion a few days ago - that an HD 3870 cannot crunch Modified Fit tasks due to a lack of OpenCL support - is true...perhaps it can be forced to crunch Modified Fit tasks as well. let me know what you find out about the GTX 260 host. |
|
5)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61320)
Posted 4 Mar 2014 by Sunny129 Post: i don't know if it was mentioned earlier in this thread or not (i haven't followed it from the beginning), but at some point we discovered that some 213.76-point tasks take x seconds to complete, while other 213.76-point tasks take 1.5x seconds to complete. thus the OP decided that' we'd use only the longer-running 213.76-point tasks, that is, the ones that take approx. 50% longer to complete. i suppose he could have just as easily told everyone to use 5 of the shorter-running tasks for their averages, or perhaps 3 short ones and 2 long ones. fortunately, for simplicity's sake, he chose to average 5 of the same kind of task, and that just happened to be the kind that takes 50% longer to complete. its not really about interpretation so much as it is about consistency...if you aren't calculating your average run time using 5 of the longer running 213.76-point tasks, then the contribution will be useless... |
|
6)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61315)
Posted 3 Mar 2014 by Sunny129 Post: nope, times consistently at 50 secs. you have a number of valid 213.76-point tasks that took ~76s to complete (in addition to the ones that only ~50s to complete)...these are the ones we're after. after having averaged the first 5 of your valid longer-running 213.76-point tasks, i got an average run time of 76.5s. |
|
7)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61295)
Posted 2 Mar 2014 by Sunny129 Post: Got your message today about the 3870 times.... I'm running an app_info shown below. What do I need to add or change to run the app you're asking for? unfortunately i don't think you can help the cause with an HD 3xxx series GPU. according to your app_info, you're using v0.82 (ati14), the non-OpenCL version of the Milkyway@Home applicaiton...i'm assuming that you're HD 3870 is not OpenCL-capable, or else your GPU would be using v1.02 (opencl_amd_ati), the OpenCL version of the Milkyway@Home application. to participate in the first place, you'll note that its the tasks that earn 213.76 points that we're after here...and the only application that sends out 213.76-point tasks is the Milkyway@Home Separation (Modified Fit) application. if you look at all the different versions of the Modified Fit application, there are no non-OpenCL GPU versions of the application - all of them are OpenCL-based...so i don't think the server will give you any Modified Fit tasks (even if you have them selected/enabled in your MW@H web preferences) b/c the project knows that HD 3xxx series GPUs cannot crunch OpenCL versions (the only ones available for either AMD or nVidia GPUs) of the app. |
|
8)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61226)
Posted 24 Feb 2014 by Sunny129 Post: For the 280X and slower units, I see times 70-75 secs so far... looks like most are around ~75s, and i see 2 outliers - a ~70s run time and an ~81s run time...if we average the 8 valid 213.76-point tasks you have so far, we get 74.7675s, or ~75s, which is just about right in line with my 7970's run times when clocked at 1100-1150MHz, give or take a second or two due to small, but irreconcilable factors... |
|
9)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61218)
Posted 23 Feb 2014 by Sunny129 Post: I've got the right card haven't I? no, his is either the HD7950-DC2T-3GD5 or the HD7950-DC2T-3GD5-V2 (different display output arrangement, same everything else). |
|
10)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61211)
Posted 23 Feb 2014 by Sunny129 Post: if you want something that'll substantially outperform a 7970/7970GE, you'll have to upgrade to a 290 or a 290X... oops...you're absolutely right - i completely forgot about the crippled FP64 performance of the Hawaii chip...so yeah, even the 290 and 290X aren't as good as the 7970 for MW@H... 7970 FP64 performance: 925,000,000 cycles per second X 2 instructions per cycles X 2048 streaming processors X 0.25 FP64 ratio = 947.2 GFLOPs R9 280 FP64 performance: 925,000,000 cycles per second (boost clock) X 2 instructions per cycles X 1792 streaming processors X 0.25 FP64 ratio = 828.8 GFLOPs R9 280X FP64 performance: 1,000,000,000 cycles per second (boost clock) X 2 instructions per cycles X 2048 streaming processors X 0.25 FP64 ratio = 1,024 GFLOPs R9 290 FP64 performance: 947,000,000 cycles per second (boost clock) X 2 instructions per cycles X 2560 streaming processors X 0.125 FP64 ratio = 606.08 GFLOPs R9 290X FP64 performance: 1,000,000,000 cycles per second (boost clock) X 2 instructions per cycles X 2816 streaming processors X 0.125 FP64 ratio = 704 GFLOPs |
|
11)
Message boards :
Number crunching :
Benchmark results - times wanted for any hardware, CPU or GPU, old or new!
(Message 61200)
Posted 23 Feb 2014 by Sunny129 Post: both WESand Leonheart appear to be using the same OS and video driver version, but WES is using BOINC v7.2.33, whereas Leonheart is using BOINC v7.0.64...now i can't imagine that has anything to do with the difference in run times, but maybe its worth looking into... Yea, i was tried hard ) Memory OC isn`t important for crunching, and pretty unstable on high clocks. Really nice to see that 7950 can be faster than stock 7970 lol. But OC 280X will beat all variants. don't be fooled by new nomenclature...the decisive advantage, if there is one, won't necessarily go to the 280X without question. it'll trade blows with the 7970/7970GE. you have to remember that a 280X is just a re-badged 7970 w/ a re-tweaked core clock (850MHz vs the 7970's 925MHz), memory clock (1500MHz vs the 7970's 1375MHz), and a boost clock (1000MHz vs the 7970's nonexistent boost clock). otherwise, they use the same 4.31 billion transistor chip built on the same 28nm manufacturing process based on the same GCN 1.0 architecture rated at the same TDP of 250W. they also have the same 3GB of VRAM and the same 384-bit memory bus. the only advantage a 280X seems to have over the 7970GE is a lower TDP (since the 7970GE's TDP is rated a bit higher than the reference 7970), while the 280X's VRAM clock is the same and its core clock and boost clock are slower. i see that Mumak has a Gigabyte 280X on the way that's factory OCed, but if we're going to compare apples to apples, even that card doesn't have the distinct advantage over certain factory OCed 7970's/7970GE's. if you want something that'll substantially outperform a 7970/7970GE, you'll have to upgrade to a 290 or a 290X...don't get me wrong, the 280X is a great card...but if you're looking for an upgrade, and you've already got a 7970, the 280X would be a waste of money b/c you're paying more for a sometimes minimal, and other times nonexistent performance gain. just some food for thought... |
|
12)
Message boards :
Number crunching :
server is out of work?
(Message 60345)
Posted 9 Nov 2013 by Sunny129 Post: it appears that the server is out of work, this according to both the server status page and the messages in my BOINC event log. this behavior started on my host not quite 2 hours ago. hopefully someone can give the server a kick before Monday morning... |
|
13)
Message boards :
News :
New Separation Modified Fit Runs Started
(Message 59785)
Posted 2 Sep 2013 by Sunny129 Post: After I had made those changes I saw, for the first time, under the "primary (default) preferences" that there was a setting "default computer location" which was set to "home". So, I should have read through the settings more carefully and picked up on why the modfits kept coming. A pink or "do better" slip is in order. Lol if your host's default location was left blank (as opposed to set to "home," or "work" or "school" for that matter), your host would have accepted the settings you originally implemented. at any rate, glad you figured it out. |
|
14)
Message boards :
News :
New Separation Modified Fit Runs Started
(Message 59759)
Posted 30 Aug 2013 by Sunny129 Post: i wanted to add that i finally did get invalid from the new run of modified fit tasks. but its the only one so far out of a few hundred that i've crunched from the new run. also, its one of those invalids whose outcome is a "success" as opposed to a "validate error," and its also the kind of invalid that didn't ruin my streak of consecutive valid tasks due to having had "too many errors" occur on various hosts (may have bug). but so far i would venture to say that >99% of the new modified fit tasks are validating on my host. |
|
15)
Message boards :
News :
New Separation Modified Fit Runs Started
(Message 59758)
Posted 30 Aug 2013 by Sunny129 Post: ok, after a 4-hour power outage and an hour of being back up and running, i can now confirm that the new modified fit tasks are validating regularly on my host. i did have one invalid since i've been back up and running, but it turned out to be one of the few remaining BPL tasks from the previous run that i guess hadn't yet filtered out of the server...so no worries there. my platform specifics are Win7 x64, BOINC 7.0.64, 7970 @ 950MHz & 7950 @ 900MHz, Catalyst 13.4. |
|
16)
Message boards :
News :
New Separation Modified Fit Runs Started
(Message 59751)
Posted 29 Aug 2013 by Sunny129 Post: just ran 4 of the new modified fit tasks, and 3 of them validated straight away. the 4th is "validation inconclusive" (checked, but no consensus yet) for now. so i went ahead and started letting them into the queue to confirm that i didn't just get lucky with the first 4 tasks. i'll post up again once they reach the top of my queue and start crunching along side the regular milkyway tasks... it may be worth noting that i'm running Win7 x64 Pro, while KIDH is running WinXP Pro SP3. |
|
17)
Message boards :
News :
Separation Modified Fit v1.26 Release
(Message 59746)
Posted 29 Aug 2013 by Sunny129 Post: Here, too, all these WUs failed (WinXP SP3 x86, BOINC 7.0.64, Radeon 4850). those modified fit tasks were all based on the Broken Power Law model. now that [almost] all of those have been flushed out of the server queue, you should start getting modified fit tasks from the new run Jake just released (provided you're host is still set to accept them). please let us know if these run any better than the previous modified fit tasks. i'll be opting back into modified fit shortly to test the new run as well... |
|
18)
Message boards :
News :
Separation Modified Fit v1.26 Release
(Message 59719)
Posted 28 Aug 2013 by Sunny129 Post: Here too - too many errors. yup...looks like you're getting the same kind of invalids i'm getting - a 50/50 mix of "validate errors" and "completed - can't validate," where the "completed - can't validate" tasks apparently failed due to "work unit error - check skipped." granted, every last one of your invalids are modified fit tasks, whereas a vast majority of my invalids are modified fit tasks with a handful of regular MW@H invalids...but the rgular MW@H invalids are older than the rest, and they've stopped rolling in since having switched to crunching only MW@H tasks (and no modified fit tasks for now). i can confirm this much b/c the number of consecutive valid MW@H tasks continues to climb, while the number of consecutive valid modified fit tasks hardly breaks into double digits before dropping back down to zero and starting again. |
|
19)
Message boards :
News :
Separation Modified Fit v1.26 Release
(Message 59713)
Posted 28 Aug 2013 by Sunny129 Post: well the number of invalids continues to grow...i'm now up to 20 of them. i've made no hardware or driver changes since the new v1.26 app showed up. 19 of the 20 invalids are modified fit tasks. |
|
20)
Message boards :
News :
Separation Modified Fit v1.26 Release
(Message 59707)
Posted 28 Aug 2013 by Sunny129 Post: i had errors at first, but it was my own fault. i run on an anonymous platform, and so the application didn't automatically update for me. i thought there was a relatively easy way to update the application for anonymous platform users too, but i couldn't figure it out of there was one...perhaps i'm just rusty and have a faded memory, b/c its been a while i've had to manually update a BOINC project application. at first i just closed BOINC and changed some entries (i changed 1.24 to 1.26 and 124 to 126) in the app_info.xml, but apparently that doesn't really let the project server know that my host needs the new v1.26 app. in fact, upon restarting BOINC i got a message stating that the app_info.xml file referencing a nonexistent application. now that much made sense, b/c only the regular v1.02 app and the modified fit v1.24 app were in the project directory. so i tried resetting the project, but that didn't help either. what i ended up having to do was close BOINC, remove the app_info.xml file and the application binaries from the project directory, restart BOINC, and reset Milkyway@Home. only then did BOINC download the new v1.26 modified fit app and re-download the regular v1.02 app. well, actually, i don't know for sure if removing the application binaries from the directory was absolutely necessary, but that's neither here nor there. now that i had the new v1.26 app in the project directory, i was able to put my app_info.xml file back into the project directory and restart BOINC. sure enough, everything is working perfectly again, almost error-free...but i'll get to that in just a second. in the mean time, is there an easier way to download an updated application binary for anonymous platform hosts than removing the app_info.xml from the project directory and resetting the project? ok, onto the handful of errors...actually they're invalids. of the 3, one's outcome is a validate error, while the other two's outcomes are "successes," but the validate state is reported as "work unit error - check skipped." this type of error is new to me...any ideas? and yes, these are modified fit tasks...here's a link to my INVALID TASKS. |
Next 20
©2024 Astroinformatics Group