Message boards :
Number crunching :
Mac OS Sierra
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9
Author | Message |
---|---|
W3Perl Send message Joined: 29 Apr 99 Posts: 251 Credit: 3,696,783,867 RAC: 12,606 |
r3588 was the old version. The new version was in the Intel-mod-1-24.zip download and named MBv8_8.22r3605_avx2_x86_64-apple-darwinBTW, did you test the last version of the CPU AVX2 App?Running app with command : MBv8_8.22r3588_avx2_x86_64-apple-darwin KWSN-Darwin-MBbench v2.1.08 Running on domisses-MacBook-Pro.local at Wed Feb 1 10:26:19 2017 --------------------------------------------------- Starting benchmark run... --------------------------------------------------- Listing wu-file(s) in /testWUs : 18dc09ah.26284.16432.6.33.125.wu reference_work_unit_r3215.wu Listing executable(s) in /APPS : MBv8_8.22r3605_avx2_x86_64-apple-darwin Listing executable in /REF_APPs : MBv8_8.22r3587_ati5-SoG_ssse3_x86_64-apple-darwin --------------------------------------------------- Current WU: 18dc09ah.26284.16432.6.33.125.wu --------------------------------------------------- Running default app with command : MBv8_8.22r3587_ati5-SoG_ssse3_x86_64-apple-darwin -device 0 834.93 real 48.56 user 80.94 sys Elapsed Time: ………………………………… 835 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 1326.32 real 1322.93 user 1.34 sys Elapsed Time : ……………………………… 1326 seconds Speed compared to default : 62 % ----------------- Comparing results Result : Strongly similar, Q= 99.48% --------------------------------------------------- Done with 18dc09ah.26284.16432.6.33.125.wu. Current WU: reference_work_unit_r3215.wu --------------------------------------------------- Running default app with command : MBv8_8.22r3587_ati5-SoG_ssse3_x86_64-apple-darwin -device 0 454.47 real 34.86 user 51.23 sys Elapsed Time: ………………………………… 455 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 966.12 real 963.23 user 0.87 sys Elapsed Time : ……………………………… 966 seconds Speed compared to default : 47 % ----------------- Comparing results Result : Strongly similar, Q= 99.50% --------------------------------------------------- Done with reference_work_unit_r3215.wu. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Current WU: 18dc09ah.26284.16432.6.33.125.wuFrom earlier, https://setiathome.berkeley.edu/forum_thread.php?id=80359&postid=1842860#1842860 Current WU: 18dc09ah.26284.16432.6.33.125.wuNice. If you could, try placing r3366 in the REF_APPs folder and run them head to head. Then place the Stock CPU App in REF_APPs and compare it with r3605. The Stock App at Beta is here http://boinc2.ssl.berkeley.edu/beta/download/setiathome_8.05_x86_64-apple-darwin You might have to set the execute bit on the downloaded file, cd to the folder and run, sudo chmod a+x setiathome_8.05_x86_64-apple-darwin See how they compare... |
W3Perl Send message Joined: 29 Apr 99 Posts: 251 Credit: 3,696,783,867 RAC: 12,606 |
Sounds good :) --------------------------------------------------- Starting benchmark run... --------------------------------------------------- Listing wu-file(s) in /testWUs : 18dc09ah.26284.16432.6.33.125.wu reference_work_unit_r3215.wu Listing executable(s) in /APPS : MBv8_8.22r3605_avx2_x86_64-apple-darwin Listing executable in /REF_APPs : MBv8_8.06r3366_avx_x86_64-apple-darwin --------------------------------------------------- Current WU: 18dc09ah.26284.16432.6.33.125.wu --------------------------------------------------- Running default app with command : MBv8_8.06r3366_avx_x86_64-apple-darwin 1529.77 real 1522.93 user 1.39 sys Elapsed Time: ………………………………… 1530 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 1328.03 real 1321.80 user 1.40 sys Elapsed Time : ……………………………… 1328 seconds Speed compared to default : 115 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% --------------------------------------------------- Done with 18dc09ah.26284.16432.6.33.125.wu. Current WU: reference_work_unit_r3215.wu --------------------------------------------------- Running default app with command : MBv8_8.06r3366_avx_x86_64-apple-darwin 1100.40 real 1094.54 user 1.03 sys Elapsed Time: ………………………………… 1101 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 964.49 real 959.79 user 0.86 sys Elapsed Time : ……………………………… 964 seconds Speed compared to default : 114 % ----------------- Comparing results Result : Strongly similar, Q= 99.98% --------------------------------------------------- Done with reference_work_unit_r3215.wu. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Well.....it certainly appears as though the avx2 App is faster. I'd like to see it run with one of the current Non-VLAR BLC tasks. It also appears to be producing the same results as the avx App. It would be nice to compare the avx2 App to the Stock App to see if the results are very similar. I'm pretty sure the speed will be notable, look at this comparison; https://setiathome.berkeley.edu/workunit.php?wuid=2417169435 Stock Intel(R) Core(TM) i7-4771 CPU @ 3.50GHz = 5,402.06 AVX Intel(R) Core(TM) i7-6700HQ CPU @ 2.60GHz = 2,523.13 It might be time to switch your machine over to ATI r3610 and CPU r3605. |
W3Perl Send message Joined: 29 Apr 99 Posts: 251 Credit: 3,696,783,867 RAC: 12,606 |
Well.....it certainly appears as though the avx2 App is faster. I'd like to see it run with one of the current Non-VLAR BLC tasks. Yes you're right. Good news :) Running on domisses-MacBook-Pro.local at Thu Feb 2 11:18:36 2017 --------------------------------------------------- Starting benchmark run... --------------------------------------------------- Listing wu-file(s) in /testWUs : 18dc09ah.26284.16432.6.33.125.wu reference_work_unit_r3215.wu Listing executable(s) in /APPS : MBv8_8.22r3605_avx2_x86_64-apple-darwin Listing executable in /REF_APPs : setiathome_8.05_x86_64-apple-darwin --------------------------------------------------- Current WU: 18dc09ah.26284.16432.6.33.125.wu --------------------------------------------------- Running default app with command : setiathome_8.05_x86_64-apple-darwin -verb -st -nog 1894.32 real 1886.08 user 2.29 sys Elapsed Time: ………………………………… 1895 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 1544.66 real 1538.45 user 1.53 sys Elapsed Time : ……………………………… 1544 seconds Speed compared to default : 122 % ----------------- Comparing results Result : Strongly similar, Q= 99.99% --------------------------------------------------- Done with 18dc09ah.26284.16432.6.33.125.wu. Current WU: reference_work_unit_r3215.wu --------------------------------------------------- Running default app with command : setiathome_8.05_x86_64-apple-darwin -verb -st -nog 1752.79 real 1745.91 user 1.49 sys Elapsed Time: ………………………………… 1753 seconds --------------------------------------------------- Running app with command : MBv8_8.22r3605_avx2_x86_64-apple-darwin 980.02 real 974.92 user 1.01 sys Elapsed Time : ……………………………… 980 seconds Speed compared to default : 178 % ----------------- Comparing results Result : Strongly similar, Q= 99.83% --------------------------------------------------- Done with reference_work_unit_r3215.wu. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
If anyone with a newer Mac with AVX2 would like to test the new CPU App post your Mac Specs and I'll post a package for your machine. I'd like to monitor how the new App preforms. |
W3Perl Send message Joined: 29 Apr 99 Posts: 251 Credit: 3,696,783,867 RAC: 12,606 |
Up and running...including the Intel GPU. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
It looks as though the minute the iGPU started the CPU tasks started taking over twice as long. It doesn't appear using the iGPU would be advantageous even if it did produce correct results. I removed the last iGPU test App awaiting another version as the last version wasn't any improvement. One has to wonder how much work is being lost by those running the iGPU instead of just using the CPU. |
W3Perl Send message Joined: 29 Apr 99 Posts: 251 Credit: 3,696,783,867 RAC: 12,606 |
It looks as though the minute the iGPU started the CPU tasks started taking over twice as long. It doesn't appear using the iGPU would be advantageous even if it did produce correct results. I removed the last iGPU test App awaiting another version as the last version wasn't any improvement. One has to wonder how much work is being lost by those running the iGPU instead of just using the CPU. iGPU tasks require around the same amount of time to process than CPU tasks....but they are freezing others CPU tasks (even if I have 4 out of 8 core free). I will finish the current iGPU tasks and will cancel iGPU tasks from the web interface. |
Urs Echternacht Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 |
Intel has changed the clock domain for iGPU with their Skylake CPUs. Its now bound to the baseclock (L3-cache, RAM, iGPU) instead of the DMI-connection (chipset and PCIe). I do see the same behaviour on Linux on a KabyLake i7 : if iGPU in use CPU tasks are slowed down to a fourth, else the CPU cores without iGPU usage are fast. Doesn't matter if one or more CPU threads are in use, if iGPU used for crunching the CPUs are slowed. The speedloss is bigger than running 8 threads on CPU. The "sweet spot" seems to be running 6 or 7 threads without iGPU. _\|/_ U r s |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13746 Credit: 208,696,464 RAC: 304 |
Intel has changed the clock domain for iGPU with their Skylake CPUs. Its now bound to the baseclock (L3-cache, RAM, iGPU) instead of the DMI-connection (chipset and PCIe). It's not related to thermal or power limits at all? Grant Darwin NT |
Urs Echternacht Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 |
It's not related to thermal or power limits at all?This is a Mac thread so lets keep this loosely related short: Seems not related (TDP lifted to 65Watt, thermal below) or i've not found the BIOS setting to decouple the iGPU ... if there is one. (legend: C = degree Celsius) 6 or 7 threads CPU @ 3.6GHz temps no iGPU 60 C - 65 C 1 - 7 threads CPU @ 3.6GHz temps with iGPU 48 C - 52 C If iGPU in use CPU is somehow throttled. Bottleneck ? _\|/_ U r s |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Intel has changed the clock domain for iGPU with their Skylake CPUs. Its now bound to the baseclock (L3-cache, RAM, iGPU) instead of the DMI-connection (chipset and PCIe). Seems they made the issue worse. The 4000 series CPUs would only be slowed by about 50% when using the iGPU. :( SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13746 Credit: 208,696,464 RAC: 304 |
If iGPU in use CPU is somehow throttled. Bottleneck ? Either bottleneck (CPU/iGPU sharing caches & data paths?), or power related- Power limit for complete package; power required to run iGPU reduces available power for CPU? Or both? Grant Darwin NT |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
If iGPU in use CPU is somehow throttled. Bottleneck ? Here are some posts starting from 2014 about the same issue. A journey: iGPU slowing CPU processing iGPU tuning Loading APU to the limit: performance considerations - ongoing research I believe the leading theory was L3 cache contention when it was being looked into last. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Urs Echternacht Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 |
First i will try to divide power different. Limit iGPU at lower wattage and give CPU more. Lets see how that works. _\|/_ U r s |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.