I know that this has been discussed for Mojave - Apple is killing CUDA Support. (As well as OpenCL being "Deprecated"....)

For High Sierra, CUDA is STILL an available option for NVIDIA Cards with the NVIDIA CUDA Driver. I've ALWAYS had the CUDA Driver installed; going back even to my old Hackintosh. (The Hackintosh was Updated to CUDA Driver 8.0.71 on El Capitan 10.11.6.)

Had to reboot the Mac Pro today due to some odd behavior with my built-in WiFi card. After the reboot, I noticed in BOINC's Event Log that CUDA was NOT Present. I was on CUDA Driver 8.0.71 here, as well and the Driver worked months ago before moving to TBar's OpenCL App that I'm on now. Yet, in System Preferences, the NVIDIA CUDA Driver Pane is Present and when Opened showed that I was still on 8.0.71. I chose to Update to the Latest CUDA Driver to see if this would help. (It installed without issue.)

[Event Log - Before CUDA Update:]

Sun Nov 4 12:58:12 2018 | | cc_config.xml not found - using defaults
Sun Nov 4 12:58:12 2018 | | Starting BOINC client version 7.8.6 for x86_64-apple-darwin
Sun Nov 4 12:58:12 2018 | | log flags: file_xfer, sched_ops, task
Sun Nov 4 12:58:12 2018 | | Libraries: libcurl/7.50.2 OpenSSL/1.1.0 zlib/1.2.11 c-ares/1.11.0
Sun Nov 4 12:58:12 2018 | | Data directory: /Library/Application Support/BOINC Data
Sun Nov 4 13:00:13 2018 | | OpenCL: NVIDIA GPU 0: GeForce GTX 1050 (driver version 10.33.0 387.10.10.10.40.105, device version OpenCL 1.2, 2048MB, 2048MB available, 547 GFLOPS peak)
Sun Nov 4 13:00:13 2018 | | OpenCL: AMD/ATI GPU 0: ATI Radeon HD 5770 (driver version 1.2 (Jun 29 2018 18:33:46), device version OpenCL 1.2, 1024MB, 1024MB available, 680 GFLOPS peak)
Sun Nov 4 13:00:13 2018 | | OpenCL CPU: Intel(R) Xeon(R) CPU E5620 @ 2.40GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Sun Nov 4 13:00:13 2018 | SETI@home | Found app_info.xml; using anonymous platform
Sun Nov 4 13:00:13 2018 | | Host name: Andromeda.local
Sun Nov 4 13:00:13 2018 | | Processor: 16 GenuineIntel Intel(R) Xeon(R) CPU E5620 @ 2.40GHz [x86 Family 6 Model 44 Stepping 2]
Sun Nov 4 13:00:13 2018 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 popcnt aes pcid
Sun Nov 4 13:00:13 2018 | | OS: Mac OS X 10.13.6 (Darwin 17.7.0)
Sun Nov 4 13:00:13 2018 | | Memory: 64.00 GB physical, 659.75 GB virtual
Sun Nov 4 13:00:13 2018 | | Disk: 930.71 GB total, 659.51 GB free
Sun Nov 4 13:00:13 2018 | | Local time is UTC -8 hours
Sun Nov 4 13:00:13 2018 | | VirtualBox version: 5.2.18r124319
Sun Nov 4 13:00:13 2018 | | A new version of BOINC is available (7.14.2). <a href=https://boinc.berkeley.edu/download.php>Download</a>
Sun Nov 4 13:00:13 2018 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7952666; resource share 365
Sun Nov 4 13:00:13 2018 | SETI@home | General prefs: from SETI@home (last modified 04-May-2018 09:14:36)
Sun Nov 4 13:00:13 2018 | SETI@home | Computer location: home
Sun Nov 4 13:00:13 2018 | SETI@home | General prefs: no separate prefs for home; using your defaults
Sun Nov 4 13:00:13 2018 | | Reading preferences override file
Sun Nov 4 13:00:13 2018 | | Preferences:
Sun Nov 4 13:00:13 2018 | | max memory usage when active: 22937.60 MB
Sun Nov 4 13:00:13 2018 | | max memory usage when idle: 32768.00 MB
Sun Nov 4 13:00:13 2018 | | max disk usage: 30.00 GB
Sun Nov 4 13:00:13 2018 | | max CPUs used: 8
Sun Nov 4 13:00:13 2018 | | (to change preferences, visit a project web site or select Preferences in the Manager)
Sun Nov 4 13:00:13 2018 | | Running CPU benchmarks
Sun Nov 4 13:00:13 2018 | | Suspending computation - CPU benchmarks in progress
Sun Nov 4 13:00:45 2018 | | Benchmark results:
Sun Nov 4 13:00:45 2018 | | Number of CPUs: 8
Sun Nov 4 13:00:45 2018 | | 3256 floating point MIPS (Whetstone) per CPU
Sun Nov 4 13:00:45 2018 | | 11798 integer MIPS (Dhrystone) per CPU


[Event Log - After CUDA Update:]

Sun Nov 4 18:31:24 2018 | | cc_config.xml not found - using defaults
Sun Nov 4 18:31:24 2018 | | Starting BOINC client version 7.8.6 for x86_64-apple-darwin
Sun Nov 4 18:31:24 2018 | | log flags: file_xfer, sched_ops, task
Sun Nov 4 18:31:24 2018 | | Libraries: libcurl/7.50.2 OpenSSL/1.1.0 zlib/1.2.11 c-ares/1.11.0
Sun Nov 4 18:31:24 2018 | | Data directory: /Library/Application Support/BOINC Data
Sun Nov 4 18:31:24 2018 | | CUDA: NVIDIA GPU 0: GeForce GTX 1050 (driver version 10.0.130, CUDA version 10.0, compute capability 6.1, 2048MB, 1203MB available, 1862 GFLOPS peak)
Sun Nov 4 18:31:24 2018 | | OpenCL: NVIDIA GPU 0: GeForce GTX 1050 (driver version 10.33.0 387.10.10.10.40.105, device version OpenCL 1.2, 2048MB, 1203MB available, 1862 GFLOPS peak)
Sun Nov 4 18:31:24 2018 | | OpenCL: AMD/ATI GPU 0: ATI Radeon HD 5770 (driver version 1.2 (Jun 29 2018 18:33:46), device version OpenCL 1.2, 1024MB, 1024MB available, 680 GFLOPS peak)
Sun Nov 4 18:31:24 2018 | | OpenCL CPU: Intel(R) Xeon(R) CPU E5620 @ 2.40GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Sun Nov 4 18:31:24 2018 | SETI@home | Found app_info.xml; using anonymous platform
Sun Nov 4 18:31:24 2018 | | Host name: Andromeda.local
Sun Nov 4 18:31:24 2018 | | Processor: 16 GenuineIntel Intel(R) Xeon(R) CPU E5620 @ 2.40GHz [x86 Family 6 Model 44 Stepping 2]
Sun Nov 4 18:31:24 2018 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 popcnt aes pcid
Sun Nov 4 18:31:24 2018 | | OS: Mac OS X 10.13.6 (Darwin 17.7.0)
Sun Nov 4 18:31:24 2018 | | Memory: 64.00 GB physical, 657.92 GB virtual
Sun Nov 4 18:31:24 2018 | | Disk: 930.71 GB total, 657.68 GB free
Sun Nov 4 18:31:24 2018 | | Local time is UTC -8 hours
Sun Nov 4 18:31:24 2018 | | VirtualBox version: 5.2.18r124319
Sun Nov 4 18:31:24 2018 | | A new version of BOINC is available (7.14.2). <a href=https://boinc.berkeley.edu/download.php>Download</a>
Sun Nov 4 18:31:24 2018 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7952666; resource share 365
Sun Nov 4 18:31:24 2018 | SETI@home | General prefs: from SETI@home (last modified 04-May-2018 09:14:36)
Sun Nov 4 18:31:24 2018 | SETI@home | Computer location: home
Sun Nov 4 18:31:24 2018 | SETI@home | General prefs: no separate prefs for home; using your defaults
Sun Nov 4 18:31:24 2018 | | Reading preferences override file
Sun Nov 4 18:31:24 2018 | | Preferences:
Sun Nov 4 18:31:24 2018 | | max memory usage when active: 22937.60 MB
Sun Nov 4 18:31:24 2018 | | max memory usage when idle: 32768.00 MB
Sun Nov 4 18:31:24 2018 | | max disk usage: 30.00 GB
Sun Nov 4 18:31:24 2018 | | max CPUs used: 8
Sun Nov 4 18:31:24 2018 | | (to change preferences, visit a project web site or select Preferences in the Manager)

[New CUDA Version:]

410.130

-------------------------------------

So, if CUDA Driver 8.0.71 was working fine and was displayed some months ago, what killed the support for it? Did the OS render 8.0.71 inoperable???

Just curious... All seems to be as it should now that CUDA Driver 410.130 is in place.


TL

[EDIT:]

Added Color for more clarity.

One other question... Why is the Dual Xeon CPU System also NOT being shown as CUDA Compatible??? I thought all CPUs could use CUDA.

[EDIT 2:]

My OS Version hasn't changed, on 10.13.6 Ver: 17G65 - 2018-001 Security Update.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

This is from the last time you asked that question;
Message 1951115 - Posted: 21 Aug 2018

Well, CUDA 8 is for El Capitan. For Sierra and High Sierra you need to use the Highest CUDA version. Just keep Updating it in CUDA Preferences until it says Nothing Newer.
There is still the problem with the 2GB GPUs. Don't connect them to a Monitor, use a GPU with at least 3GB of RAM, such as my 1050Ti. Or connect the monitor to an ATI GPU.
The 2GB NV GPUs work fine as long as they are not connected to a Monitor. See this, CUDA version 9.2, compute capability 6.1, 4096MB, 3550MB available
That's how much vRAM OSX uses Just to run the monitor. That's a pretty big chunk out of a 2 GB GPU.

You should have updated to a CUDA 9 driver back when Sierra first appeared. That was some time ago. I doubt CUDA 8 has worked since Sierra, I dunno, I've never bothered to try it in Sierra or H.S. The CUDA driver is built for a specific OS version, 8 for El Capitan, 9-9.1 for Sierra, 9.2-10 for High Sierra, etc. It is different for the actual Apps, i.e., a CUDA 8 App will still work with a CUDA 10 driver. You shouldn't even have a cuda 8 driver installed in H.S, much less ask why it doesn't work. I don't think CUDA has Ever worked for CPUs, last time I checked it was strictly for nVidia GPUs.

This is from the last time you asked that question;
Message 1951115 - Posted: 21 Aug 2018

Well, CUDA 8 is for El Capitan. For Sierra and High Sierra you need to use the Highest CUDA version. Just keep Updating it in CUDA Preferences until it says Nothing Newer.
There is still the problem with the 2GB GPUs. Don't connect them to a Monitor, use a GPU with at least 3GB of RAM, such as my 1050Ti. Or connect the monitor to an ATI GPU.
The 2GB NV GPUs work fine as long as they are not connected to a Monitor. See this, CUDA version 9.2, compute capability 6.1, 4096MB, 3550MB available
That's how much vRAM OSX uses Just to run the monitor. That's a pretty big chunk out of a 2 GB GPU.

You should have updated to a CUDA 9 driver back when Sierra first appeared. That was some time ago. I doubt CUDA 8 has worked since Sierra, I dunno, I've never bothered to try it in Sierra or H.S. The CUDA driver is built for a specific OS version, 8 for El Capitan, 9-9.1 for Sierra, 9.2-10 for High Sierra, etc. It is different for the actual Apps, i.e., a CUDA 8 App will still work with a CUDA 10 driver. You shouldn't even have a cuda 8 driver installed in H.S, much less ask why it doesn't work. I don't think CUDA has Ever worked for CPUs, last time I checked it was strictly for nVidia GPUs.

1.) At the time that Sierra was first available, I was still on the Hackintosh and stuck on El Capitan, 10.11.6 due to the Mac Pro 3,1 Profile that best matched my old PC's hardware. I, therefore, was unable to go beyond 8.0.71 then.

2.) A few months ago, when I made the purchase of the used Mac Pro 5,1, I was provided an Image Install of High Sierra 10.13.0 that I immediately Upgraded to 10.13.3. On 10.13.3, I still made use of the 8.0.71 CUDA Driver to first Test the very old CUDA App that you created that I was also using on the Hackintosh. It was then that you mentioned and highly recommended that I move to your OpenCL App, and I did make that change. BUT, I still kept the CUDA Driver at 8.0.71. Since then, I Upgraded 10.13.3 to 10.13.4, 10.13.5, and now 10.13.6 with Sec Update 2018-001. Perhaps one of these Upgrades removed the CUDA Support for BOINC because I kept the CUDA Driver at 8.0.71. I never checked the Event Log in BOINC specifically for the CUDA Line due to moving to your OpenCL App - until now. It's, therefore, quite possible that because I DID NOT Upgrade the CUDA Driver that MacOS removed or blocked the CUDA Driver Support in BOINC after Upgrading beyond 10.13.3.

2A.) When I made the Post in August, I was making use of one of the GTX-750TI SC Cards from the Hackintosh that I moved over to the 5,1. (Later changed out for the newer 1050 Card.)

3.) My GTX-1050 2GB Card is NOT connected to a monitor. ONLY the ATI HD-5770 is connected to the Apple Cinema Display. (Has been this way since I purchased the 5,1.) Due to the 3 Drives with OSes on them, (first SSD is High Sierra, second SSD is Sierra, third SSD is Boot Camp - Win 7 Pro x64 SP-1), and the need to make use of the "Option Key" during Boot Up to select Drives I had no choice BUT to connect the Apple Cinema Display to the ATI Card. The EVGA GTX-1050 is NOT Flashed, and therefore is UNABLE to give me a Display at Boot Up to show me my choices with use of the "Option Key".

4.) In Windows, I thought that under Lunatics 0.44 that I remember CUDA50 on CPU was used - in addition to GPU support; this is why I ask now about CPU CUDA compatibility on Mac. I guess Apple never allowed this... But, I thought that, (even though not allowed on Apple products), that all processors are CUDA Capable... (Maybe I'm remembering wrong/incorrectly... Been quite a while since I've had 0.44; it was on Old Prometheus from 4 years ago, now... New Prometheus now uses the current 0.45 Version of Lunatics.)

Anyway, now I know that I need to keep the CUDA Driver current and will do so. Until I can get a GPU in the system that has more than 2GB VRAM, and is Pre-Flashed for Mac, I won't be trying out the newer CUDA App.

Thanks again for the clarification. :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

[Hardware Clarification:]

(I'm NOT going to be able to get this in on the other Post - too close to the 1 Hr. Time Limit...)

I forgot!!! Actually, when I picked up the 5,1 and the Hackintosh had NOT yet died, I had on hand TWO EVGA GTX-1050's new in box... I DID open one and installed it into the 5,1... THEN, when the Hackintosh died and I later had New Prometheus built, (in hopes of resurrecting the Hackintosh), and found that I could NOT get a Hack built out of it - I pulled the two 750TI SC Cards out of New Prometheus, pulled the 1050 out of the 5,1, installed the good 750TI SC into the 5,1 and installed both 1050s into New Prometheus!

So, in August's Post, I most likely was still on the first 1050 in the 5,1... BUT after the Hack died this changed. A couple months later, the EVGA GTX-1050s went on sale at my local source for $149 + Tax so I picked up a third 1050 which is currently in the 5,1.

Now, the two 750TI SC Cards from the Hack are NOT being used anywhere, AND I retired Exeter, (the Win XP Pro x64 SP-2 System), and pulled out the EVGA GTX-760. I was hoping to make use of all three of the 7x0 Cards in the New Hack in the next 6 Mos; BUT, with the issues of Mojave NOT having Maxwell/Pascal/Turing Drivers, yet, I may end up getting an ATI Sapphire Pulse RX580 Card for that System. (I won't be crunching with it, if this is the case...)

Thanks again, TBar for the Mac CUDA clarification. :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Someone found the version of the Web Driver that works with the current System at nVidia. I'm still not sure why it hasn't been offered publicly, but, it does appear to be an official nVidia Web Address. I tried it last night since I really didn't have anything to lose seeing as how I had stopped crunching on my real Mac. It seems to be the BEST driver yet for High Sierra on My Mac Pro 3,1. I had been having small glitches with all the previous HS drivers, this one solved all those. I've been running it with a single 750 Ti since last night and it hasn't missed a Pulse yet. It was recently posted on the board at nVidia, I used a link at MacRumors last night, https://forums.geforce.com/default/topic/1069883/geforce-drivers/web-driver-387-10-10-15-15-108-osx-lag/post/5907744/#5907744
The machine that's been running it since last night is here, https://setiathome.berkeley.edu/results.php?hostid=8097309
It's the only GPU in the machine, I've been running FireFox with 6 or less tabs open while crunching and it still hasn't missed any Pulses....very encouraging.

OK - I do have 387.10.10.15.15.108 already Downloaded from InsanelMac when they caught the release of it. However; they, too, indicate that NVIDIA has pulled it from release.

I just want to ask, then, does this new Driver work fully with the new 2018-002 Security Release??? If yes, I will perform the Security Update and install 387.10.10.15.15.108. [EDIT:] I also have the 387.10.10.10.40.108 Driver that was released and pulled from release, too.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Yes, it works with the 2018-002 Security Release. The machine running it has been fully updated. I was running an older card that works with the stock Apple driver. I updated the Security Patch a couple days ago so it would give a message when the Updated drivers appeared. So far, No updated drivers have been posted. The only one that works is the one in the link, 387.10.10.10.40.108. Still nothing Officially from nVidia through the usual channels.

OK - I'm getting ready to perform the Security Update and NVIDIA Driver Update. FIRST, I will finish up my queue of OpenCL work.

I've also Downloaded and Unzipped the V0.97b2 Special CUDA App and will Install this after performing the OS and NVIDIA Updates. I'm looking forward to FASTER Crunching in High Sierra with your new CUDA App! ;-) :-)

Thanks, TBar, for all you do. :-). :-D


TL

[EDIT:]

WOW!!! 4 Min. 32 Sec on your 750TI.

What should I expect on my 1050??? (Close to the same???)


TL

[EDIT 2:]

42 OpenCL Units to go... At almost 11 Min a piece... If I did my math right, 7.7 Hours from now, which is a 2:23 PM - PST Start Time. So close to 10:30 PM - PST


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

28 OpenCL Units Left to Crunch... Then I Upgrade OS & NVIDIA Driver. Then Install Special CUDA App V0.97b2!!! :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

All OpenCL Units are now complete. Unfortunately, I'm REALLY tired AND I have to be at the Polls early for work.

I'm heading to bed for now. Polls close at 8 PM Pacific, so looks like Wednesday I'll be starting the change over to CUDA Special App.

Goodnight. :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

YEAH!!!!! :-) :-D

CUDA 90 now Crunching!!!!! :-) :-D

[EDIT:]

Units now crunching in UNDER 3 Min and 50 Sec! :-D


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

First CUDA 90 Unit Finished on New Andromeda, Computer ID: 7952666

[EDIT:]

Using "Unroll" of 5.


Task 7116934126

Name blc04_2bit_guppi_58405_79976_HIP86036_0010.14527.0.22.45.62.vlar_1

Workunit 3204078617
Created 6 Nov 2018, 7:13:00 UTC
Sent 6 Nov 2018, 11:15:19 UTC
Report deadline 29 Dec 2018, 16:15:01 UTC
Received 6 Nov 2018, 11:26:06 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x00000000)
Computer ID 7952666
Run time 3 min 42 sec
CPU time 44 sec
Validate state Initial
Credit 0.00
Device peak FLOPS 1,862.40 GFLOPS
Application version SETI@home v8
Anonymous platform (NVIDIA GPU)
Peak working set size 495.94 MB
Peak swap size 82,820.61 MB
Peak disk usage 0.13 MB


Stderr output

<core_client_version>7.8.6</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GTX 1050, 2047 MiB, regsPerBlock 65536
computeCap 6.1, multiProcs 5
pciBusID = 3, pciSlotID = 0
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 1050 is okay
SETI@home using CUDA accelerated device GeForce GTX 1050
Using unroll = 5 from command line args

setiathome v8 enhanced x41p_V0.97b2, Cuda 9.10 special
Modifications done by petri33, compiled by TBar

Detected setiathome_enhanced_v8 task. Autocorrelations enabled, size 128k elements.
Work Unit Info:
...............
WU true angle range is : 0.009559
Sigma 62
Sigma > GaussTOffsetStop: 62 > 2
Thread call stack limit is: 1k
Pulse: peak=10.0134, time=45.9, period=28.01, d_freq=7562232813.54, score=1.033, chirp=0.22847, fft_len=2k
Autocorr: peak=18.44298, time=74.45, delay=5.107, d_freq=7562231928.34, chirp=-14.001, fft_len=128k
Pulse: peak=8.116181, time=45.9, period=18.55, d_freq=7562237496.66, score=1.073, chirp=71.453, fft_len=2k
Pulse: peak=4.489587, time=45.9, period=9.574, d_freq=7562237510.01, score=1.037, chirp=71.987, fft_len=2k
Pulse: peak=2.237178, time=45.82, period=3.81, d_freq=7562233098.6, score=1.086, chirp=-83.057, fft_len=256

Best spike: peak=23.01102, time=6.085, d_freq=7562232300.25, chirp=45.478, fft_len=8k
Best autocorr: peak=18.44298, time=74.45, delay=5.107, d_freq=7562231928.34, chirp=-14.001, fft_len=128k
Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.124e+11, d_freq=0,
score=-12, null_hyp=0, chirp=0, fft_len=0
Best pulse: peak=2.237178, time=45.82, period=3.81, d_freq=7562233098.6, score=1.086, chirp=-83.057, fft_len=256
Best triplet: peak=0, time=-2.124e+11, period=0, d_freq=0, chirp=0, fft_len=0

Spike count: 0
Autocorr count: 1
Pulse count: 4
Triplet count: 0
Gaussian count: 0

03:24:31 (16542): called boinc_finish(0)

</stderr_txt>
]]>

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Just completed MacOS Update for High Sierra 2018-002.

MacOS: 10.13.6 - (17G3025)

NVIDIA Web Driver Update: 387.10.10.10.40.108

CUDA Driver: 410.130

[BOINC Event Log Update:]

Wed Nov 7 10:38:53 2018 | | cc_config.xml not found - using defaults
Wed Nov 7 10:38:53 2018 | | Starting BOINC client version 7.8.6 for x86_64-apple-darwin
Wed Nov 7 10:38:53 2018 | | log flags: file_xfer, sched_ops, task
Wed Nov 7 10:38:53 2018 | | Libraries: libcurl/7.50.2 OpenSSL/1.1.0 zlib/1.2.11 c-ares/1.11.0
Wed Nov 7 10:38:53 2018 | | Data directory: /Library/Application Support/BOINC Data
Wed Nov 7 10:38:54 2018 | | CUDA: NVIDIA GPU 0: GeForce GTX 1050 (driver version 10.0.130, CUDA version 10.0, compute capability 6.1, 2048MB, 1343MB available, 1862 GFLOPS peak)
Wed Nov 7 10:38:54 2018 | | OpenCL: NVIDIA GPU 0: GeForce GTX 1050 (driver version 10.33.0 387.10.10.10.40.108, device version OpenCL 1.2, 2048MB, 1343MB available, 1862 GFLOPS peak)
Wed Nov 7 10:38:54 2018 | | OpenCL: AMD/ATI GPU 0: ATI Radeon HD 5770 (driver version 1.2 (Oct 10 2018 23:57:11), device version OpenCL 1.2, 1024MB, 1024MB available, 680 GFLOPS peak)
Wed Nov 7 10:38:54 2018 | | OpenCL CPU: Intel(R) Xeon(R) CPU E5620 @ 2.40GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Wed Nov 7 10:38:54 2018 | SETI@home | Found app_info.xml; using anonymous platform
Wed Nov 7 10:38:54 2018 | | Host name: Andromeda.local
Wed Nov 7 10:38:54 2018 | | Processor: 16 GenuineIntel Intel(R) Xeon(R) CPU E5620 @ 2.40GHz [x86 Family 6 Model 44 Stepping 2]
Wed Nov 7 10:38:54 2018 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni pclmulqdq dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 popcnt aes pcid
Wed Nov 7 10:38:54 2018 | | OS: Mac OS X 10.13.6 (Darwin 17.7.0)
Wed Nov 7 10:38:54 2018 | | Memory: 64.00 GB physical, 658.96 GB virtual
Wed Nov 7 10:38:54 2018 | | Disk: 930.71 GB total, 658.72 GB free
Wed Nov 7 10:38:54 2018 | | Local time is UTC -8 hours
Wed Nov 7 10:38:54 2018 | | VirtualBox version: 5.2.18r124319
Wed Nov 7 10:38:54 2018 | | A new version of BOINC is available (7.14.2). <a href=https://boinc.berkeley.edu/download.php>Download</a>
Wed Nov 7 10:38:54 2018 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7952666; resource share 365
Wed Nov 7 10:38:54 2018 | SETI@home | General prefs: from SETI@home (last modified 04-May-2018 09:14:36)
Wed Nov 7 10:38:54 2018 | SETI@home | Computer location: home
Wed Nov 7 10:38:54 2018 | SETI@home | General prefs: no separate prefs for home; using your defaults
Wed Nov 7 10:38:54 2018 | | Reading preferences override file
Wed Nov 7 10:38:54 2018 | | Preferences:
Wed Nov 7 10:38:54 2018 | | max memory usage when active: 22937.60 MB
Wed Nov 7 10:38:54 2018 | | max memory usage when idle: 32768.00 MB
Wed Nov 7 10:38:54 2018 | | max disk usage: 30.00 GB
Wed Nov 7 10:38:54 2018 | | max CPUs used: 8
Wed Nov 7 10:38:54 2018 | | (to change preferences, visit a project web site or select Preferences in the Manager)
Wed Nov 7 10:38:55 2018 | | Suspending computation - time of day

--------------------------------------------------------------------------------------

I'm now going to take the Time Of Day restriction off, crunch a few more Units to make sure that everything is "Normal".


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Task 7120370694

Name blc04_2bit_guppi_58405_85640_HIP85417_0027.14513.818.22.45.56.vlar_1
Workunit 3205744144
Created 7 Nov 2018, 9:15:32 UTC
Sent 7 Nov 2018, 12:09:24 UTC
Report deadline 30 Dec 2018, 17:09:06 UTC
Received 7 Nov 2018, 16:30:52 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x00000000)
Computer ID 7952666
Run time 3 min 44 sec
CPU time 44 sec
Validate state Valid
Credit 38.86
Device peak FLOPS 1,862.40 GFLOPS
Application version SETI@home v8
Anonymous platform (NVIDIA GPU)
Peak working set size 492.80 MB
Peak swap size 82,829.53 MB
Peak disk usage 0.13 MB


Stderr output

<core_client_version>7.8.6</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GTX 1050, 2047 MiB, regsPerBlock 65536
computeCap 6.1, multiProcs 5
pciBusID = 3, pciSlotID = 0
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 1050 is okay
SETI@home using CUDA accelerated device GeForce GTX 1050
Using unroll = 5 from command line args

setiathome v8 enhanced x41p_V0.97b2, Cuda 9.10 special
Modifications done by petri33, compiled by TBar

Detected setiathome_enhanced_v8 task. Autocorrelations enabled, size 128k elements.
Work Unit Info:
...............
WU true angle range is : 0.005629
Sigma 105
Sigma > GaussTOffsetStop: 105 > -41
Thread call stack limit is: 1k
Pulse: peak=2.038818, time=45.84, period=3.028, d_freq=7617824442.09, score=1.017, chirp=3.0755, fft_len=512
Spike: peak=26.2785, time=35.79, d_freq=7617829303.26, chirp=-8.0828, fft_len=32k
Spike: peak=28.27209, time=35.79, d_freq=7617829303.23, chirp=-8.1031, fft_len=32k
Spike: peak=29.08867, time=35.79, d_freq=7617829303.2, chirp=-8.1234, fft_len=32k
Spike: peak=28.62465, time=35.79, d_freq=7617829303.17, chirp=-8.1437, fft_len=32k
Spike: peak=26.93102, time=35.79, d_freq=7617829303.14, chirp=-8.164, fft_len=32k
Spike: peak=24.20635, time=35.79, d_freq=7617829303.11, chirp=-8.1843, fft_len=32k
Spike: peak=24.45892, time=17.18, d_freq=7617830080.16, chirp=-11.441, fft_len=128k
Triplet: peak=10.23132, time=38.3, period=34.81, d_freq=7617829214.58, chirp=20.763, fft_len=2k
Pulse: peak=9.728954, time=45.9, period=26.13, d_freq=7617830042.92, score=1.006, chirp=39.142, fft_len=2k
Pulse: peak=2.644168, time=45.9, period=5.548, d_freq=7617830658, score=1.025, chirp=-48.985, fft_len=2k
Spike: peak=24.00273, time=81.6, d_freq=7617829856.82, chirp=-60.722, fft_len=32k
Spike: peak=25.01837, time=81.6, d_freq=7617829856.91, chirp=-60.743, fft_len=32k
Spike: peak=24.63848, time=81.6, d_freq=7617829856.83, chirp=-60.804, fft_len=32k
Spike: peak=25.29839, time=81.6, d_freq=7617829856.92, chirp=-60.824, fft_len=32k
Spike: peak=24.18237, time=81.6, d_freq=7617829856.84, chirp=-60.885, fft_len=32k
Spike: peak=24.6733, time=81.6, d_freq=7617829856.93, chirp=-60.905, fft_len=32k
Pulse: peak=0.1896814, time=45.82, period=0.06772, d_freq=7617825682.69, score=1.003, chirp=-66.442, fft_len=64
Pulse: peak=2.569423, time=45.99, period=5.727, d_freq=7617827991.07, score=1.018, chirp=-77.899, fft_len=4k
Pulse: peak=7.807442, time=45.9, period=20.34, d_freq=7617828761.92, score=1.028, chirp=86.589, fft_len=2k
Pulse: peak=5.360508, time=45.99, period=14.76, d_freq=7617828080.18, score=1.006, chirp=91.28, fft_len=4k

Best spike: peak=29.08867, time=35.79, d_freq=7617829303.2, chirp=-8.1234, fft_len=32k
Best autocorr: peak=17.41364, time=5.727, delay=5.554, d_freq=7617828328.65, chirp=-7.0699, fft_len=128k
Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.124e+11, d_freq=0,
score=-12, null_hyp=0, chirp=0, fft_len=0
Best pulse: peak=7.807442, time=45.9, period=20.34, d_freq=7617828761.92, score=1.028, chirp=86.589, fft_len=2k
Best triplet: peak=10.23132, time=38.3, period=34.81, d_freq=7617829214.58, chirp=20.763, fft_len=2k

Spike count: 13
Autocorr count: 0
Pulse count: 7
Triplet count: 1
Gaussian count: 0

08:19:33 (31208): called boinc_finish(0)

</stderr_txt>
]]>

----------------------------------------------------------------------

Everything seems to be fine. Unit Crunched in 3 Min and 44 Sec. :-) :-D


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Just completed MacOS Update for High Sierra 2018-002.
MacOS: 10.13.6 - (17G3025)
NVIDIA Web Driver Update: 387.10.10.10.40.108...

The good news is I've been testing the correct Web Driver for the past few days. The Bad news is although better, the driver still conflicts with the CUDA Special App when using the GPU to surf the Web. Best use would be to use a different GPU to run the Monitor, or, do as I do with the Hackintoshes. On the Hacks I start the App, make sure they aren't missing the Best Pulse, then turn off the monitor and let them run unmolested. They do fine sitting over in the corner crunching SETI with the monitors Off. Maybe Mojave will be better...someday.

Oh, don't forget to set the checkpoint in BOINC Manager/Disks & Memory to higher than your run-time. If it takes 3.75 minutes to run the task, set the checkpoint to 4 minutes so you aren't resuming suspended tasks.

Just completed MacOS Update for High Sierra 2018-002.
MacOS: 10.13.6 - (17G3025)
NVIDIA Web Driver Update: 387.10.10.10.40.108...

The good news is I've been testing the correct Web Driver for the past few days. The Bad news is although better, the driver still conflicts with the CUDA Special App when using the GPU to surf the Web. Best use would be to use a different GPU to run the Monitor, or, do as I do with the Hackintoshes. On the Hacks I start the App, make sure they aren't missing the Best Pulse, then turn off the monitor and let them run unmolested. They do fine sitting over in the corner crunching SETI with the monitors Off. Maybe Mojave will be better...someday.

Oh, don't forget to set the checkpoint in BOINC Manager/Disks & Memory to higher than your run-time. If it takes 3.75 minutes to run the task, set the checkpoint to 4 minutes so you aren't resuming suspended tasks.

Monitor is attached to the Native ATI HD-5770 GPU - this GPU does NOT crunch.

I can't find "checkpoint" in "Disks & Memory", BUT, the Checkbox about leaving Units Suspended In Memory IS NOT checked. [EDIT:] (I'm on BOINC 7.8.6)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

I've been monitoring my BOINC/SETI@Home Crunching aspects over the past couple days of implementation of CUDA 90.

1.) My Local BOINC Settings of Crunching between 6 PM and 9 AM Pacific work.

That is, in resuming crunching from Suspension Hours - there seems to be NO adverse effects. BOINC picks up right where it left off, no crashing of Units, no hanging, no errors. (So far.)

2.) Today, I tested even further by purposely Suspending BOINC manually when a newly Crunching Unit reached 49 Seconds into the start of said Unit. The Default "Checkpoint Apparatus" in BOINC, (I believe), is 60 Seconds; so upon resuming work, as anticipated, the Unit reset and started back at 0 Seconds Crunched. Again, no hanging, no crashing, no errors in resuming work from being Suspended.

NOTE: For the BRIEF week or two that I tried Petri's "Special Sauce"-App, I remember reporting the opposite, and this was investigated and a warning was posted to others that Petri's App DOES NOT respond well to Suspending and Resuming BOINC and that in so doing, Units crashed and error-ed.

I then went to TBar's OpenCL App having Upgraded from TBar's original CUDA 75 App. (This was on the Mac Pro 5,1 System.)

Please don't take this reporting as a complaint against Petri's App. I think that App has worked well for the masses, especially those whom are able to run their Systems 24/7. I just am NOT in a position, (electricity Usage-wise), to allow my Systems to run 24/7.

I MUST engage Crunching Windows. Due to HIGHER PG&E Bills when Crunching during the day, I've resorted to ONLY Crunching from 6 PM to 9 AM year round.

For now, I find that TBar's CUDA 90 App to work flawlessly; just as his OpenCL and CUDA 75 Apps have done for me in the past.


Thanks, again, TBar for all you do and have done. Also, thank you to Petri and all the other App Developers for SETI@Home. :-) I know LOTS about Hardware, (NOT everything, but lots), BUT I AM NOT a Programmer/Coder!

I could NEVER do what you all do/have done. I stand in awe and amazement. Especially since all of you are doing this on a completely Volunteer basis.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

The app doesn't always fail when restarting, just sometimes. When it does fail it generally doesn't start at all, and just hangs (with 0 GPU use) until the 20x estimated time has elapsed, until BOINC stops it with a Time Exceeded error. If such a task is acting that way (say elapsed time is 3x normal) you can suspend it, clear the slot for that task, resume it, and it will complete normally.

Task don't appear to ever have problems until a checkpoint is dropped. And then only sometimes. It 'could' be only if a triplet, significant pulse, etc was recorded - who knows why, but only happens sometimes. As you said, if you suspend within the first 60s before a checkpoint they are fine ...

Hence the extended checkpointing warning to users of the app.

One other question... Why is the Dual Xeon CPU System also NOT being shown as CUDA Compatible??? I thought all CPUs could use CUDA.

You're confusing it with OpenCL.

CUDA is a parallel computing platform and application programming interface model created by Nvidia, and is proprietary, works only on Nvidia hardware.
OpenCL is a parallel computing platform and application programming interface framework created by the Khronos Group, is Open Source and works on any computing hardware made and maintained by Altera, AMD, Apple (although now deprecated in favor of Metal 2), ARM, Creative, IBM, Imagination, Intel, Nvidia, Qualcomm, Samsung, Vivante, Xilinx, and ZiiLABS.
OpenCL will together with OpenGL eventually be merging into of Vulkan.

---

<----- RAC Up to almost 8.9K!!! This is a jump up of approx 4.2K in about 6.5 days!!!!! :-) :-D

[EDIT:]

Time, currently, is 4:40 AM - PST

Crunching will stop for awhile, (at 9 AM - PST), and resume at 6 PM - PST.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos