Posts by David Riese

1) Message boards : Cafe : Hack? Gaming the system? (Message 1990)
Posted 14 Oct 2019 by David Riese
Post:
Hey, Mike, thanks for pushing on this issue!
2) Message boards : Cafe : The Last Person To Post Here Wins 10 (Message 1899)
Posted 15 Aug 2019 by David Riese
Post:
Yep, activity on this thread has slowed dramatically.

It has been a very busy summer. Since my last post to the LPTPHW threads, I have traveled to DC (multiple times - consulting with DoD and Pharma; speechifying at trade group meetings), Indiana (visiting the in-laws and attending the Indy 500), Hawaii (visiting my Mom's family on the Big Island), and Wisconsin (family reunion - Dad's side).

PharmD students have already returned to class here at Auburn. Undergraduates and MS/PhD students start next Monday. I have a heavy teaching load this fall, including portions of 3 new courses.

Continue to be excited about our melanoma genetics/drug target identification research. We are pushing hard to get three papers in print by the end of 2019. Going to try to get 3 major grant applications out the door by the end of 2019, too.

Been having fun upgrading the GPU in my old Macs to GTX 980s, 1060s, and 1070s. Don't have any Macs that reliably run OS Mojave, so I *gasp* bought a new Mac - 2x dual core i5 iMac - so I could run a very nice open source statistical analysis/programming environment software package ("R"). Being the dedicated cruncher, I connected it to a pair of eGPUs, each containing an AMD RX 580. Awesome Collatz crunching performance (for a Mac, that is).

It certainly has been nice to have Collatz running so well for the past few months. My only glitches have been caused by network/power outages at home/work.

Hope all of you are doing well. Post when you get a chance - would love to hear how all of you are doing.

Dave
3) Message boards : Number crunching : Computation errors (Message 1861)
Posted 26 Jun 2019 by David Riese
Post:
David,
I had the same problem with WUs taking twice as long (but only on one machine). The fix for that problem, in my case, was to add the parameters back into the collatz_sieve_1.30_windows_x86_64__opencl_nvidia_gpu file, which seemed to have been cleared. I hope that's all your problem is too.

Steve


Steve:

Thanks for the suggestion! Yep, the custom config file had been replaced by a blank file. Restored the custom file and the throughput returned. Hoorah!

Regards,

Dave
4) Message boards : Number crunching : Computation errors (Message 1859)
Posted 25 Jun 2019 by David Riese
Post:
Echoing Barry's comments, this appears to be a tough nut to crack and we should have patience while Jon tries to sort it out. By the way, in my experience, the tasks that terminate with an error in 2 seconds don't hurt my machines' throughput very much - typically after no more than 5-6 such tasks, the machine will start working on a task that can be crunched to completion. So, relatively speaking, this aspect of the problem doesn't waste a lot of machine time and electricity.

However, in my experience, throughput is heavily impacted by the fact that tasks that can be crunched to completion now take twice as long to complete. For example, my computer 850946 (Mac Pro 5,1; GTX 1060) used to take 12-13 minutes to crunch a GPU task but now takes 24-25 minutes to crunch a GPU tasks. I also note that CPU usage has increased from 12-13 seconds per task to 160-170 seconds per task. Thus, this machine used to average 27-31K credits/completed task but now averages 35-39K credits/completed task. So, even though it takes twice as long to crunch a task to completion, the throughput of this machine has been cut by only ~40%. Fortunately, only two of my boxes have been affected by this problem and I hope that you too have had only a fraction of your boxes affected. Hang in there! I am sure that Jon is working as hard as he can to fix things.
5) Message boards : Number crunching : Computation errors (Message 1836)
Posted 24 Jun 2019 by David Riese
Post:
I am having a similar problem on at least one of my machines; when my GTX 1060 is capable of crunching a work unit, it takes approximately twice as long as before.
6) Message boards : Number crunching : Computation errors (Message 1835)
Posted 24 Jun 2019 by David Riese
Post:
Jon has returned from Wisconsin and fixed the file uploading problem. Alas, it does not appear that the computation error problem has been fixed. Any idea if the problem is restricted to a specific brand and model of board? I am having problems with my only NVIDIA GTX 1060 and one of my three NVIDIA GTX 980s - all in Macs.
7) Message boards : Number crunching : Three new problems this week. (Message 1834)
Posted 24 Jun 2019 by David Riese
Post:
Welcome back, Jon, and thanks for taking care of the file_upload_handler daemon and providing us with an update. I am no longer having any trouble uploading work.
8) Message boards : Number crunching : Three new problems this week. (Message 1831)
Posted 24 Jun 2019 by David Riese
Post:
Don't bother trying to solve the problem by downloading and installing an updated version of BOINC - boinc.berkeley.edu is down, probably until Monday at the earliest:

setiathome.berkeley.edu/forum_thread.php?id=84081&postid=1999220
9) Message boards : Number crunching : Computation errors (Message 1830)
Posted 24 Jun 2019 by David Riese
Post:
Doesn't appear to be specific for windows, as my OS X 10.11 and 10.13 Macs have generated an unusually high number of computation errors.
10) Message boards : Cafe : The Last Person To Post Here Wins 10 (Message 1700)
Posted 25 Apr 2019 by David Riese
Post:
Good morning. 😀

Time for 🥓 and ☕️.
TL


Good morning!!

Did your computer ever get any tasks here?


I gave up on Daedalus getting tasks. She's off due to the HIGH electricity costs needed to keep her running. I'm only on Hackintosh-Andromeda, and Farragut, (dad's i7 7700 using the Integrated HD-630 for SETI), for Crunching. Farragut seldom Crunches... Maybe 2 Hours every 3 Days on average. Andromeda, (as previously stated), only Crunches from 18:00 to 09:00 Daily and is currently Crunching SETI.
TL


That's too bad and I still have no clue why it won't get any tasks!!

David I think that's a great idea but it will have some bumps in the road as there really aren't many people using both brands in a single pc, obviously some people do and it works but lots of people have given up trying.


Mikey, at this moment I have three Macs that have an AMD GPU that is crunching Einstein tasks and an NVIDIA GPU that is crunching Collatz tasks: two have a GTX 960 and an HD 5770 and one has a GTX 750Ti and an R9 380. Next week I am going to take delivery of an AMD RX 580 that I intend to pair with a GTX 750Ti to replace a GTX 680 in a Mac 3,1. So, by next week I will have the following eleven NVIDIA GPUs crunching for Collatz: GTX 980, GTX 960 (2), GTX 770, GTX 680, and GTX 750Ti (6). And I will have the following four AMD GPUs crunching for Einstein: RX 580, R9 380, and HD 5770 (2). And I will have the following spares looking for new homes: R9 380, HD 5870 (2), HD 5770 (2) and GeForce 8800 GT (3).

You are correct in that it took a bit of setup work to install the cc_config.xml with the correct switches to crunch only Collatz tasks on the NVIDIA GPUs and only Einstein tasks on the AMD GPUs. And, it has been a bit of ongoing work to make sure that the machines don't get so overburdened with CPU tasks that they stop crunching GPU tasks. I suspect my problem is that my network connectivity at work is so spotty I tend to keep 10 days of tasks stored on each computer at all times. Thus, it takes only a brief power outage to throw everything out of whack on my work computers - once the power is restored, the computers typically can't finish all of the tasks on hand by the task deadlines. So, my standard operating procedure following a power outage is to manually abort all of the tasks that I don't think the computer can crunch by the dead line. The problem is that my BOINC crunchers are typically the least of my problems in recovering from a power outage at work - I takes time to move my 9 freezers and 6 incubators from backup power outlets to standard outlets.

I think part of the success is that all of these macs are 10-11 years old, meaning that there is a lot of information available on how to upgrade them and make them "work". Another part is that I have avoided the temptation to upgrade to the latest and greatest versions of the Mac operating system (10.13 and 10.14). All but one of my crunchers are running 10.11, with the exception running 10.12.

BTW, I've got to admit that I have been very happy with my decision last summer to surrender my administrative responsibilities (Associate Dean for Research and Graduate Programs for the School of Pharmacy) and return to a typical faculty role of teaching, research, and outreach. Indeed, this change in workload has allowed me to pursue a very exciting project that has the potential to identify new biomarkers for particularly aggressive forms of metastatic melanoma and the potential to identify novel strategies for treating these aggressive tumors. It's been very gratifying to work closely with my students to attack problems of potential societal benefit. I really can't tell you how much fun it is to stare at a screen full of data for 2-3 hours and have the "aha" moment that leads to a novel explanation or implication of the observations that we have been making. I have had about 5 of those moments in the last week, and it had been **years** since I had my last such moment. What a rush!

Well, gotta run. One of my chemical engineering graduate students is finishing his PhD in absentia while working full-time for a clinical testing laboratory on the other side of Alabama. This means that I mentor him primarily by videoconference meetings at 7 am on Thursday mornings. Because of my schedule tomorrow, I need to conduct that meeting from my work office, meaning that I really need to leave home by 6:00 am. Fun times!
11) Message boards : Cafe : The Last Person To Post Here Wins 10 (Message 1697)
Posted 24 Apr 2019 by David Riese
Post:
Good morning. 😀

Time for 🥓 and ☕️.
TL


Good morning!!

Did your computer ever get any tasks here?

I gave up on Daedalus getting tasks. She's off due to the HIGH electricity costs needed to keep her running. I'm only on Hackintosh-Andromeda, and Farragut, (dad's i7 7700 using the Integrated HD-630 for SETI), for Crunching. Farragut seldom Crunches... Maybe 2 Hours every 3 Days on average. Andromeda, (as previously stated), only Crunches from 18:00 to 09:00 Daily and is currently Crunching SETI.


TL


TL, I am sorry to hear that electricity costs are preventing you from doing more crunching. Nonetheless, thanks for sharing your crunching experiences with us - they have been particularly helpful for me as I try to squeeze every last bit of utility out of my collection of old Macs! In the Macintosh forum I discussed how I have replicated your problems with Macs that have two different NVIDIA GPUs. To make a long story short, I am out of ideas. Think I am going to shift gears and try to configure boxes that have an NVIDIA GPU and an AMD GPU. This will allow me to crunch C@H tasks using the NVIDIA GPU and Einstein@Home tasks using the AMD GPU - this will increase my engagement with other BOINC projects.

Until later,

Dave
12) Questions and Answers : Macintosh : Optimising the Mac apps (Message 1696)
Posted 24 Apr 2019 by David Riese
Post:
Again, TL, thanks for sharing your experiences with us.

On an impulse I bought a relatively cheap GTX 980. Have a GTX 750Ti and an AMD Radeon R9 380 in a MacPro 4,1 here at home (main home computer) running El Capitan (10.11.6). So, decided to see what would happen if I replaced the AMD R9 380 with the GTX 980. Machine booted up just fine. System information reported that the 980 and the 750Ti were indeed installed. But, as you would predict, only the 980 is crunching Collate GPU tasks. See following entries from my BOINC log file:

Tue Apr 23 20:26:38 2019 | | Starting BOINC client version 7.10.3 for x86_64-apple-darwin
Tue Apr 23 20:26:38 2019 | | log flags: file_xfer, sched_ops, task
Tue Apr 23 20:26:38 2019 | | Libraries: libcurl/7.58.0 OpenSSL/1.1.0g zlib/1.2.5 c-ares/1.13.0
Tue Apr 23 20:26:38 2019 | | Data directory: /Library/Application Support/BOINC Data
Tue Apr 23 20:26:38 2019 | | OpenCL: NVIDIA GPU 0 (ignored by config): GeForce GTX 750 Ti (driver version 10.11.14 346.03.15f16, device version OpenCL 1.2, 2048MB, 2048MB available, 602 GFLOPS peak)
Tue Apr 23 20:26:38 2019 | | OpenCL: NVIDIA GPU 1: GeForce GTX 980 (driver version 10.11.14 346.03.15f16, device version OpenCL 1.2, 4096MB, 4096MB available, 1866 GFLOPS peak)
Tue Apr 23 20:26:38 2019 | | OpenCL CPU: Intel(R) Xeon(R) CPU X5570 @ 2.93GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)

Bummer. Going to take out the 750Ti. Don't feel like trying to sort this out right now and I think I have a home for the 750Ti/R9 380 combo. May have to do what you suggested and buy a bare bones 5,1 and build something fun.

BTW, the 980 is crunching C@H GPU tasks in ~925 seconds, which is much faster than my 960s, which crunch them in ~1600 seconds. So, at least there is some good news.
13) Questions and Answers : Macintosh : Optimising the Mac apps (Message 1692)
Posted 21 Apr 2019 by David Riese
Post:
Huh. Thanks for clarifying and I am sorry you are continuing to have this problem. I have never been able to replicate your particular problem, as I have never been able to boot a combination of a GTX 960 (single power connector) and a GTX 750Ti (no power connector) in a Mac3,1 - I suspect it is a power draw issue through the bus. I have been able to boot combinations of GTX 680/GTX 750Ti and GTX 770/GTX 750 Ti, but that doesn't replicate your problem - the GTX 680/770 cards don't rely on the NVIDIA Web Driver. If I can buy a cheap GTX 960 (single power connector), I may try to configure a Mac 4,1 with a GTX 960/GTX 750Ti combination and see if I can sort this out. There are times I wish that I had a dedicated cruncher box that would allow me to tackle issues like this .... maybe I need to buy a used Mac 4,1 as a birthday present to myself!

SKIP the 4,1 UNLESS you Flash the ROM to make it "think" it's a 5,1... IF it were me, I'd go for a REAL 5,1; MORE bang for the buck, IF Dual Xeon 3.46GHz 12c/24t you can run 1333MHz ECC DDR3 System RAM! (iBuildMacs has this RAM PRETTY inexpensively compared to standard Computer Shops, Amazon, or e-Bay. I actually got my 5,1 1066MHz RAM CHEAP!!! (8x8GB Sticks ECC DDR3: $280, NO Tax, Plus Shipping.) I had to go 1066MHz RAM because my Dual 2.4GHz WILL NOT take 1333MHz RAM. I'd have to spend $700 on the Dual 3.46GHz Westmere Upgrade Package to utilize 1333MHz RAM. $280 for 64GB of 1066MHz ECC DDR3 RAM is CHEAP! My Local Source wanted $600 for the SAME style RAM. (This was almost 2 years ago, now.)

Just checked iBuildMacs and the iBuildMacs' price for the 1333MHz ECC DDR3 RAM at 64GB, (8x8GB Sticks for Dual Xeon System), IS: $280, NO Tax, Plus Shipping. (I think shipping was like $14 from L.A. to Danville, CA.)

DON'T buy the WHOLE System from iBuildMacs, as a Complete Custom 5,1 from them is MORE expensive than going e-Bay or Private Party. For example, I got my Dual 2.4GHz System Private Party with 12GB RAM, AMD/ATI HD5770, four Seagate 500GB HDs - (2 with Operating Systems and Software), Apple White/Clear Full Size USB Keyboard with 10-Key, a second White/Clear Full Size Bluetooth Keyboard with 10-Key, Original WiFi and Bluetooth Card, one SuperDrive DVD-Burner, and an Apple 23" Cinema Display - ALL for $579 almost 2yrs ago.

To which I changed ALL the HDs, replacing 3 with Samsung 860 EVO vNAND SATA III 1TB SSDs, and one Western Digital Black 2TB HD, Upgraded the Wifi Card to Airport Extreme Equivalent WiFi-AC and Bluetooth 4 - (Kueren WiFi-AC & Bluetooth 4 Card off e-Bay for $111 + $25 Shipping from Denmark), pulled the AMD/ATI Card, installed one Mac GTX-970 4GB GDDR5 VRAM Card, and a Secondary EVGA GTX-1050 2GB GDDR5 VRAM Card, pulled the 12GB DDR3 RAM out and Upgraded to the aforementioned iBuildMacs' RAM of 64GB: Total Invested is approx - $2,300 - (Estimated System Value: $3,200) (Software package on Sierra 10.12.1 Drive, (Bay 2), includes MS Office for Mac, and the FULL ADOBE CS-6 Suite.) (High Sierra on Primary Drive in Bay 1: 10.13.6) (BootCamp: Windows 7 SP-1 x64 on SSD in Drive Bay 3) (Western Digital Drive partitioned into TWO 1TB Partitions - Part 1: "Time Machine", Part 2: "Games" for Blizzard Games.) I pulled both Apple Keyboards and Upgraded to the MacAlly ACEBTKey Bluetooth Keyboard - Full Size with 10-Key, BUT is "Chicklet"-thick. ($59 + Tax, Amazon. (Included in the $2,300 Investment.))

I HIGHLY recommend a REAL 5,1 for longevity and Maximum Upgrade Capabilities.

[EDIT:] AND, the GTX-970 is Dual 6-Pin! So, I'm pretty much FULL BORE on MOBO Power Draw for that GPU, PLUS the 65-75 Watts' Draw on the GTX-1050.

I've read somewhere, (I think it was MacVidCards'), that the REAL 5,1 CAN allow up to a 300 Watt Draw from "add-on Cards" in the PCI-e Slots combined with the 2 6-Pin Connectors.

I also found that the Total PSU output is AT LEAST 800 Watts on the REAL 5,1.

[EDIT 2:]

Welcome to what is Daedalus! 😀


TL


Thanks for the recommendation to purchase a 5,1 rather than a 4,1 - I think that is great advice, particularly since the price of 5,1 boxes continues to drop. And, thanks for your description of your 5,1 box - sounds like a very impressive piece of hardware and a real bargain, too.
14) Message boards : Cafe : The Last Person To Post Here Wins 9. (Message 1657)
Posted 10 Apr 2019 by David Riese
Post:
New night, STILL CANNOT get Tasks on Daedalus!!! 😳😱😨😯😧😮😲😤😠😡🤬🤯
TL


Do you have Seti tasks? If so maybe your cache is full so Collatz isn't sending you any? What's the percentage number for each project? Maybe change Seti to a zero and Collatz to say 50% it might work.


Remember, Daedalus has been OFF due to the Electric Bill. So, NO, no other Projects' Tasks in queue.
TL


What does the log say when it tries to get tasks?

"Sending scheduler request: To fetch work.
Requesting new tasks for NVIDIA GPU
Scheduler request completed: got 0 new tasks"

NO Errors are being mentioned, AND no reasons why "0" Tasks are being issued.


TL


Sorry that you are having problems getting C@H tasks. Will have to think about your problem. Will post again if I get some inspiration. In the meantime, I am winning. Just like my Auburn Tiger's men's basketball team, until they ran into those Virginia Cavaliers. :-(
15) Questions and Answers : Macintosh : Optimising the Mac apps (Message 1656)
Posted 10 Apr 2019 by David Riese
Post:


I find, (from vigorous testing at SETI@Home), that in MacOS Sierra and High Sierra that the NVIDIA Web Drivers are FLAWED! It seems that they REFUSE to Enable OpenCL Support on secondary and tertiary GPUs that are installed into the System. Without OpenCL Support on my Secondary EVGA GTX-1050 GPU, ONLY the MacVidCards' GTX-1070 GPU is running here at Collatz.

Further, in Hackintosh Systems, it's been reported that the NVIDIA Web Driver confuses and reverses the PCI-e Slot Order from the Physical Slot Locations on the MOBO. So, in effect, PCI-e Slot 1 is seen in MacOS as Slot 2; and Slot 2 on the MOBO is seen as Slot 1. For me, on this new System, this had the effect of killing OpenCL AND OpenGL Support on the GTX-1070 Card.

To remedy this, and get the 1070 Card running with OpenCL and OpenGL I had to physically transpose the two Cards. (AND, in the BIOS, Set PCI-e Slot 2 to be the Video Boot Slot.) This rendered the EVGA GTX-1050 to be without OpenCL and OpenGL Support from the Web Driver.

So, in close, ONLY my GTX-1070 is Crunching here at Collatz... Even though the "cc_config" state to use ALL GPUs... (See my detailed Posts discussing the issue with Mikey and David in The Last Person To Post Here Wins Thread in the Cafe.) This Faulty Web Driver issue was FIRST Reported by Petri33 and TBar in the SETI@Home Number Crunching Forum.

Still, I hope to have Andromeda run a LOOOOONG time here at Collatz. 😀


TL


Good luck with Andromeda! BTW, were you ever able to crunch C@H work on both GPUs? Note that I have two rigs with dual GTX-750Tis and I was able to tweak them to crunch C@H work on both GPUs. My problem is that projects were utilizing the CPU cores and I was unable to commit two CPU cores to support the C@H work on both GPUs. My recollection is that I had to do several things to solve the problem:

1. Adjust preferences to set C@H priority much higher than any other project's priority.
2. Adjust settings to prevent C@H from crunching CPU work units
3. Adjust preferences to load lots of work (minimum 5 days, up to 10 days if available)
4. Delete all projects from BOINC Manager
5. Add C@H to BOINC Manager; crunch for 24 hours to all the system to "equilibrate"
6. Then add other projects to BOINC Manager as desired

Dave

In Sierra and High Sierra, the NVIDIA Web Driver, (by design), STRIPS OUT OpenCL Support from Secondary and Tertiary GPUs in the System. So, NO, I CANNOT get Dual GPUs to Crunch in Collatz on MacOS on NVIDIA Cards.

I DON'T honestly remember if El Capitan had this same issue with the NVIDIA Web Driver. I DO KNOW that one other Web Driver issue that WAS in El Capitan that has carried upward to High Sierra is the fact that the Web Driver REVERSES the Slot Order of the GPUs in the System. That is, the GPU in Physical PCI-e Slot 1 is seen by the Driver as Slot 2, (or Secondary GPU), and Physical PCI-e Slot 2 is seen as Slot 1. (Or Primary GPU.) IF you actually have a case big enough to accommodate the third PCI-e Slot on the Motherboard using Double Wide GPUs like EVGA or MacVidCards, THEN, it is the THIRD PCI-e Slot that is made Primary by the Web Driver. The other two slots fight over 2nd and 3rd Positions.

In El Capitan, all of this is kind of moot IF all the cards are the same make and model... As you state, both of your cards are 750TI Cards. In my case, I have "mixed" Cards. My Primary Card in PCI-e Slot 2 is the MacVidCards' GTX-1070 8GB GPU, and the Secondary Card is the EVGA GTX-1050 2GB GPU in PCI-e Slot 1.

NOW, in using the CUDA Driver at SETI@Home - this is a WHOLE OTHER ball of wax!!! CUDA is NOT affected in the way that the Web Driver is. So, BOTH my GPUs work at SETI under TBar's CUDA90 App under NVIDIA CUDA Driver 410.130 in High Sierra. (CUDA Driver 8.0.71 for El Capitan.)


TL


Huh. Thanks for clarifying and I am sorry you are continuing to have this problem. I have never been able to replicate your particular problem, as I have never been able to boot a combination of a GTX 960 (single power connector) and a GTX 750Ti (no power connector) in a Mac3,1 - I suspect it is a power draw issue through the bus. I have been able to boot combinations of GTX 680/GTX 750Ti and GTX 770/GTX 750 Ti, but that doesn't replicate your problem - the GTX 680/770 cards don't rely on the NVIDIA Web Driver. If I can buy a cheap GTX 960 (single power connector), I may try to configure a Mac 4,1 with a GTX 960/GTX 750Ti combination and see if I can sort this out. There are times I wish that I had a dedicated cruncher box that would allow me to tackle issues like this .... maybe I need to buy a used Mac 4,1 as a birthday present to myself!
16) Questions and Answers : Macintosh : Optimising the Mac apps (Message 1549)
Posted 2 Mar 2019 by David Riese
Post:
Hello everyone. 😀 I'm Posting this from my new Hackintosh named Andromeda.

First, Unoptimized, WU ran for 17 Min and 49 Seconds. Below is the Optimized results on the Second WU run on the new System.


Task 19785828

Name collatz_sieve_2c5fc237-866f-4831-b358-fabf33ff0ad3_0
Workunit 18012781
Created 8 Jan 2019, 23:20:14 UTC
Sent 8 Jan 2019, 23:31:21 UTC
Report deadline 22 Jan 2019, 23:31:21 UTC
Received 8 Jan 2019, 23:43:06 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x00000000)
Computer ID 845444
Run time 11 min 6 sec
CPU time 1 min 13 sec
Validate state Valid
Credit 32,527.57
Device peak FLOPS 6,468.31 GFLOPS
Application version Collatz Sieve v1.40 (opencl_nvidia)
x86_64-apple-darwin
Peak working set size 29.14 MB
Peak swap size 4,381.71 MB
Peak disk usage 6.74 MB
Stderr output

<core_client_version>7.8.6</core_client_version>
<![CDATA[
<stderr_txt>
Collatz Conjecture Sieve 1.40 OS X x86_64 for OpenCL
Written by Slicker (Jon Sonntag) of team SETI.USA
Based on the AMD Brook+ kernels by Gipsel of team Planet 3DNow!
Sieve code and OpenCL optimization provided by Sosiris of team BOINC@Taiwan
kernels_per_reduction=48
threads=10
lut_size=16
sieve_size=27
cache_sieve=1
sleep=0
Collatz Config Settings:
verbose 1 (yes)
kernels/reduction 48
threads 2^10 (1024)
lut_size 16 (524288 bytes)
sieve_size 2^27 (7049172 bytes)
sleep 1
cache_sieve 1 (yes)
reducecpu 0 (no)
Processor Type NVIDIA
Max Dimensions 3
Max Work Items 1024 1024 64
Max Work Groups 1024
Max Kernel Threads 1024
Device Vendor NVIDIA
Name GeForce GTX 1070
Driver Version 10.33.0 387.10.10.10.40.113
OpenCL Version OpenCL 1.2
Device Vendor NVIDIA
Name GeForce GTX 1070
Driver Version 10.33.0 387.10.10.10.40.113
OpenCL Version OpenCL 1.2
actual threads 1024
5138089700750857954623 - 1339 steps @ 0.48066
5138089700755767532315 - 1357 steps @ 0.481332
5138089700752750990015 - 1370 steps @ 0.481786
5138089700753829899111 - 1507 steps @ 0.482237
5138089700761398598431 - 1675 steps @ 0.573218
5138089700855489692315 - 1750 steps @ 1.76171
5138089701311847959835 - 1874 steps @ 7.26315
5138089710667954675711 - 1887 steps @ 123.181
5138089725371704350719 - 2192 steps @ 309.298
Start 5138089700750253883392
Stop 5138089753526812016640
Best 5138089725371704350719
Highest steps 2192
Total steps 420336201408229
CPU time 73.8044 seconds
Elapsed time 00:11:04
15:42:31 (16495): called boinc_finish

</stderr_txt>
]]>

----------------------------- End Std_Err Report. -----------------------------

["cc_config.xml"]

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
<skip_cpu_benchmarks>1</skip_cpu_benchmarks>
</options>
</cc_config>


["app_config.xml"]

<app_config>
<app>
<name>collatz_sieve</name>
<max_concurrent>8</max_concurrent>
<gpu_versions>
<gpu_usage>1.0</gpu_usage>
<cpu_usage>0.5</cpu_usage>
</gpu_versions>
</app>
</app_config>


[".config" File]

[Based on my Windows GTX-1050 Optimizations.]

verbose=1
kernels_per_reduction=48
threads=10
lut_size=16
sieve_size=27
cache_sieve=1
sleep=0

------------------------- End of Config Files. ------------------------

I find, (from vigorous testing at SETI@Home), that in MacOS Sierra and High Sierra that the NVIDIA Web Drivers are FLAWED! It seems that they REFUSE to Enable OpenCL Support on secondary and tertiary GPUs that are installed into the System. Without OpenCL Support on my Secondary EVGA GTX-1050 GPU, ONLY the MacVidCards' GTX-1070 GPU is running here at Collatz.

Further, in Hackintosh Systems, it's been reported that the NVIDIA Web Driver confuses and reverses the PCI-e Slot Order from the Physical Slot Locations on the MOBO. So, in effect, PCI-e Slot 1 is seen in MacOS as Slot 2; and Slot 2 on the MOBO is seen as Slot 1. For me, on this new System, this had the effect of killing OpenCL AND OpenGL Support on the GTX-1070 Card.

To remedy this, and get the 1070 Card running with OpenCL and OpenGL I had to physically transpose the two Cards. (AND, in the BIOS, Set PCI-e Slot 2 to be the Video Boot Slot.) This rendered the EVGA GTX-1050 to be without OpenCL and OpenGL Support from the Web Driver.

So, in close, ONLY my GTX-1070 is Crunching here at Collatz... Even though the "cc_config" state to use ALL GPUs... (See my detailed Posts discussing the issue with Mikey and David in The Last Person To Post Here Wins Thread in the Cafe.) This Faulty Web Driver issue was FIRST Reported by Petri33 and TBar in the SETI@Home Number Crunching Forum.

Still, I hope to have Andromeda run a LOOOOONG time here at Collatz. 😀


TL


Good luck with Andromeda! BTW, were you ever able to crunch C@H work on both GPUs? Note that I have two rigs with dual GTX-750Tis and I was able to tweak them to crunch C@H work on both GPUs. My problem is that projects were utilizing the CPU cores and I was unable to commit two CPU cores to support the C@H work on both GPUs. My recollection is that I had to do several things to solve the problem:

1. Adjust preferences to set C@H priority much higher than any other project's priority.
2. Adjust settings to prevent C@H from crunching CPU work units
3. Adjust preferences to load lots of work (minimum 5 days, up to 10 days if available)
4. Delete all projects from BOINC Manager
5. Add C@H to BOINC Manager; crunch for 24 hours to all the system to "equilibrate"
6. Then add other projects to BOINC Manager as desired

Dave
17) Message boards : Cafe : The Last Person To Post Here Wins 8 (Message 1538)
Posted 26 Feb 2019 by David Riese
Post:
<<--- Been sick for a week and a few days.

FINALLY got the doctor to prescribe antibiotics! (Saw him this past Tuesday the 19th.) Antibiotics were prescribed on Friday and picked up Saturday morning. Also have Prednisone to be used as needed to keep the asthma in check while the antibiotics do their thing.

This started in the back of my sinuses and moved into my lungs on the 19th. 😞😔😟🙁

Being sick SUCKS!!!
TL


Yes it does!! I'm glad the Dr finally came around and you are on the road to recovery!!

I guess due to the abuse by many people in the past NOT finishing their antibiotic course and causing mutant-antibiotic-resistant infections, now doctors try to minimize antibiotic use... He actually told me he wanted to wait 10 full days before prescribing antibiotics.

By Friday, I was SOOOO miserable, I called and insisted on an antibiotic prescription to be filled that day. [EDIT:] AND, I've always been medicine compliant. Unless I have a bad reaction to something, I ALWAYS finish medicine prescribed for me. Especially antibiotics.


TL


TL, I hope you are feeling better soon! Note that the prednisone can help relieve some of the symptoms of your crud. In fact, when I have an upper respiratory tract infection, my primary care provider will frequently give me a steroid shot but not prescribe an antibiotic unless s/he is convinced that it is needed. And yes, antibiotic-resistant bacteria are a major public health concern, so kudos for completing your pharmacotherapies as prescribed!

And I am a big proponent of TSA-precheck, even for folks who do not fly a lot. If it helps you avoid missing one flight, it pays for itself. Plus, I love the fact that I don't have to take off my shoes and I don't have to take my laptop out of my carryon bag.
18) Message boards : Cafe : The Last Person To Post Here Wins 8 (Message 1517)
Posted 14 Feb 2019 by David Riese
Post:
Hey, Mikey and TL:

I used to use the K-Cup Universal Reusable Filters to brew 100% Kona in my Keurig. However, I gave it up in favor of cold-brewed coffee, in part because I like to make my coffee in large batches. I grind my 100% Kona Coffee (Ferrari Coffee Company) or 100% Kauai Coffee (can't remember the vendor - I buy 24 pounds at a time at the Costco in Kona, HI) on the french press setting in a Cuisinart DBM-8 automatic burr mill. I then add 3/4 cup of grounds and filtered water to a 34 ounce double-walled stainless steel french press (Coffee Gator). I mix with a spoon to wet all of the grounds, cover the french press with saran wrap (to reduce diffusion loss of the aromatics), and allow the coffee to brew in my refrigerator for 24-48 hours. (I have four french presses so that I always have coffee available to drink.) I filter the grinds (using the french press) and transfer the coffee to a vacuum bottle (Yeti, Hydro, or Stanley). I add ice (to keep the coffee cold longer) and a little 2% milk (to smooth the taste and to keep the aromatics from diffusing out of the coffee). One batch typically lasts me the entire day. I am a bit of a coffee snob, as my maternal grandparents owned a small coffee farm in Holualoa, Big Island, HI as a side business.

David

P.S. Got my Federal tax refund less than 1 week after filing my return. Woo-hoo! Still can't file my state return, as the state has not yet finalized one of the forms that I need to file!
19) Message boards : Cafe : The Last Person To Post Here Wins 8 (Message 1491)
Posted 2 Feb 2019 by David Riese
Post:
A load of dark clothes in the washing machine. A 32 oz hydro flask of iced coffee on my desk. Bluetooth headphones are connecting me to hours of podcasts. It's time to do the taxes! Woo-hoo! At least I am winning!
20) Message boards : Cafe : The Last Person To Post Here Wins 8 (Message 1480)
Posted 30 Jan 2019 by David Riese
Post:
Yes, it is! 😀

Just got Daedalus, (Mac Pro 5,1), Updated to High Sierra's 2019-001 Security Update AND getting the New NVIDIA Web Driver 387.10.10.10.40.118 Updated with it. 😀😃😄😁

Andromeda is next in about half an hour. 😀
TL


That is very good!! Congratulations!!

I myself got my next to last pc updated to a 240gb SSD from the 120gb SSD it was using so it's back up and crunching again, it was down to 400mb of free space on the hardrive due to win10 not deleting stuff after it's done with it! The last pc isn't filling up the harddrive at this point so I'm just waiting for it to do that, it's still got almost 50gb free for some unknown reason, I don't think it's because it's running Win7 but it could be I guess.

I also switched 2 more pc's over to Linux Mint because the Win10 updates don't like my older hardware very much. I have 2 Dell Intel 6 core pc's that thru hyper-threading think they are 12 core cpu's that one of them just flat out refused to boot up under the new Win10 updates so it went to Linux Mint and is running just fine. The only problem with them is they can't handle any gpu newer than an Nvidia 750Ti so they aren't going to be around long I guess.

I also have 5 off-lease Intel dual Xeon quad core machines I bought very cheaply, there are 2 quad core cpu's in each pc that has 8 cpu cores thru hyper-threading, so 16 cpu cores for crunching in each pc. One of the problems is they are only clocked in the mid to low 2ghz range so aren't very fast, the 2nd problem is that most will only take an Nvidia 750Ti gpu as the max it can handle which isn't very fast either. I did put an Nvidia 760 in one of them and it did work but it does not work trying to put that gpu, or any other Nvidia 760, in any of the other pc's like it. As I move them over to Linux I may try again but they wouldn't even boot up with the 760 gpu at all the first time so I'm not hopeful. They are also struggling with the latest Win10 updates so may go to Linux Mint too.

YOU HAVE XEONS!!!!!?????!!!!! 😱😳😮

DUDE!!! Turn those bad boys into Hackintoshes!!! You could use the TonyMacx86 method and through MultiBeast Settings make them think they are Classic Mac Pros!!! (Up to a 5,1.) Although, from your description of the CPU Cores on them, a 2008 3,1 MultiBeast Setting would be closer to match what they actually are. On El Capitan, the 760 and 750 Cards would even give you the Mac Boot Screen. (The 760 is Native to MacOS, so you wouldn't even need the NVIDIA Web Driver, just the CUDA Driver. (CUDA Driver 8.0.71 works best in El Capitan.))

[EDIT:] Never mind the stricken through text... Check to see if they are E-Series!!! My real 5,1 has E5620 CPUs in it - Dual Xeon Quad Core 2.4GHz CPUs, 8 Cores, (like you), so 16 Threads... IF they are E-Series, you could use MultiBeast and go to the Mac Pro 5,1 Setting on each System. Otherwise, the Mac Pro 3,1 Setting would be best. (Especially if they are pre-2010 Xeons.) IF they are 2010 or newer, you could use the Mac Pro 5,1 Settings and go right to High Sierra.
TL


Yes 3 of them are the E Series ones and 2 are the X Series ones. But what advantage does going to Mac help me over Windows or Linux besides another whole new OS to learn?


I apologize for butting in, but one advantage of the 2008-2012 MacPros (MacPro 3,1; MacPro 4,1; MacPro 5,1) over PCs of a comparable vintage is that these MacPros have proven to be remarkably amenable to upgrades. My (2008) MacPro 3,1 and (2010) MacPro 4,1 units contain GTX 750Ti, GTX 680, GTX 770, and GTX 960 cards and others have put much more powerful cards into theirs. The GTX 680 and GTX 770 cards are rock solid stable out of the box through MacOS 10.14 Mohave, but note that Mohave requires a MacPro 4,1 or newer. ***More recent/powerful NVIDIA GPUs DO NOT function under Mohave*** Moreover, I hear a lot of users with newer NVIDIA cards (more contemporary than the GTX 680 and GTX 770) run into problems with MacOS 10.13 High Sierra security updates - NVIDIA drivers have to be updated every time that there is a security update to the MacOS and NVIDIA is sometimes slow in releasing those updates. One (but by no means only) place to learn about these issues is MacVideoCards.com, particularly the blog hosted there. To be candid, this is why I continue to run OS X 10.11 ElCapitan on my MacPro 3,1 and 4,1 units - I don't want to worry about OS upgrades that muck up NVIDIA GPU compatibility.

So, if you have an opportunity to "build" a hackintosh that behaves as a MacPro 4,1 or 5,1, you might want to grab it, particularly if you can't otherwise drive a more powerful NVIDIA GPU using your existing motherboard and CPUs.

By the way, I am composing this post on a 2008 MacPro 3,1 with 16 GB RAM, an NVIDIA GTX 960, an ATI Radeon HD 5770, a 1 TB SSD boot drive, and a 3 TB HD drive partitioned into an internal backup drive and a scratch drive. This is my main computer here in my work office, meaning that I use it for biostatistical analyses (GraphPad Prism), chemical structure depiction and analyses (ChemDraw Prime), digital image processing (Photoshop), desktop publishing (InDesign), as a terminal (for accessing my supercomputing resources) and to run other mainstream software. I continue to be amazed how much an 11-year-old platform can be upgraded and how useful that platform can continue to be.


Next 20


©2019 Jon Sonntag; All rights reserved