Advanced search

Message boards : Graphics cards (GPUs) : Problem with GTX 1070

Author Message
Gorgar66
Send message
Joined: 20 Aug 17
Posts: 3
Credit: 58,522,925
RAC: 23
Level
Thr
Scientific publications
wat
Message 48749 - Posted: 26 Jan 2018 | 16:19:09 UTC

Hi all,
since the windows 10 update to 1709, my graphics card often stuck during crunching. Time is counting, but no progress with task. After I restarted Pc, it is running fine again. Before the update, I never had errors,after task is done, but now there are some. It's no difference, if I overclock it, or not. It happens in both ways.
Is this because of surfing in Internet, or doing other things with pc? No bigger tasks, just unpacking, or starting programs, or copying data to external Hard drive.
Greetings
Gorgar66

mmonnin
Send message
Joined: 2 Jul 16
Posts: 173
Credit: 288,961,689
RAC: 43,983
Level
Asn
Scientific publications
wat
Message 48754 - Posted: 26 Jan 2018 | 19:47:24 UTC - in response to Message 48749.

What driver do you have. W10 update can install some drivers that don't work very well for crunching on some projects. Typically thats for OpenCL projects but I'd check anyway.

Gorgar66
Send message
Joined: 20 Aug 17
Posts: 3
Credit: 58,522,925
RAC: 23
Level
Thr
Scientific publications
wat
Message 48756 - Posted: 27 Jan 2018 | 9:11:32 UTC

I'm using Version 390.65.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 666
Credit: 2,498,095,550
RAC: 2,354
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 48757 - Posted: 27 Jan 2018 | 12:43:54 UTC - in response to Message 48749.

Hi all,
since the windows 10 update to 1709, my graphics card often stuck during crunching. Time is counting, but no progress with task. After I restarted Pc, it is running fine again. Before the update, I never had errors,after task is done, but now there are some. It's no difference, if I overclock it, or not. It happens in both ways.
Is this because of surfing in Internet, or doing other things with pc? No bigger tasks, just unpacking, or starting programs, or copying data to external Hard drive.
Greetings
Gorgar66


Read this thread https://gpugrid.net/forum_thread.php?id=4660

Gorgar66
Send message
Joined: 20 Aug 17
Posts: 3
Credit: 58,522,925
RAC: 23
Level
Thr
Scientific publications
wat
Message 48758 - Posted: 27 Jan 2018 | 22:52:03 UTC

27.01.2018 23:41:09 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><cc_status> <network_status>2</network_status> <ams_password_error>0</ams_password_error> <task_s'
27.01.2018 23:41:09 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_messages> <seqno>247919</seqno> <translatable/></get_messages></boinc_gui_rpc_request>'
27.01.2018 23:41:09 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><msgs><msg> <project></project> <pri>1</pri> <seqno>247920</seqno> <body><![CDATA[[gui_rpc] GUI RPC '
27.01.2018 23:41:09 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_results><active_only>1</active_only></get_results></boinc_gui_rpc_request>'
27.01.2018 23:41:09 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><results><result> <name>e84s24_e68s24p0f46-PABLO_p27_wild_0_sj403_IDP-1-2-RND0747_0</name> <wu_na'
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_cc_status/></boinc_gui_rpc_request>'
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><cc_status> <network_status>2</network_status> <ams_password_error>0</ams_password_error> <task_s'
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_messages> <seqno>247925</seqno> <translatable/></get_messages></boinc_gui_rpc_request>'
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><msgs><msg> <project></project> <pri>1</pri> <seqno>247926</seqno> <body><![CDATA[[gui_rpc] GUI RPC '
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_results><active_only>1</active_only></get_results></boinc_gui_rpc_request>'
27.01.2018 23:41:10 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><results><result> <name>e84s24_e68s24p0f46-PABLO_p27_wild_0_sj403_IDP-1-2-RND0747_0</name> <wu_na'
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_cc_status/></boinc_gui_rpc_request>'
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><cc_status> <network_status>2</network_status> <ams_password_error>0</ams_password_error> <task_s'
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_messages> <seqno>247931</seqno> <translatable/></get_messages></boinc_gui_rpc_request>'
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><msgs><msg> <project></project> <pri>1</pri> <seqno>247932</seqno> <body><![CDATA[[gui_rpc] GUI RPC '
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC Command = '<boinc_gui_rpc_request><get_results><active_only>1</active_only></get_results></boinc_gui_rpc_request>'
27.01.2018 23:41:11 | | [gui_rpc] GUI RPC reply: '<boinc_gui_rpc_reply><results><result> <name>e84s24_e68s24p0f46-PABLO_p27_wild_0_sj403_IDP-1-2-RND0747_0</name>
Here is a part of the log, when it got stuck.
Thx to Betting Slip

kksplace
Send message
Joined: 4 Mar 18
Posts: 12
Credit: 5,192,825
RAC: 0
Level
Ser
Scientific publications
wat
Message 49387 - Posted: 2 May 2018 | 20:27:40 UTC

It seems I have at least a similar problem and the related threads don't seem to quite answer the (perceived?) problem.

First, this is all new to me -- so be gentle...

My rig is a Dell 8930 with an i5-8400 (8th gen) 6-core and a GTX 1070 with 391.35 driver.

I notice that my GPUGRID work units require a little more care. They sometimes will 'stall' -- they will show "Running" but not progress. A small percentage of the time, they will eventually "reset" and start back, but this can take minutes to hours. However, a majority of the time when this happens I must suspend the task and/or GPU operations, and then resume for them to start back up.

This mostly happens when either browsing the web (this is a family computer) or performing a CPU intensive task. (Happens on all browsers, not just FireFox.) However, my Einstein@Home GPU tasks will start back up nicely on their own in the same circumstances.

The only thing that has seemed to help some is to change my "Suspend when non-BOINC CPU usage is above" to 50% (from 25%), which has decreased the number of times this happens.

Any ideas for a Newbie (or is it Noobie?)?

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 64
Credit: 2,079,781,557
RAC: 6,674,771
Level
Phe
Scientific publications
watwatwat
Message 49388 - Posted: 3 May 2018 | 4:15:45 UTC - in response to Message 49387.
Last modified: 3 May 2018 | 4:15:55 UTC

Make of 1070? Are you overclocking it in any way?

kksplace
Send message
Joined: 4 Mar 18
Posts: 12
Credit: 5,192,825
RAC: 0
Level
Ser
Scientific publications
wat
Message 49389 - Posted: 3 May 2018 | 10:42:44 UTC - in response to Message 49388.

NVIDIA GeoForce... and no, not overclocking.

I have installed MSI Afterburner and made the fan curve a little more 'aggressive', per a recommendation listed on other forums. While running GPUGRID workunits along with 3 other CPU cores working at the same time, my GPU temp stabilizes about 77 degrees C. It doesn't seem to voltage limit very much -- very occasionally.

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 399
Credit: 2,745,215,917
RAC: 1,166,933
Level
Phe
Scientific publications
watwat
Message 49390 - Posted: 3 May 2018 | 11:18:47 UTC - in response to Message 49387.



I notice that my GPUGRID work units require a little more care. They sometimes will 'stall' -- they will show "Running" but not progress. A small percentage of the time, they will eventually "reset" and start back, but this can take minutes to hours. However, a majority of the time when this happens I must suspend the task and/or GPU operations, and then resume for them to start back up.

If you ever have a GPUGrid Work Unit (WU) show running and it's back to a low percentage like 1% but the elapsed time is already many hours, I would abort this task as it will not give a valid result even after many hour of completion resulting in a lot of wasted time, energy and heat. These WUs are a bit more finicky than other GPU projects. For example after clicking suspend in BOINC, it will take at least 10 seconds for the GPUGrid app to stop.

The only thing that has seemed to help some is to change my "Suspend when non-BOINC CPU usage is above" to 50% (from 25%), which has decreased the number of times this happens.

This largely has to do with the previous issue I mentioned, if your CPU usage raises and drops fast, which happens a lot at 25% and less at 50%, it will try to disable computing all the time. Since the GPUGrid app does not like to be stopped in a timely manner I would straight up disable this feature. Windows will take care of you in terms of determining which app should perform best, in this case the app you're actively using such as chrome. If you wish to have access to the full resources, for example during a game, you can add "exclusive applications" in BOINC which will completely disable computing while you have this application running. I've found this to be the most fool-proof and seamless way to compute with no slow downs.


kksplace
Send message
Joined: 4 Mar 18
Posts: 12
Credit: 5,192,825
RAC: 0
Level
Ser
Scientific publications
wat
Message 49393 - Posted: 3 May 2018 | 23:30:07 UTC - in response to Message 49390.

Thank you for the recommendations. I will check back in a couple of weeks with how it works.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 666
Credit: 2,498,095,550
RAC: 2,354
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 49394 - Posted: 4 May 2018 | 8:03:50 UTC - in response to Message 49390.

I think a lot of people have a problem with the over cautious default settings in Boinc Manager. They should either relax them or produce a PDF manual which is downloaded with Boinc for newcomers.
____________
Radio Caroline, the world's most famous offshore pirate radio station.
Great music since April 1964. Support Radio Caroline Team -
Radio Caroline

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 399
Credit: 2,745,215,917
RAC: 1,166,933
Level
Phe
Scientific publications
watwat
Message 49395 - Posted: 4 May 2018 | 11:18:51 UTC - in response to Message 49394.

I think a lot of people have a problem with the over cautious default settings in Boinc Manager. They should either relax them or produce a PDF manual which is downloaded with Boinc for newcomers.

I agree, most people have multicore machines with operating systems that allocate resources to what the user is doing. Most of the time the user can't even tell the CPU is pinned at 100%. I don't think the "CPU Usage" and "CPU time" settings should even be enabled at stock.

kksplace
Send message
Joined: 4 Mar 18
Posts: 12
Credit: 5,192,825
RAC: 0
Level
Ser
Scientific publications
wat
Message 49448 - Posted: 11 May 2018 | 10:23:53 UTC - in response to Message 49393.

It has only been a week, but I think I can safely reply -- it worked! I have processed all my GPUGRID WUs without problem this past week. And due to your recommendations I have added another core to the crunching as well (for another project).

There have been a lot of new WUs this week, which may also had an affect, but you are right that Windows will handle my short-lived attempts to go beyond 100% very well. I only added one game (Overwatch for daughter) to the 'exclusive applications' list.

Happy crunching!

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 399
Credit: 2,745,215,917
RAC: 1,166,933
Level
Phe
Scientific publications
watwat
Message 49449 - Posted: 11 May 2018 | 10:47:11 UTC - in response to Message 49448.

It has only been a week, but I think I can safely reply -- it worked! I have processed all my GPUGRID WUs without problem this past week. And due to your recommendations I have added another core to the crunching as well (for another project).

Now just keep an eye on temps. Now that you're running with more load, the only thing to care about are CPU and GPU temps. If they look good (less than 80s) and the motherboard around the CPU has a bit of airflow, I would say you're good for the long haul.

Happy Crunching!

JoergF
Avatar
Send message
Joined: 20 Apr 15
Posts: 270
Credit: 765,019,820
RAC: 1,310,879
Level
Glu
Scientific publications
watwat
Message 49450 - Posted: 11 May 2018 | 11:29:25 UTC - in response to Message 49449.

Now just keep an eye on temps. Now that you're running with more load, the only thing to care about are CPU and GPU temps. If they look good (less than 80s) and the motherboard around the CPU has a bit of airflow, I would say you're good for the long haul.


+1 Good Point.

But I would even use MSI Afterburner and keep the temps below 70°C as GPUs have a single sensor only and dont tell you the whole truth about RAM chip, voltage Regulator, capacitor and other temperatures on the board. I have seen an EVGA brand board thermogram with significant temperature variations, so better be on the safe side. Also running 2 concurrent jobs on one GPU helps to dampen sudden temperature drops when one job is finished and your PC applies for new work.

____________
I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 399
Credit: 2,745,215,917
RAC: 1,166,933
Level
Phe
Scientific publications
watwat
Message 49451 - Posted: 11 May 2018 | 12:35:28 UTC - in response to Message 49450.
Last modified: 11 May 2018 | 12:51:50 UTC

Now just keep an eye on temps. Now that you're running with more load, the only thing to care about are CPU and GPU temps. If they look good (less than 80s) and the motherboard around the CPU has a bit of airflow, I would say you're good for the long haul.

But I would even use MSI Afterburner and keep the temps below 70°C

I couldn't agree more, it's just not always possible, especially if you have a founders edition or stock cooler of any sort.

I like to keep all temps as low as possible to mitigate the effects of thermal expansion and contraction on GPUs. I also care about fan life so I don't run anything near 100% fan speed. Less than 70°C for any processor is its happy place.

kksplace
Send message
Joined: 4 Mar 18
Posts: 12
Credit: 5,192,825
RAC: 0
Level
Ser
Scientific publications
wat
Message 49460 - Posted: 11 May 2018 | 21:56:50 UTC - in response to Message 49450.

I installed MSI Afterburner several weeks ago, and it has helped. Before, when crunching GPUGRID, my GPU temp would stay at 81 degrees and often therm-throttle/limit. With Afterburner, I played with the fan profile and now have the GPU settle out at 75 to 77 degrees when crunching and 72 degrees when Overwatch is played. (Fan speed at 80% at 77 degrees).The voltage limits at 1043mV in this setup. Afterburner shows the CPU temps averaging at 72 degrees with 5 cores working and 69-70 degrees with 4 cores.

Afterburner is only able to see/control the GPU fan -- not the stock CPU cooler or case fan. The case fan is definitely working hard though.

There isn't much room in the Dell 8930 case. Only a couple of coolers will fit according to several forums and a liquid cooler setup has problems since the radiator and fan apparently don't fit in the hole where the case fan now sits.

My point -- I am learning a lot and know a lot more about what to look for (maybe build?) when my wife finally lets me buy the next computer. Not really unhappy at the moment. What I have now is far better than anything before -- just smarter for next time.

Thank you for the help, and happy to be crunching.

Post to thread

Message boards : Graphics cards (GPUs) : Problem with GTX 1070