Open Beta test: SoG for NVidia, Lunatics v0.45 - Beta6 (RC again)

Message boards : Number crunching : Open Beta test: SoG for NVidia, Lunatics v0.45 - Beta6 (RC again)
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 20 · 21 · 22 · 23 · 24 · 25 · 26 . . . 32 · Next

AuthorMessage
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1818369 - Posted: 20 Sep 2016, 4:35:45 UTC - in response to Message 1818364.  

Note that this means I have nothing else to compare my readings to.

From Intel's ATX power specs ver. 2.1:

4.1.4  Voltage  Tolerances  
Tolerance for the motherboard power rails should comply with the values listed in Table 6. 
Table 6.  Voltage Tolerances 
Voltage Rail  Tolerance 
+5VDC           ± 5 % 
-5VDC (if used) ± 10 % 
+12VDC          ± 5 % 
-12VDC          ± 10 % 
+3.3VDC         ± 4 % 
+5VSB           ± 5 % 

Hope this helps ...

I think this is too wide a tolerance for graphic cards under full load to still produce valid results, especially the right-on-the-edge GTX 560. That was certainly the case when I ran a GTX 560. I would be looking closer to a 1.5% tolerance on the +12V rail. This being measured with a multimeter and not depending on a motherboard chipset sensor that can have very large errors depending on how it was setup in calibration by the motherboard vendor. Main thing to measure is the +12V rail NOT under load, that is not crunching, and then under full load of BOINC with whatever tasks per card you are using in your configuration. I would like to see the voltage sag under load.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1818369 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818370 - Posted: 20 Sep 2016, 4:35:59 UTC - in response to Message 1818364.  
Last modified: 20 Sep 2016, 4:41:52 UTC

Note that this means I have nothing else to compare my readings to.

From Intel's ATX power specs ver. 2.1:

4.1.4  Voltage  Tolerances  
Tolerance for the motherboard power rails should comply with the values listed in Table 6. 
Table 6.  Voltage Tolerances 
Voltage Rail  Tolerance 
+5VDC           ± 5 % 
-5VDC (if used) ± 10 % 
+12VDC          ± 5 % \
-12VDC          ± 10 % 
+3.3VDC         ± 4 % 
+5VSB           ± 5 % 

Hope this helps ...



. . Helpful info to have but he won't be testing the 3.3V rail from a molex connector :), nor the negative 5V/12V rails either for that matter. But still good to have for a reference.

Stephen

. . PS Never try and test a switch mode PSU's voltages without a load, they don't like it and tend to go ... POOF! Just in case you didn't know.

[Edit] to clarify that in view of Keith's remark I am talking about no load at all as in not connected to the motherboard or equivalent device, while Keith's comment refers to a nominal load such as a motherboard et al at idle which is safe to do.

.
ID: 1818370 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1818372 - Posted: 20 Sep 2016, 4:40:13 UTC - in response to Message 1818364.  
Last modified: 20 Sep 2016, 4:43:05 UTC

[pre]
4.1.4 Voltage Tolerances
Tolerance for the motherboard power rails should comply with the values listed in Table 6.
Table 6. Voltage Tolerances
Voltage Rail Tolerance
+5VDC ± 5 %
+12VDC ± 5 %
Hope this helps ...

I've had issues when the error was less than 2%, however in those cases it was due to faulty caps; so lots of noise on the rails & the DC voltage still being pretty close to what it should have been.
Lower voltages than specified will generally give more issues than higher voltages for a given deviance from the specified value.
eg
12.0V nominal
11.8v is more likely to give problems than 12.2V.


EDIT- would be a good idea to monitor the voltages while booting up. Ideally there should be no change from when the system is first turned on to when the desktop is up & BOINC is crunching.
Grant
Darwin NT
ID: 1818372 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818373 - Posted: 20 Sep 2016, 4:45:30 UTC - in response to Message 1818372.  


I've had issues when the error was less than 2%, however in those cases it was due to faulty caps; so lots of noise on the rails & the DC voltage still being pretty close to what it should have been.
Lower voltages than specified will generally give more issues than higher voltages for a given deviance from the specified value.


. . I think you mean deviation, deviance is me on a Saturday night :)

Stephen

.
ID: 1818373 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1818375 - Posted: 20 Sep 2016, 4:49:31 UTC - in response to Message 1818370.  

Note that this means I have nothing else to compare my readings to.

From Intel's ATX power specs ver. 2.1:

4.1.4  Voltage  Tolerances  
Tolerance for the motherboard power rails should comply with the values listed in Table 6. 
Table 6.  Voltage Tolerances 
Voltage Rail  Tolerance 
+5VDC           ± 5 % 
-5VDC (if used) ± 10 % 
+12VDC          ± 5 % \
-12VDC          ± 10 % 
+3.3VDC         ± 4 % 
+5VSB           ± 5 % 

Hope this helps ...



. . Helpful info to have but he won't be testing the 3.3V rail from a molex connector :), nor the negative 5V/12V rails either for that matter. But still good to have for a reference.

Stephen

. . PS Never try and test a switch mode PSU's voltages without a load, they don't like it and tend to go ... POOF! Just in case you didn't know.

Good, accurate information. The 3.3V rail is mainly used on the CPU but unfortunately DOES make an appearance on the PCIe slot connector. The -12V rail can mostly be ignored in modern hardware, it only had a use in the RS-232C serial port which has mostly disappeared from current motherboards.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1818375 · Report as offensive
robertmiles
Volunteer tester

Send message
Joined: 16 Jan 12
Posts: 213
Credit: 4,117,756
RAC: 6
United States
Message 1818379 - Posted: 20 Sep 2016, 5:39:46 UTC

I finally caught a task postponed by a suspicious spike.

[url]http://setiathome.berkeley.edu/result.php?resultid=5166941169
[/url]
Appears to have some kind of restart. Mentioned a spike in the output, without mentioning suspicious.
ID: 1818379 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1818381 - Posted: 20 Sep 2016, 5:49:08 UTC - in response to Message 1818379.  

I finally caught a task postponed by a suspicious spike.

http://setiathome.berkeley.edu/result.php?resultid=5166941169

Appears to have some kind of restart. Mentioned a spike in the output, without mentioning suspicious.


Corrected to make link clickable
ID: 1818381 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818382 - Posted: 20 Sep 2016, 5:51:56 UTC - in response to Message 1818375.  




. . Helpful info to have but he won't be testing the 3.3V rail from a molex connector :), nor the negative 5V/12V rails either for that matter. But still good to have for a reference.

Stephen

. . PS Never try and test a switch mode PSU's voltages without a load, they don't like it and tend to go ... POOF! Just in case you didn't know.

Good, accurate information. The 3.3V rail is mainly used on the CPU but unfortunately DOES make an appearance on the PCIe slot connector. The -12V rail can mostly be ignored in modern hardware, it only had a use in the RS-232C serial port which has mostly disappeared from current motherboards.


. . Some hardware used +/- 5V to drive their RS232 but the good stuff used the +/- 12V as per the RS232 specs. As you point out, USB has rendered RS232 obsolete unless you really need to interface with older equipment that lacks USB ports. Though even then most would use a USB to RS232 adaptor. Can you think of what else uses the -5V? In any case even the 3.3V on the PCIe slot does not have a major function that I am aware of. Mainly it is the +5V and the +12V that do the heavy lifting. The -5V and -12V only ever had low current ratings as I remember. But for the +5V and +12V better to be a little high than a little low as Grant pointed out.

Stephen
ID: 1818382 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1818387 - Posted: 20 Sep 2016, 6:14:57 UTC - in response to Message 1818326.  
Last modified: 20 Sep 2016, 6:15:43 UTC

There's a suitable power plug or two. I just ordered a multimeter.

Will be very interesting to see how the 5/12v moves when BOINC is and is not running... can't say I've tested that, either ...

If you really want to see something scary you should use an oscilloscope to view your power rails at idle & with BOINC running apps.
I did that a few years ago on one of my machines.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1818387 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1818388 - Posted: 20 Sep 2016, 6:15:07 UTC - in response to Message 1818382.  


. . Some hardware used +/- 5V to drive their RS232 but the good stuff used the +/- 12V as per the RS232 specs. As you point out, USB has rendered RS232 obsolete unless you really need to interface with older equipment that lacks USB ports. Though even then most would use a USB to RS232 adaptor. Can you think of what else uses the -5V? In any case even the 3.3V on the PCIe slot does not have a major function that I am aware of. Mainly it is the +5V and the +12V that do the heavy lifting. The -5V and -12V only ever had low current ratings as I remember. But for the +5V and +12V better to be a little high than a little low as Grant pointed out.

Stephen

And of course, RS-232 to USB adapters have been the bane of my hobby since I still have to use legacy hardware with serial ports. I typically have 3 in use to drive my scopes, mount and camera interfaces every night. Only in the last couple of years or so have the manufacturers started transitioning to USB or Ethernet interfaces. I believe the -5V and -12V were only used for balanced transmission systems like RS-232 or RS-432 or sometimes in early DC-DC inverters or analog op-amp audio circuits. The ATX12V 1.3 standard dropped the negative rails entirely.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1818388 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818400 - Posted: 20 Sep 2016, 7:42:26 UTC - in response to Message 1818387.  

There's a suitable power plug or two. I just ordered a multimeter.

Will be very interesting to see how the 5/12v moves when BOINC is and is not running... can't say I've tested that, either ...

If you really want to see something scary you should use an oscilloscope to view your power rails at idle & with BOINC running apps.
I did that a few years ago on one of my machines.



. . I suspect I don't really want to see that :)

Stephen

.
ID: 1818400 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1818423 - Posted: 20 Sep 2016, 10:37:44 UTC - in response to Message 1818400.  
Last modified: 20 Sep 2016, 10:38:16 UTC

There's a suitable power plug or two. I just ordered a multimeter.

Will be very interesting to see how the 5/12v moves when BOINC is and is not running... can't say I've tested that, either ...

If you really want to see something scary you should use an oscilloscope to view your power rails at idle & with BOINC running apps.
I did that a few years ago on one of my machines.



. . I suspect I don't really want to see that :)

Stephen

.


Have done it out of curiosity since my now aged Corsair 1200AX PSU came with ferrite chokes on the PCIe power cables. turned out the GTX480 I was running at the time was quite capable of polluting the 12V on the single rail PSU, and the chokes did the trick. Can easily see multi-GPU rigs suffering stability problems without some or another approach to mitigating the noise. While I believe newer GPUs would tend to use higher frequency converters and use less current than a GTX 480, it was a bit of an eyeopener as to the possibilities for subtle system instability, and the importance of quality power.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1818423 · Report as offensive
robertmiles
Volunteer tester

Send message
Joined: 16 Jan 12
Posts: 213
Credit: 4,117,756
RAC: 6
United States
Message 1818527 - Posted: 20 Sep 2016, 23:41:19 UTC

Until the voltmeter arrives, I won't be able to continue the planned tests.

So, I've decided to try my other graphics boards in place of the 560 to see if they are also affected by whatever the problem is.
ID: 1818527 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818530 - Posted: 21 Sep 2016, 0:02:51 UTC - in response to Message 1818527.  

Until the voltmeter arrives, I won't be able to continue the planned tests.

So, I've decided to try my other graphics boards in place of the 560 to see if they are also affected by whatever the problem is.


. . A very good idea. That should prove the problem into either the GTX560 or the rig itself. Should have tried that earlier. You may have to reset your fan profile in Afterburner as it will react to the change of GPU.

Stephen

.
ID: 1818530 · Report as offensive
robertmiles
Volunteer tester

Send message
Joined: 16 Jan 12
Posts: 213
Credit: 4,117,756
RAC: 6
United States
Message 1818546 - Posted: 21 Sep 2016, 1:56:48 UTC
Last modified: 21 Sep 2016, 1:58:42 UTC

GTX 560 alternates:

I first tried the GTS 450, previously used in the same computer. The software initially thought it was a 9800 GT (the original graphics board for this computer), and BOINC didn't even recognize it as usable.

I then tried reinstalling the 362.00 driver; the driver installation program didn't recognize the GTS 450 as compatible with that driver.

I then tried installing the oldest driver I already had downloaded, 340.52. It installed, but still considered the board a 9800 GT.

I'm out of ideas, so I asked Nvidia. Do you have any other ideas?
ID: 1818546 · Report as offensive
robertmiles
Volunteer tester

Send message
Joined: 16 Jan 12
Posts: 213
Credit: 4,117,756
RAC: 6
United States
Message 1818547 - Posted: 21 Sep 2016, 2:21:18 UTC - in response to Message 1817199.  

Any other ideas on what to try after that?

I go back to power. If you have access to a voltmeter, look at your 12v (yellow). Especially see what happens when you start up and shut down BOINC.


Looks like I'll need to know what the other wires to the Molex connectors are also. Two black and one red on some, black, brown, and an empty space on others.

I've noticed that one of the electrolytic capacitors on the motherboard may be bulging - the top is rounded on that one, but flat on all the others. Not something I can still fix without help, if that's what it is.
ID: 1818547 · Report as offensive
Profile Jimbocous Project Donor
Volunteer tester
Avatar

Send message
Joined: 1 Apr 13
Posts: 1849
Credit: 268,616,081
RAC: 1,349
United States
Message 1818549 - Posted: 21 Sep 2016, 2:35:24 UTC - in response to Message 1818547.  

Any other ideas on what to try after that?

I go back to power. If you have access to a voltmeter, look at your 12v (yellow). Especially see what happens when you start up and shut down BOINC.


Looks like I'll need to know what the other wires to the Molex connectors are also. Two black and one red on some, black, brown, and an empty space on others.

I've noticed that one of the electrolytic capacitors on the motherboard may be bulging - the top is rounded on that one, but flat on all the others. Not something I can still fix without help, if that's what it is.

"Normal" wiring on a Molex, specifically the larger 4-pin connector used for powering disk drives and the like, is black centers or sometimes a black and a brown (ground), Red (+5v) and Yellow (+12v) to the outsides. Should be able to tell easily with a meter;,nothing will be that far gone.
Bulging capacitor tops are usually the kiss of death, and could quite possibly explain your various weirdnesses. I had an old MSI mobo that developed that, and it did begin throwing a lot of error WUs.
ID: 1818549 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818560 - Posted: 21 Sep 2016, 4:06:29 UTC - in response to Message 1818546.  

GTX 560 alternates:

I first tried the GTS 450, previously used in the same computer. The software initially thought it was a 9800 GT (the original graphics board for this computer), and BOINC didn't even recognize it as usable.

I then tried reinstalling the 362.00 driver; the driver installation program didn't recognize the GTS 450 as compatible with that driver.

I then tried installing the oldest driver I already had downloaded, 340.52. It installed, but still considered the board a 9800 GT.

I'm out of ideas, so I asked Nvidia. Do you have any other ideas?


. . With boards of that era there are some tricky things with drivers, the later drivers don't activate OpenCL on the older cards, nor (on some) the CUDA potential so Boinc will not see it as a viable GPU. There is a message somwhere about what drivers to use for pre-fermi (and maybe early fermi) GPUs. I was crunching for a while on an 8600GT with 256 MB ram. About as minimal as you can get and still crunch with it but I had to use an early driver 337.88.

. . I think though that the GTS450 is actually a Fermi card, so I would do a complete uninstall of all video drivers and then try a clean install with the latest driver from Nvidia, it might then see the card for what it is. I use Iobit Advanced system care which has a Power uninstall option that cleans that stuff out pretty well, it removes all traces so the new install should be clean.

Stephen

.
ID: 1818560 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1818561 - Posted: 21 Sep 2016, 4:14:20 UTC - in response to Message 1818547.  
Last modified: 21 Sep 2016, 4:23:11 UTC

Any other ideas on what to try after that?

I go back to power. If you have access to a voltmeter, look at your 12v (yellow). Especially see what happens when you start up and shut down BOINC.


Looks like I'll need to know what the other wires to the Molex connectors are also. Two black and one red on some, black, brown, and an empty space on others.

I've noticed that one of the electrolytic capacitors on the motherboard may be bulging - the top is rounded on that one, but flat on all the others. Not something I can still fix without help, if that's what it is.



. . On the Molex connector the black wires are ground and the red is +5V while the Yellow is +12 (unless I have misremembered that part :{ ) but the connectors are keyed to avoid issues, that is why they only go in one way. Usually a brown wire is only substituted for a red wire (brown&blue replace red&black respectively due to colour blindness confusion) so I would expect to have brown black black yellow. I don't know why a molex would only have +5V and no +12V. If the molex connectors feeding the adapter for the GPU do not have yellow wires then you have a problem because the purpose of external PCIe power connectors is to provide more power at +12v not +5V.

. . If there is a dying cap on the MoBo good luck, replacing them is not fun, and it may be a symptom on something else not right. But caps can get like that just from age, they "dry out" and swell up, then pop eventually.

. . Sadly old electronics don't last forever.

Stephen

.
ID: 1818561 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13161
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1818567 - Posted: 21 Sep 2016, 5:00:21 UTC - in response to Message 1818561.  

Yeah, have to agree about the "kiss of death" regarding bulging capacitors on motherboards. Unlikely too that he will be able to see small ripple voltages with just a simple DVM unless he's ordered a really expensive one which the bulging capacitor almost certainly is producing. If I remember an earlier post correctly, the motherboard is a 1999 Dell vintage. That puts in right at the beginning of the bad capacitor plague caused by the bad stolen industrial espionage electrolyte recipe.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1818567 · Report as offensive
Previous · 1 . . . 20 · 21 · 22 · 23 · 24 · 25 · 26 . . . 32 · Next

Message boards : Number crunching : Open Beta test: SoG for NVidia, Lunatics v0.45 - Beta6 (RC again)


 
©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.