How can one read these dmp files Jacob?

---

Greetings from TJ

I don't know. I think, if you have the Windows SDK and development tools installed, and have Windows symbols available, you might be able to step through them. But that's all beyond my ability.

The short story is: If you have a .dmp file in your C:\Windows\LiveKernelReports\WATCHDOG directory, it means your GPU had a TDR... and either an application was faulty, or the GPU was faulty, or (the most common case) you are pushing your GPU too hard in terms of Core Clock or Memory Clock.

I still fully recommend Heaven 4.0, on the maximum settings I described a couple posts up, running overnight, to confirm stability. Once I did that for my 2 GTX 660 Ti's, and found that I had to decrease the clock on one and could increase the clock on the other, and since then I have had 0 problems with GPUGrid and with iRacing.

Not trying to spam this thread. Retvari, I hope you can get your issue figured out, and if my suggestion doesn't help you, then I apologize.

How can one read these dmp files Jacob?

Having Visual 2013 helps with reading certain files, but most can be read with notepad, if text is involved. (watchdog are mostly text files)

If you have a .dmp file in your C:\Windows\LiveKernelReports\WATCHDOG directory

You can read these files with notepad. Run as admin, you'll see a prompt "user doesn't have access" if in non-admin mode.

If you have game that hard on a GPU (BF4, Metro2033/Last Light) if you don't have any games on you're disk- Heaven is great tool to stress, Or 3Dmark Vantage benchmark has looping for TMU, ROP, Memory test that strain cards to limits. The extreme Firestrike benchmark loops, and will fail an card overclocked. If have you Nvidia Cuda samples: the n-body test can be looped, a card will also fail the random number samples, if over clocked too high. This how I Found my cards best temps and voltage.

With a custom bios and Nvidia Inspector Bat files, as Jacob has shown for setting "Max boost", works wonders once know cards limit for core/memory speeds and voltage. New Gm204 can run at 1.000V with a 1.2 GHz speed. 1.025 voltage also. Overclocking records past 2GhZ (GM204 card is first ever to break 2Ghz) with L2N. Many 1.5 GHz speeds have be reached with air cooling and stock voltage. GM204 is truly an engineering feat. The amount features added along with new filtering tech really raises Molecular Dynamics function for single precision.

Have you considered running Heaven, to determine how far you may need to downclock? If you can get Heaven to run at max settings overnight with no issues, then I'd consider it stable.

When I first tested the card, I did. The only application failed is GPUGrid's.
See the first post of this thread.
BTW the card seems to be stable at 3.2GHz, but different workunit batches could use different parts of the GPU.
I suspect that something messed up with the GDDR5 voltage, or the PSU of the memory subsystem on this card series.

I see this card runs ~80C. Do you know Voltage control temps on card? VRM runs over 100C on some GTX780ti cards.(rated for 110C for you're card.) Gigabyte been worse offender, from viewing 780ti owner boards. eVGA and Asus 780ti's VRM is rated at 120-125c.

Do you know temps for DDR memory? These temps run really hot on certain GTX780ti's

New Zotac's GTX970/980 along eVGA's 900 series have highest rated core/boost speeds.

Thanks for your help eXaPower, but I have tried notepad, wordpad but no normal reading is possible.
I did not game but have a few dmp files, perhaps when a drives crashed with GPUGRID? And now I am interested in what is in those files.

---

Greetings from TJ

Thanks for your help eXaPower, but I have tried notepad, wordpad but no normal reading is possible.
I did not game but have a few dmp files, perhaps when a drives crashed with GPUGRID? And now I am interested in what is in those files.

I neglected to mention Win8.1 notepad will open these type files- only if Visual been installed prior, but license expired or is current. During a period you're host can be tweaked- Try fiddling with some windows System32 program list to see if one will allow it opening it or......

You can gain access to newest Visual version (To create/test custom made programs/prior or custom files) with new CUDA 6.5.19 toolkit, if want full visual VC++ redis, Microsoft has developer account (you block all info being sent to them- Just do a custom install.) Trial period for 60-90 days. Once CUDA toolkit/Visual are linked together a world learning DIY programs is opened. Nvidia has a debugging program with they're Registered Developer program. Intel has a great AVX/FMA3 DIY programming tool. AMD is a HSA member.

Linux is intertwined with NVidia HSA. A lot of options are available. Freedom of choice.

I see this card runs ~80C.

No. This card (GPU1) runs at 65-70°C.
The other card (GPU0) - which is a standard NVidia design - runs fine on 3.5GHz at 80°C.

Do you know Voltage control temps on card?

I don't know voltage control temps, but I think it should be lower than the other card's, as this card has more phase on that VR, and this card has better cooling.

Do you know temps for DDR memory? These temps run really hot on certain GTX780ti's

I don't know that either, but the same reasoning applies to the RAM chips as for the VRM chips.

I see this card runs ~80C.

No. This card (GPU1) runs at 65-70°C.
The other card (GPU0) - which is a standard NVidia design - runs fine on 3.5GHz at 80°C.

Do you know Voltage control temps on card?

I don't know voltage control temps, but I think it should be lower than the other card's, as this card has more phase on that VR, and this card has better cooling.

Do you know temps for DDR memory? These temps run really hot on certain GTX780ti's

I don't know that either, but the same reasoning applies to the RAM chips as for the VRM chips.

If time permits - thin gauge wires with correct metal probes to attach? (Do you have tools for you're Gigabyte Ti?) , you can manually read temp with proper equipment. (Or if you already have kit for electrical/ or temp readouts.)

How can one read these dmp files Jacob?

Hi TJ you can read those .dmp files with bluescreenviewer
http://www.nirsoft.net/utils/blue_screen_view.html#DownloadLinks

Just download the app, unzip it, and run it from the resulting file called BlueScreenView.exe, then go to the options menu & click on "advanced Options", then click the radio button that says "load a single minidump file" then just direct it to the folder that was mentioned. C:\Windows\LiveKernelReports\WATCHDOG, & pick the .dmp file you want.

I hope the results give you what your looking for.

I had another failed workunit on this card, so I took another 100MHz off, it's now running at 3.1GHz GDDR5 clock.

Thanks JugNut I will try it over the weekend.

---

Greetings from TJ

Hello,

Wanted to try GPUGRID today, but UNFORTUNATELY, still ONLY ERRORS within 15 seconds :/

Tried 1 GPU only (0 and then 1) but same result.

Temp 65°

What's happening ?

What can we do ?

EDIT : Have decreased the Power target down to 60 % + short WU's, but same problem. All tasks = errors.

388-NOELIA_20MGK36I-2-5-RND2053_0 10116890 160926 27 Sep 2014 | 9:40:02 UTC 27 Sep 2014 | 9:50:20 UTC Erreur en cours de calculs 8.27 2.54 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
14-NOELIA_20MGK36I-2-5-RND3099_0 10116873 160926 27 Sep 2014 | 9:33:35 UTC 27 Sep 2014 | 9:41:41 UTC Erreur en cours de calculs 67.69 12.40 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
522-NOELIA_20MGWT-2-5-RND4515_0 10116837 160926 27 Sep 2014 | 9:33:35 UTC 27 Sep 2014 | 9:50:20 UTC Erreur en cours de calculs 2.40 0.00 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
249-NOELIA_20MGK36I-2-5-RND5686_0 10116835 160926 27 Sep 2014 | 9:33:35 UTC 27 Sep 2014 | 9:41:41 UTC Erreur en cours de calculs 10.39 2.96 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
747-NOELIA_20MGK36I-2-5-RND8713_0 10116750 160926 27 Sep 2014 | 9:41:42 UTC 27 Sep 2014 | 9:58:09 UTC Erreur en cours de calculs 7.78 2.95 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
232-NOELIA_20MGK36I-2-5-RND3755_0 10116603 160926 27 Sep 2014 | 9:33:35 UTC 27 Sep 2014 | 9:40:02 UTC Erreur en cours de calculs 7.30 2.34 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)
406-NOELIA_20MGWT-2-5-RND6078_0 10116559 160926 27 Sep 2014 | 9:41:42 UTC 27 Sep 2014 | 9:58:09 UTC Erreur en cours de calculs 4.18 2.22 --- Long runs (8-12 hours on fastest card) v8.41 (cuda60)

Best Regards

Philippe

Your 3.1GHz ties in well with what I thought the situation might be,

Assuming H5GQ2H24AFR R2C, these require 1.6V to support 3.5GHz.

Excluding the possibility of bad GDDR5 and bad circuitry (which we can do nothing about anyway), my guess is that the card isn't supplying the necessary 1.6V, and is either supplying 1.5V or 1.35V - possibly 1.5V for some people and 1.35V for others; with 1.35V perhaps being sufficient for 2.7GHz and 1.5V sufficient for 3.1GHz. This ties in with what has been reported here and suggests a firmware, driver or OS issue.

My solution would be to stick with it at 3.1GHz, if it proves to be stable, or sell the card and get an equivalent second hand card that does run at 3.5GHz.

288larsson who posted in this thread also has a Gigabyte GTX 780 Ti OC (Windforce 3x) GPU.

Alas I don't know how to change the GDDR5 voltage.

Hello !

How comes these 780Ti are doing OK on all other projects but GPUGRID ?

If it was a hardware issue, one should have problems on all projects I guess ?

Thank You

Hello !

How comes these 780Ti are doing OK on all other projects but GPUGRID ?

If it was a hardware issue, one should have problems on all projects I guess ?

Thank You

Hello Philippe,

It's because the GPUGrid app is the most advanced one. It's compiled with the latest CUDA version, so it can utilize the card like no other project's app can. The "GPU usage" measurement is misleading.

Could you please specify all details of your GTX780Ti (Manufacturer, model, clocks), and your PSU (Manufacturer, model, wattage, efficiency)?

Hello Zoltan,

Thank you for your message.

the 2 * GTX780Ti = Gigabyte Windforce GV-N78TWF3 - 3GD

http://www.gigabyte.fr/products/product-page.aspx?pid=4912#sp

The PSU is a brand new CORSAIR RM1000 / 1000W 80+ Gold

MB = ASUS Z87PRO / CPU i7-4770K / WC AIO NEPTON 140 XL / RAM DDR3 Corsair vengeance 2 x 8 go 1600 Mhz CL10 LP

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


NB Should one allow 1 CPU core / WU or is 0.5 still OK ?

I see that your CPU time = GPU time, and on the only WU I finished, the CPU use is about 25 % of GPU time ? http://www.gpugrid.net/result.php?resultid=13111203

I have finished only 1 WU since I plugged in these cards last week ...

Thank You very much for your help !

Best Regards

Philippe

EDIT : Have modified app_config to 1 CPU / 1 GPU + Have decrased the "MEM CLOCK" from 3500 down to 3100, and it looks like the WU won't crash. At least, no error during the 6 first minutes.

Should one keep this 3100 Mhz as standard for GPUGRID or can I increase it step by step up to ???

Is this method extending the time it takes to complete the WU's ?

All WU's crashing / errors =>

Collatz (GPU use = 99 % thanks to a .config ad hoc file) until PrimeGrid recovers.

Hope to be able to run GPUGRID one day without any problem, as I bought these cards having in mind to increase my participation in this BIO project.

Thank You

Best,

Philippe

I see this card runs ~80C.

No. This card (GPU1) runs at 65-70°C.
The other card (GPU0) - which is a standard NVidia design - runs fine on 3.5GHz at 80°C.

Do you know Voltage control temps on card?

I don't know voltage control temps, but I think it should be lower than the other card's, as this card has more phase on that VR, and this card has better cooling.

Do you know temps for DDR memory? These temps run really hot on certain GTX780ti's

I don't know that either, but the same reasoning applies to the RAM chips as for the VRM chips.

In this review of your card (right one?), they measured temps using thermal imaging and found the VRM is running quite hot (87C @ load)

http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_780_ti_windforce_3x_review,9.html

Hello !

Thank you for your message.

These Gigabyte 780Ti are in an open case, with 3 extra fans helping cooling.

Or supposed to help cooling.

They don't exceed (GPU) 65° C, but no idea about the VRM temp ...

Have decreased the MEM CLOCK, unsuccessfully :/


On the other hand, they run just fine on Collatz, with an extra .config file
that utilizes the card at 99 %.

verbose=1
items_per_kernel=22
kernels_per_reduction=9
threads=9
sleep=1

They also run OK on PPS Sieve (PrimeGrid) ...

Will probably build a 100 % WC crunchbox, but not with the 780Ti, but

will wait until the 980 are accepted by GPUGRID.

In the meantime, any idea what I can do in order to be able to crunch on GPUGRID ?

I can use EVGA Precision X in order to decrease power or temp ...

Thank You

Philippe

Interesting, I remember having to cool the back of a GPU to keep it stable (might have been a ref GTX660 or 650TiBoost). I just used a case fan angled up at the bottom of the card.

Not sure about the VRM but the memory (H5GQ2H24AFR-R2C) is only rated to 70℃,
http://component.iiic.cc/index.php?main_page=product_info&products_id=1198893

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

The GPU temp is monitored using EVGA Precision X, and the fan speed (in %) = Temp + 10 => the 3 "Windforce" fans are running already very fast + the 3 external fans are helping with heat dispersal ...

Do you think the origin of the problem could be heat ?

NB : http://www.gpugrid.net/forum_thread.php?id=2507&nowrap=true#21073


This WU crashed after 1 hour only, GPU Temp about 62 ° ... :
http://www.gpugrid.net/result.php?resultid=13143650