I am on Windows 10 Insider Fast Ring Build 19002, using NVIDIA HotFix driver 436.51, on a PC that has 3 GPUs: RTX 2080, GTX 980 Ti, GTX 980.

I keep seeing my SETI GPU tasks show "Postponed" as the Task Status.
When I looked at the stderr.txt file for them, I see the following lines that look suspect:

ERROR: OpenCL kernel/call 'clEnqueueMapBuffer(gpu_GPUState)' call failed (-36) in file ..\analyzeFuncs.cpp near line 1995.
Waiting 30 sec before restart...

Does any expert have any additional information about how to resolve this?

PC Tasks:
https://setiathome.berkeley.edu/results.php?hostid=7987187

Full log:

Running on device number: 2
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
OpenCL platform detected: NVIDIA Corporation
BOINC assigns device 2
Info: BOINC provided OpenCL device ID used

Build features: SETI8	Non-graphics	OpenCL	USE_OPENCL_NV	OCL_ZERO_COPY	SIGNALS_ON_GPU	OCL_CHIRP3	FFTW	USE_SSE3	x86	
     CPUID: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz 

     Cache: L1=64K L2=256K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3 SSSE3 FMA3 SSE4.1 SSE4.2 AVX 
OpenCL-kernels filename : MultiBeam_Kernels_r3584.cl 
ar=2.728229  NumCfft=99655  NumGauss=0  NumPulse=27847483539  NumTriplet=27847483539
Currently allocated 201 MB for GPU buffers
In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768

Windows optimized setiathome_v8 application
Based on Intel, Core 2-optimized v8-nographics V5.13 by Alex Kan
SSE3xj Win32 Build 3584 , Ported by : Raistmer, JDWhale

SETI8 update by Raistmer

OpenCL version by Raistmer, r3584

Number of OpenCL platforms:				 1


 OpenCL Platform Name:					 NVIDIA CUDA
Number of devices:				 3
  Max compute units:				 46
  Max work group size:				 1024
  Max clock frequency:				 1815Mhz
  Max memory allocation:			 2147483648
  Cache type:					 Read/Write
  Cache line size:				 128
  Cache size:					 1507328
  Global memory size:				 8589934592
  Constant buffer size:				 65536
  Max number of constant args:			 9
  Local memory type:				 Scratchpad
  Local memory size:				 49152
  Queue properties:				 
    Out-of-Order:				 Yes
  Name:						 GeForce RTX 2080
  Vendor:					 NVIDIA Corporation
  Driver version:				 436.51
  Version:					 OpenCL 1.2 CUDA
  Extensions:					 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_copy_opts cl_nv_create_buffer
  Max compute units:				 22
  Max work group size:				 1024
  Max clock frequency:				 1291Mhz
  Max memory allocation:			 1610612736
  Cache type:					 Read/Write
  Cache line size:				 128
  Cache size:					 1081344
  Global memory size:				 6442450944
  Constant buffer size:				 65536
  Max number of constant args:			 9
  Local memory type:				 Scratchpad
  Local memory size:				 49152
  Queue properties:				 
    Out-of-Order:				 Yes
  Name:						 GeForce GTX 980 Ti
  Vendor:					 NVIDIA Corporation
  Driver version:				 436.51
  Version:					 OpenCL 1.2 CUDA
  Extensions:					 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_copy_opts cl_nv_create_buffer
  Max compute units:				 16
  Max work group size:				 1024
  Max clock frequency:				 1278Mhz
  Max memory allocation:			 1073741824
  Cache type:					 Read/Write
  Cache line size:				 128
  Cache size:					 786432
  Global memory size:				 4294967296
  Constant buffer size:				 65536
  Max number of constant args:			 9
  Local memory type:				 Scratchpad
  Local memory size:				 49152
  Queue properties:				 
    Out-of-Order:				 Yes
  Name:						 GeForce GTX 980
  Vendor:					 NVIDIA Corporation
  Driver version:				 436.51
  Version:					 OpenCL 1.2 CUDA
  Extensions:					 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_copy_opts cl_nv_create_buffer


Work Unit Info:
...............
Credit multiplier is :  2.85
WU true angle range is :  2.728229
Used GPU device parameters are:
	Number of compute units: 16
	Single buffer allocation size: 128MB
	Total device global memory: 4096MB
	max WG size: 1024
	local mem type: Real
	FERMI path used: yes
	LotOfMem path: yes
	LowPerformanceGPU path: no
	HighPerformanceGPU path: no
period_iterations_num=50
Triplet: peak=9.063154, time=46.4, period=0.4096, d_freq=1419162902.83, chirp=0, fft_len=32 
ERROR: OpenCL kernel/call 'clEnqueueMapBuffer(gpu_GPUState)' call failed (-36) in file ..\analyzeFuncs.cpp near line 1995.
Waiting 30 sec before restart...

I would use an older video drive. There seems to be a problems with the newest drivers. I did not do the analysis you did but I would get errors on my Ryzen when I installed the latest drivers.
(you do install the drivers from the nVidia site every time you install an new build, don't you?)

---

SETI@home classic workunits 4,019
SETI@home classic CPU time 34,348 hours

Is there any confirmation that this particular error is related to driver version? I use the latest drivers available, intentionally, for security and gaming. If there is a problem, let's identify it and fix it, instead of ignoring it, please. Thank you.

If an application which has been in constant use since 5 Jan 2017 suddenly starts throwing errors, and the only thing which has changed is a driver update, then it's probable that the component which needs debugging is the new driver.

Is there any confirmation that this particular error is related to driver version? I use the latest drivers available, intentionally, for security and gaming. If there is a problem, let's identify it and fix it, instead of ignoring it, please. Thank you.

there is confirmation of the 436 drivers causing issues with certain types of tasks. notably Arecibo VHAR tasks. even if the driver isnt the cause of this specific issue, you WILL have a problem processing certain types of tasks with these drivers. but browsing through your errored tasks, they all look like this type of tasks...

there's no "NEED" to have the latest drivers for "gaming". what games do you play?

the security need is also highly blown out of proportion, as 99.99% of normal people are not even a target, or the security threat requires physical access to your system to exploit.

but if you insist, 431.60 is a game ready driver, and includes the security fixes. I would try this driver and see if the issue goes away.

you can get the 431.60 drivers here: https://www.nvidia.com/download/driverResults.aspx/148865/en-us

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

I'm not asking to find a workaround. I'm asking for a fix! If there's a driver issue, let's work to fix it.

Am I the first to report this particular issue?

Also, are there any steps I can take to get it to repro outside of BOINC, in order to send to my NVIDIA contacts for analysis?

I'm not asking to find a workaround. I'm asking for a fix! If there's a driver issue, let's work to fix it.

Am I the first to report this particular issue?

Also, are there any steps I can take to get it to repro outside of BOINC, in order to send to my NVIDIA contacts for analysis?

unless you work for nvidia, you can't "fix" a driver issue... the fix is to revert to known working driver. which is 431.60 like I posted, until nvidia gets around to pushing out a new version that fixes it. people have already filed bug reports with nvidia about it.

you are not the first one to report this issue with 436 drivers. there's several other threads in this forum about it.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

The fix is to download 431.xx drivers and do a "clean install" which removes all traces of the previous set. NVidia hasn't even gotten as far as acknowledging the issue with 436.xx as it's such a tiny minority of their customers who are affected by this.

---

Do you have a link to another report of my particular error? I searched the forums and could not find it.

Also, I have a little sway with NVIDIA, and can get priority put towards a fix, if someone could help me get a solid repro.

Let's get the issues fixed.

unless you work for nvidia, you can't "fix" a driver issue...

But you can assess where the fault lies, and then you can address your fault to the forum used by the responsible body - in the case of a driver bug, to the manufacturer's bug tracking service (NVidia, in this case). The OP of this particular thread is experienced in doing that.

Although at least one of the SETI@Home developers has submitted the issue to NVidia already, by all means if you'd like to do so that would be appreciated, as the more reports they get the more likely they are to do something to resolve it (or at least acknowledge!)

I personally haven't seen that particular error before, but as it's been established there are known bad drivers present, they are going to be blamed. The best thing at this point would be to revert to 431.xx, even just briefly for testing, to see if the issue is resolved. I'm betting it would be.

---

Do you have a link to another report of my particular error?

The key threads are

NVidia 436.xx drivers can cause very long compute times especially on Arecibo VHAR work units (sticky)
SoG custom application stuck on 1660Ti

here: https://setiathome.berkeley.edu/forum_thread.php?id=84694
here: https://setiathome.berkeley.edu/forum_thread.php?id=84761

what games do you play?

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

Does anybody have instructions to run a SETI task outside of BOINC, in order to hopefully get the 2 lines of text in post 1 to show outside of BOINC?

Here's a link we can use to try the OpenCL SDK samples:
https://developer.nvidia.com/opencl

Side note: I've had to suspend SETI on my machine, because it actually caused a BSOD after tons of TDR failures. I'm still willing to get a task for testing purposes if need be.

Mr. Kevvy:
Yes. I'd like to isolate the issue, and help NVIDIA to get it fixed. I've done this with previous issues before.
My plan might be to test:
- NVIDIA drivers: 431.60, 431.68, 436.51
- against Windows 10 Builds: 18362, 19002
... but again, I need help getting a solid repro, preferably outside of BOINC.

Richard:
Thanks. Hell yeah, let's get this fixed. Maybe I'll have to run another set of OpenCL SDK tests, against the 3 drivers and 2 builds above, for the 3 GPUs in my rig. Gonna take a lot of caffeine.

I looked at the links you mentioned, even before posting this post, but I never saw my particular error. Perhaps people weren't digging into the stderr.txt enough to find it?

Ian&Steve C.:
I read the thread, before I posted and again now. Again, didn't see my particular error.
I play iRacing, Hearthstone, Rocket League, and occasionally a newly released title. Please stop supporting the idea that using older drivers is a fix. It's a workaround. I'm here to fix this. Are you?

There are the Multiband benchmark tools at Lunatics that you can use to run tasks outside of BOINC.
http://lunatics.kwsn.info/index.php?action=downloads;cat=45

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

[voice=ArnoldSchwarzenegger action=PicksUpBazooka] Excellent! [/voice]

Does anybody have instructions to run a SETI task outside of BOINC, in order to hopefully get the 2 lines of text in post 1 to show outside of BOINC?

download the WU files (preferably several, of varying types), and run them in the benchmarking tool. I've only done this personally on Linux, but there may be one for Windows.

look here: http://lunatics.kwsn.info/index.php?action=downloads;cat=45

Ian&Steve C.:
I read the thread, before I posted and again now. Again, didn't see my particular error.
I play iRacing, Hearthstone, Rocket League, and occasionally a newly released title. Please stop supporting the idea that using older drivers is a fix. It's a workaround. I'm here to fix this. Are you?

I'm all for a fix, but I'm being realistic here. You can't personally "fix" anything. you can only forward off some info to nvidia and hope they allocate some resources to actually pushing a fix. the immediate fix is to revert to the 431 drivers like so many others have done successfully. I looked through your errors tasks, and all of them are Arecibo VHAR tasks exceeding their time limit, and these issues are widely reported recently with 436 drivers. it's not hard to see when the square peg fits in the square hole.

and I asked about games, because if the games you play aren't listed in the driver notes, then you wont see any improvement. so many people have this misconception that they NEED the latest and greatest, and refuse to accept that they really don't.

Games called out for improvements in 436 drivers:

436.02
Game Ready
Provides increased performance and the optimal gaming experience for Apex Legends, Battlefield V, Forza Horizon 4, Strange Brigade, and World War Z

Gaming Technology
Adds Beta support for GPU Integer Scaling
Adds Beta support for Ultra-Low Latency Mode
Adds support for new Freestyle Sharpen Filter
Adds support for new G-SYNC compatible monitors

436.15
Game Ready
Provides increased performance and the optimal gaming experience for Control 

436.30
Game Ready
Provides increased performance and the optimal gaming experience for the Call of Duty: Modern Warfare PC Open Beta, Gears 5, and Borderlands 3

Gaming Technology
Adds support for six new G-SYNC compatible monitors 

436.48
Game Ready
Provides increased performance and the optimal gaming experience for Tom Clancy’s Ghost Recon Breakpoint and Asgard’s Wrath

Gaming Technology
Adds support for the HTC VIVE Cosmos VR HMD

you mentioned security, so I pointed out that 431.60 includes the security fixes. 431.60 checks all the boxes for you. wont cause issues with SETI/OpenCL, works with the games you play, and includes the security fixes. there's no real need in your case for 436 other than a false assumption that they are better.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

in addition to the issues with using the wrong driver, you should be open to the possibility that you have one or more additional issues with the system. You're also running 436.51, which doesnt seem to be available to the general public, it's not listed on nvidia's site. is this a pre-release beta driver or something?

one thing that I noticed while looking through your errored tasks, was that your system seems to have a problem with devices 1 and 2, the GTX 980ti and 980 respectively. Every time a job tries to run on these cards, you get the error in question, and it keeps retrying until GPU0 (the RTX 2080) becomes availabe (but then times out because of the well known issues with this driver). Either a hardware problem with the card itself, or a problem with the installation of the OpenCL components of the drivers when you upgraded.

Problem #1 looks to be a bad driver install, maybe you didnt do a DDU + clean install. maybe bad hardware.
Problem #2 looks to be the previous issues with the 436 drivers in general

but again, this is only happening on Arecibo VHAR tasks, and it looks like its processing BLC tasks OK on your 980(ti) cards (i see the same behavior on your 1050ti system also), so my guess is it's the driver.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

Oh I see now that 436.51 is a "hotfix" driver version, not a standard release.

This Hotfix driver addresses the following:

FIFA 19/FIFA 20/Star Wars: Battlefront II (2017) – Game may randomly crash to desktop
Apex Legends: Random flicker during gameplay

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

If an application which has been in constant use since 5 Jan 2017 suddenly starts throwing errors, and the only thing which has changed is a driver update, then it's probable that the component which needs debugging is the new driver.

Actually, there are a few other SETI Apps that have problems with only the Arecibo VHARs.
1) The CUDA Special App will produce the wrong best Pulse about 50% of the time with Arecibo VHARs
2) The Linux CPU App 3711 will fail to track the progress with Arecibo VHARs
3) nVidia OpenCL builds for MacOS GPUs haven't worked correctly since the release of El Capitan, since then the Macs are using the Intel OpenCL build on nVidia GPUs
4) The MacOS CPU App 3605 will fail with a SIGPIPE Error on a few of the recent Arecibo VHARs;

Crashed executable name: MBv8_8.22r3605_avx2_x86_64-apple-darwin
Machine type Intel x86-64h Haswell (64-bit executable)
System version: Macintosh OS 10.14.6 build 18G95
Thu Oct 17 10:28:51 2019
WU true angle range is :  2.732913

atos cannot load symbols for the file MBv8_8.22r3605_avx2_x86_64-apple-darwin for architecture x86_64.
0   MBv8_8.22r3605_avx2_x86_64-apple-darwin 0x00000001079bb0e4  
1   MBv8_8.22r3605_avx2_x86_64-apple-darwin 0x00000001079a2a34  
SIGPIPE: write on a pipe with no reader
2   libsystem_pthread.dylib             0x00007fff58cb42eb  
SIGPIPE: write on a pipe with no reader
SIGPIPE: write on a pipe with no reader
3   libsystem_pthread.dylib             0x00007fff58cb7249  
4   libsystem_pthread.dylib             0x00007fff58cb340d  
SIGPIPE: write on a pipe with no reader

The Apps work fine on Non Arecibo VHARs.
With all these Window's stalled tasks and BSOD I suppose the other OSes are lucky.