And a heads up for those running Boinc alpha clients. We've had reports that the ATI AP app dies when you suspend it on Boinc 7.0.22/.23.

Not just AP, but MB on ATI as well. Had that happen a couple of times. Exit code 1, latest crash happened this morning when BOINC suspended itself for time-of-day.

Good news, this is fixed in BOINC 7.0.24

I've tested with an MB and an AP, both can suspend without trouble now.
Even exiting BOINC will no longer crash the task.

With thanks to David for that speedy fix.

---

Question
There are two new files in my project folder


AstroPulse_Kernels_r555.cl_GeForce GTX 460.bin_V6

r555_AMDPhenomtmIIX61100TProcessor.wisdom


What are these???

The OpenCL compiled binary for your GTX 460, for AP v6 Wu's

and

the fftw Wisdom file for your CPU, contains codelets suitable for your CPU, this used to be generated in the slot every time the Stock CPU AP, MB or AP r409 apps started initially, and deleted once the Wu finished,
now it's just generated once in the project directory, and is reused.

Claggy

Thanks Claggy

Just curious

Michael Miles

Crunching stability reached, ah sweet stability, 285.62 is good...

Oh and I owe It all to some experimentation and some reading of some comments in the lunatics own forum, 98/99 on both 590 gpus all night while I slept, no increase in priority needed and no cpu fall back either...

I went thru several revisions of BoincTasks, all the way back to 1.17, currently at 1.24 and holding, of course TThrottle had to be reduced to 5.10 from 5.50 and x41g had to used, but now I have stability, instability would show up after 2 to 3 hours, no I don't need JasonG, I have what I need, He can ignore Me until Hell freezes over for all I care.

I am still having a driver issue though, stock speed is 630MHz for this 590, it's now running at 553MHz, so Nvidia has a driver problem it seems.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Have you tried Nvidia system tools current ver is AFAIK 6.06 on Nvidia site
of NvidiaInspector? Either will allow user to set clocks etc.

I used the latter to downclock my 560's for a bit when the temps got to high:-)

Also I've had some problems with that driver under Win7, every now and then it just quits for no apparent reason, you get a blank screen, driver reloads and Boinc restarts apps, but with warning about no output files found and if problem persists project reset might be required..

But otherwise no problems:-)

Cheers,

---

Cliff,
Been there, Done that, Still no damm T shirt!

Have you tried Nvidia system tools current ver is AFAIK 6.06 on Nvidia site
of NvidiaInspector? Either will allow user to set clocks etc.

I used the latter to downclock my 560's for a bit when the temps got to high:-)

Also I've had some problems with that driver under Win7, every now and then it just quits for no apparent reason, you get a blank screen, driver reloads and Boinc restarts apps, but with warning about no output files found and if problem persists project reset might be required..

But otherwise no problems:-)

Cheers,

The no output file problem is usually because the rig is so tied up tight that it cannot write to the hard drive in a reasonable time...

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

The no output file problem is usually because the rig is so tied up tight that it cannot write to the hard drive in a reasonable time...

No.

No output file is a symptom that the task failed - could be any any number of reasons why. You need to dig deeper into the task report on this website to find out what happened to it.

Hi Richard,
Trouble is I cant remember which tasks they were:-/ And AFAIK they actually resumed, I checked tasks and didnt see any error'd out tasks..

But it was immediatly after the 285.62 driver dropped out and then restarted.
So I assume when the driver went toes up, it stopped any GPU tasks dead without
the output files.. But that shouldnt have stopped the manager or client altogether I'd have thought..

Regards,

---

Cliff,
Been there, Done that, Still no damm T shirt!

Just asking. It's normal that the AP6_win_x86_SSE2_OpenCL_ATI_r555.exe app has no ATI GPU load for hours upon end? I've been tracking it for the past 1.5 hour+ with GPU-Z and SIV (yay, Ray!) and both show no GPU load.

It's progressing through the AP v6, but only with (seemingly high) CPU cycles: average 18% (with a max of 25%, since it's a 4 core CPU). So what then, has it moved back to running on the CPU only, by chance?

stderr.txt so far (72% run):

Number of app instances per device setted to:1
DATA_CHUNK_UNROLL setted to:4
FFA thread block override value:2048
FFA thread fetchblock override value:1024
Maximum single buffer size setted to:128MB
Running on device number: 0
DATA_CHUNK_UNROLL at default:4
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
OpenCL platform detected: Advanced Micro Devices, Inc.
BOINC assigns 0 device, slots 0 to 0 (including) will be checked
Used slot is 0; Used GPU device parameters are:
Number of compute units: 12
Single buffer allocation size: 128MB
max WG size: 256

Build features: Non-graphics OpenCL COMBINED_DECHIRP_KERNEL FFTW USE_INCREASED_PRECISION USE_SSE2 x86
CPUID: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz

Cache: L1=64K L2=256K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
AstroPulse v. 6.00
Non-graphics FFTW USE_CONVERSION_OPT
Windows x86 rev 555, 6.00 match, by Raistmer with support of Lunatics.kwsn.net team. SSE2

OpenCL version by Raistmer

oclFFT fix for ATI GPUs by Urs Echternacht
ffa threshold mods by Joe Segur
SSE3 dechirping by JDWhale
Combined dechirp kernel by Frizz
Number of OpenCL platforms: 1


OpenCL Platform Name: AMD Accelerated Parallel Processing
Number of devices: 1
Max compute units: 12
Max work group size: 256
Max clock frequency: 775Mhz
Max memory allocation: 536870912
Cache type: None
Cache line size: 0
Cache size: 0
Global memory size: 2147483648
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Scratchpad
Local memory size: 32768
Queue properties:
Out-of-Order: No
Name: Barts
Vendor: Advanced Micro Devices, Inc.
Driver version: CAL 1.4.1703 (VM)
Version: OpenCL 1.1 AMD-APP (898.1)
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_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_popcnt cl_khr_d3d10_sharing cl_amd_meminfo


Info : Building Program (binary, clBuildProgram):main kernels: OK code 0

Number of app instances per device setted to:1
DATA_CHUNK_UNROLL setted to:4
FFA thread block override value:2048
FFA thread fetchblock override value:1024
Maximum single buffer size setted to:128MB
Running on device number: 0
DATA_CHUNK_UNROLL at default:4
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
OpenCL platform detected: Advanced Micro Devices, Inc.
BOINC assigns 0 device, slots 0 to 0 (including) will be checked
Used slot is 0; Used GPU device parameters are:
Number of compute units: 12
Single buffer allocation size: 128MB
max WG size: 256

Build features: Non-graphics OpenCL COMBINED_DECHIRP_KERNEL FFTW USE_INCREASED_PRECISION USE_SSE2 x86
CPUID: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz

Cache: L1=64K L2=256K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
### Restart at 11.71 percent.
Info : Building Program (binary, clBuildProgram):main kernels: OK code 0

Number of app instances per device setted to:1
DATA_CHUNK_UNROLL setted to:4
FFA thread block override value:2048
FFA thread fetchblock override value:1024
Maximum single buffer size setted to:128MB
Running on device number: 0
DATA_CHUNK_UNROLL at default:4
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
OpenCL platform detected: Advanced Micro Devices, Inc.
BOINC assigns 0 device, slots 0 to 0 (including) will be checked
Used slot is 0; Used GPU device parameters are:
Number of compute units: 12
Single buffer allocation size: 128MB
max WG size: 256

Build features: Non-graphics OpenCL COMBINED_DECHIRP_KERNEL FFTW USE_INCREASED_PRECISION USE_SSE2 x86
CPUID: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz

Cache: L1=64K L2=256K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
### Restart at 56.76 percent.
Info : Building Program (binary, clBuildProgram):main kernels: OK code 0

---

That happens sometimes sadly.
Driver issue.
High CPU load indicates high blanking.
Blanking is processed by CPU only.

Suspend GPU for a few minutes and resume again.

Should solve it.

---

With each crime and every kindness we birth our future.

Just asking. It's normal that the AP6_win_x86_SSE2_OpenCL_ATI_r555.exe app has no ATI GPU load for hours upon end? I've been tracking it for the past 1.5 hour+ with GPU-Z and SIV (yay, Ray!) and both show no GPU load.

It's progressing through the AP v6, but only with (seemingly high) CPU cycles: average 18% (with a max of 25%, since it's a 4 core CPU). So what then, has it moved back to running on the CPU only, by chance?

If it's got higher CPU usage it's probably got more Blanking (Blanking is done on the CPU and causes lower GPU usage),

But there is also a ATI driver Bug/app Bug about, it affects both Raistmer's MB/v7 and AP GPU apps, Suspending the task with low GPU usage, watching task manager until it exits, then resuming the task sometimes gets higher GPU use,

Claggy

Suspending the task with low GPU usage, watching task manager until it exits, then resuming the task sometimes gets higher GPU use

I read someone saying that before and tried it. Didn't work. The task is now at 90% and still only working on the CPU. Oh well, at least it still goes reasonably fast.

---

OK, so in the end it was the amount of blanking.
End of the task says:

    single pulses: 2
repetitive pulses: 0
  percent blanked: 84.90
class T_remove_radar:	total=1.26e+009,	N=1,	<>=1.26e+009,	min=1.26e+009,	max=1.26e+009
class T_main_loop_L1:	total=3.81e+013,	N=47,	<>=8.10e+011,	min=3.31e+011,	max=2.58e+012
 class T_FFT_forward:	total=9.06e+009,	N=192512,	<>=4.71e+004,	min=5.42e+003,	max=1.67e+008
 class T_remove_radar_randomize:	total=5.19e+012,	N=769672,	<>=6.75e+006,	min=3.06e+002,	max=1.45e+009
 class T_build_chirp_table:	total=0.00e+000,	N=0,	<>=0.00e+000,	min=1.84e+019,	max=0.00e+000
 class T_DataWrite:	total=3.43e+009,	N=164688,	<>=2.08e+004,	min=4.26e+003,	max=1.69e+007
  class T_DataWrite_ns:	total=0,	N=0,	<>=0,	min=0	max=0
 class T_oclReadBuf:	total=2.71e+007,	N=192512,	<>=1.40e+002,	min=1.80e+001,	max=5.50e+006
   class T_ChirpWrite:	total=0.00e+000,	N=0,	<>=0.00e+000,	min=1.84e+019,	max=0.00e+000
    class T_ChirpWrite_ns:	total=0,	N=0,	<>=0,	min=0	max=0
 class T_dechirp:	total=3.75e+009,	N=192512,	<>=1.95e+004,	min=8.01e+003,	max=1.47e+008
  class Dechirp_ns:	total=0,	N=0,	<>=0,	min=0	max=0
  class Half_ns:	total=0,	N=0,	<>=0,	min=0	max=0
 class T_PC_single_pulse_kernel_FFA_update:	total=1.02e+013,	N=192512,	<>=5.28e+007,	min=1.01e+007,	max=1.34e+009
  class PC_ns:	total=0,	N=0,	<>=0,	min=0	max=0
class T_oclReadBuf:	total=2.71e+007,	N=192512,	<>=1.40e+002,	min=1.80e+001,	max=5.50e+006
class T_oclWriteBuf:	total=3.50e+009,	N=164688,	<>=2.13e+004,	min=4.39e+003,	max=1.69e+007
  class T_FFT_inverse:	total=1.64e+009,	N=192512,	<>=8.51e+003,	min=4.75e+003,	max=1.47e+008
 class T_ffa:	total=2.25e+013,	N=846,	<>=2.66e+010,	min=4.23e+009,	max=3.80e+011
class T_GPU_buffer_read_backs:	total=1,	N=1,	<>=1,	min=1	max=1
USE_OPENCL	OPENCL_WRITE	USE_INCREASED_PRECISION	SMALL_CHIRP_TABLE	COMBINED_DECHIRP_KERNEL	
rev 555
22:55:08 (2772): called boinc_finish

The MB running on the GPU now is taking 60-70% GPU, so normal load. :)

---

And the CPU usage is normal as well.

---

With each crime and every kindness we birth our future.

I have noticed huge blanking in almost all AP wu

What does cause the blanking
Radio interference? Sun giving out huge blasts of solar radiation?

I am not sure what is blanking but in AP tasks that I have been doing with NV r555 GPU the normal run time is usually under 2 hours. All the ones I have don with blanking high have taken 3 hours to do


They all run high priority under 7.0.24 x64


Michael Miles

Have you tried Nvidia system tools current ver is AFAIK 6.06 on Nvidia site
of NvidiaInspector? Either will allow user to set clocks etc.

I used the latter to downclock my 560's for a bit when the temps got to high:-)

Also I've had some problems with that driver under Win7, every now and then it just quits for no apparent reason, you get a blank screen, driver reloads and Boinc restarts apps, but with warning about no output files found and if problem persists project reset might be required..

But otherwise no problems:-)

Cheers,

Well then what do You suggest Cliff? Or somebody? As I just went thru a 4 way cpu backoff, the gpus were not even being used, It made My computer almost unusable, My browser had to wait until I killed BoincTasks and Boinc before I could use it...

Or should I just quit while I'm ahead...

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Have you tried Nvidia system tools current ver is AFAIK 6.06 on Nvidia site
of NvidiaInspector? Either will allow user to set clocks etc.

I used the latter to downclock my 560's for a bit when the temps got to high:-)

Also I've had some problems with that driver under Win7, every now and then it just quits for no apparent reason, you get a blank screen, driver reloads and Boinc restarts apps, but with warning about no output files found and if problem persists project reset might be required..

But otherwise no problems:-)

Cheers,

Nvidia System Tools does not support the 590 and they updated the tools to 6.08...

I tried 285.79, It's as bad as 285.62, crap...

http://www.nvidia.com/object/nvidia-system-tools-6.08-driver.html

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Oh this You'll love the PCI-E bus was being throttled, I had to dig thru Nvidia Inspectors 2nd panel, but once I set that control to off, the gpu went from 553MHz to 630MHz! Instantly in fact, go figure.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

I have noticed huge blanking in almost all AP wu

What does cause the blanking
Radio interference? Sun giving out huge blasts of solar radiation?

Military radar usually. It has been somewhat well-documented in the tech news posts over the past 2-3 years. The staff knew there was radar there and there wasn't anything that could be done about it for a while. Then a very smart/clever person figured out that it uses a very predictable pattern, so the data can be marked where the radar is interfering and thus.. we have our blanking.

There may be other sources of interference, but I'm pretty sure radar is the major source.

And yes, I have also had a lot of high-blanked APs. Crunched and returned 31 APs as of right now and according to my spreadsheet, I finally have the 10 needed for the good ETA estimates. Problem is that I just have to wait for wingmen to crunch their half and report.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

I have noticed huge blanking in almost all AP wu

What does cause the blanking
Radio interference? Sun giving out huge blasts of solar radiation?

Military radar usually. It has been somewhat well-documented in the tech news posts over the past 2-3 years. The staff knew there was radar there and there wasn't anything that could be done about it for a while. Then a very smart/clever person figured out that it uses a very predictable pattern, so the data can be marked where the radar is interfering and thus.. we have our blanking.

There may be other sources of interference, but I'm pretty sure radar is the major source.

And yes, I have also had a lot of high-blanked APs. Crunched and returned 31 APs as of right now and according to my spreadsheet, I finally have the 10 needed for the good ETA estimates. Problem is that I just have to wait for wingmen to crunch their half and report.

Out here radar isn't a problem, bombs shaking the ground for miles is(an IE).

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's