I did let the betas run thinking it would give more points for being beta... Oh well!
The new/beta work units are just 2 work units run in serial. The first goes to 50% and then runs the second wu to 100%.
Printable View
I did let the betas run thinking it would give more points for being beta... Oh well!
The new/beta work units are just 2 work units run in serial. The first goes to 50% and then runs the second wu to 100%.
dangaroos is right. I indeed have a hidden ProgramData file. Now everything works perfectly. Thanks all for your replies and patience, and for those that submitted the xml files.:up:Quote:
Originally Posted by Movieman
My HD 7970 boards now crunch happily 8 gpu threads in parallel plus 4 CPU threads.:)
I also successfully crunched GPU beta WU's.
Oh no, wait..ahh, we need to change the file...otherwise I can't catch up to you! :DQuote:
Originally Posted by EtaCarinae
Glad to hear it's working! :up:
Now it is time to do the same with the Nvidia Boards. let's see how many threads they can handle.Quote:
Originally Posted by Movieman
In the past I needed about 20 machines fully overclocked to do a little over 1 million ppd.
Let's see how many we need for going over this mark.
nVidia cards don't respond as well as the AMD's do but they beat the crap out of cpu's.Quote:
Originally Posted by EtaCarinae
Here I was trying to score some 15 core IB xeons and all I want now are a few measly 7970's..:rofl:
Off to bed, after 2am here..Glad you got it all working.
Indeed. The worst is the GTX680. A real laggard. Running two in parallel brings nothing as elapsed time is double.Quote:
Originally Posted by Movieman
The GTX 580 is already faster per thread and you have a gain running two in parallel, this gain is about 150%. Running more brings nothing.
The Ati HD7090 is a real winner. It is already at least twice as fast per thread as a GTX 580, and you have a full gain up to 4 threads is parallel again measuring elapsed time. With 8 threads you have no more gain any more and everything on top brings nothing. This is without any OC.
Maybe you have similar results.
I have found a small benefit from running 2 parallel on NVIDIA .... the trick is to not have both start processing at the same time ... if they are offset by 50% runtime you can get an overall improvement of 5-10%
Did anyone save the executables for the beta? If they only change the name of the file when going to production I think was can just drop the new files in, update app_info and restart.
Code:<app_info>
<app>
<name>hcc1</name>
<user_friendly_name>XtremeSystems Crunches Cancer</user_friendly_name>
</app>
<file_info>
<name>wcg_hcc1_img_7.05_windows_intelx86__ati_hcc1</name>
<executable/>
</file_info>
<file_info>
<name>hcckernel.cl.7.05</name>
<executable/>
</file_info>
<app_version>
<app_name>hcc1</app_name>
<version_num>656</version_num>
<platform>windows_intelx86</platform>
<plan_class>ati_hcc1</plan_class>
<avg_ncpus>0.500000</avg_ncpus>
<max_ncpus>0.500000</max_ncpus>
<flops>30000000000.000000</flops>
<coproc>
<type>ATI</type>
<count>0.1250000</count>
</coproc>
<file_ref>
<file_name>wcg_hcc1_img_7.05_windows_intelx86__ati_hcc1</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>hcckernel.cl.7.05</file_name>
<open_name>hcckernel.cl</open_name>
</file_ref>
</app_version>
</app_info>
In my Results Status page for a Beta I let run on my Home PC GPU:
0.03 CPU / 0.04 Elapsed
58.5 Claimed / 58.5 Granted
Also here is the Result Log:
Code:<core_client_version>7.0.38</core_client_version>
<![CDATA[
<stderr_txt>
Commandline: projects/www.worldcommunitygrid.org/wcg_beta3_img_7.05_windows_intelx86__ati_hcc1
--zipfile X0930075140273200609061426.zip --imagelist images.txt --device 0
<app_init_data>
<major_version>7</major_version>
<minor_version>0</minor_version>
<release>38</release>
<app_version>705</app_version>
<app_name>beta3</app_name>
<project_preferences>
<color_scheme>Tahiti Sunset</color_scheme>
<max_frames_sec>2</max_frames_sec>
<max_gfx_cpu_pct>1.0</max_gfx_cpu_pct>
</project_preferences>
<project_dir>F:\BOINC/projects/www.worldcommunitygrid.org</project_dir>
<boinc_dir>F:\BOINC</boinc_dir>
<wu_name>BETA_X0930075140273200609061426</wu_name>
<result_name>BETA_X0930075140273200609061426_0</result_name>
<comm_obj_name>boinc_0</comm_obj_name>
<slot>0</slot>
<wu_cpu_time>0.000000</wu_cpu_time>
<starting_elapsed_time>0.000000</starting_elapsed_time>
<using_sandbox>0</using_sandbox>
<user_total_credit>5541894.954475</user_total_credit>
<user_expavg_credit>33483.922655</user_expavg_credit>
<host_total_credit>2065.246548</host_total_credit>
<host_expavg_credit>203.961944</host_expavg_credit>
<resource_share_fraction>1.000000</resource_share_fraction>
<checkpoint_period>60.000000</checkpoint_period>
<fraction_done_start>0.000000</fraction_done_start>
<fraction_done_end>1.000000</fraction_done_end>
<gpu_type>ATI</gpu_type>
<gpu_device_num>0</gpu_device_num>
<gpu_opencl_dev_index>0</gpu_opencl_dev_index>
<ncpus>1.000000</ncpus>
<rsc_fpops_est>26433014799882.000000</rsc_fpops_est>
<rsc_fpops_bound>528660295997640.000000</rsc_fpops_bound>
<rsc_memory_bound>78643200.000000</rsc_memory_bound>
<rsc_disk_bound>50000000.000000</rsc_disk_bound>
<computation_deadline>1352544838.000000</computation_deadline>
<vbox_window>0</vbox_window>
</app_init_data>
INFO: gpu_type set in init_data.xml to ATI
INFO: gpu_device_num set in init_data.xml to 0
Boinc requested ATI gpu device number0
Unzipping input
images ../../projects/www.worldcommunitygrid.org/BETA_X0930075140273200609061426_X0930075140273200609061426.zip
Processing jobdescription
Number of Images defined in image list is 2
Found compute platform Advanced Micro Devices, Inc.
Selecting this platform
CL_DEVICE_NAME: Tahiti
CL_DEVICE_VENDOR: Advanced Micro Devices, Inc.
CL_DEVICE_VERSION: 1084.2 (VM)
CL_DEVICE_MAX_COMPUTE_UNITS:
CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS: 3
CL_DEVICE_MAX_WORK_ITEM_SIZES: 256 / 256 / 256
CL_DEVICE_MAX_WORK_GROUP_SIZE: 256
CL_DEVICE_MAX_CLOCK_FREQUENCY: 1000 MHz
CL_DEVICE_ADDRESS_BITS: 32
CL_DEVICE_MAX_MEM_ALLOC_SIZE: 512 MByte
CL_DEVICE_GLOBAL_MEM_SIZE: 2048 MByte
CL_DEVICE_ERROR_CORRECTION_SUPPORT: no
CL_DEVICE_LOCAL_MEM_TYPE: local
CL_DEVICE_LOCAL_MEM_SIZE: 32 KByte
CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KByte
CL_DEVICE_QUEUE_PROPERTIES: CL_QUEUE_PROFILING_ENABLE
CL_DEVICE_EXTENSIONS:
cl_khr_fp64
cl_amd_fp64
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_int64_base_atomics
cl_khr_int64_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_khr_dx9_media_sharing
cl_amd_c1x_atomics
Estimated kernel execution time = 0.21818 [sec]
Starting analysis of X0930075140273200609061426.jp2...
Extracting GLCM features...
Total kernel time: 22.441486 (1026 kernel executions)
Total memory transfer time: 0.832911
Average kernel time: 0.021873
Min kernel time: 0.020121 (dx=25 dy=2 sample_dist=24 )
Max kernel time: 0.024428 dx=14 dy=6 sample_dist=14
INFO: GPU calculations complete.
Total time for X0930075140273200609061426.jp2: 71 seconds
Finished Image #0, pctComplete = 0.500000
Starting analysis of X0930075140274200609061426.jp2...
Extracting GLCM features...
Total kernel time: 23.764744 (1026 kernel executions)
Total memory transfer time: 1.665711
Average kernel time: 0.023163
Min kernel time: 0.021740 (dx=5 dy=25 sample_dist=24 )
Max kernel time: 0.024648 dx=4 dy=3 sample_dist=4
INFO: GPU calculations complete.
Total time for X0930075140274200609061426.jp2: 73 seconds
Finished Image #1, pctComplete = 1.000000
CPU time used = 95.332211
19:21:16 (5980): called boinc_finish
</stderr_txt>
]]>
Beta info per Uplinger on WCG Forum:
Quote:
Greetings,
We will be doing a new Beta test later today for multiple images in a single work unit. This will be a two part test.
1. Test a regular batch of 1536 work units in their current state ( 1 image ). This will test the new version of the code still works and the new assimilators/validators work.
2. Test 5 batches of 768 work units. So 2 images per work unit. 3840 total work units.
These two tests will make sure that we can just change out the science app without having to stop the project.
This version will be 7.05.
Thanks,
-Uplinger
now it runs fine a while
then
2108 World Community Grid 9.11.2012 20:58:55 [error] No app version found for app faah platform windows_intelx86 ver 640 class ; discarding faah36463_ZINC24782239_xh2_xtal_02_0
The beta is now complete. https://secure.worldcommunitygrid.or...d_thread,34225Quote:
Originally Posted by bluestang
I now have both CPU and GPU 7.05 wus running on my only none app info computer.Quote:
Greetings,
We have updated the science application for Help Conquer Cancer. This application change will allow work units to process multiple images. The purpose of this change is to increase the average runtime of the work units and decrease the server scheduler requests.
This will be a two part change.
1. The application will be released and process single image work units.
2. In about 3 days we will start sending multiple image work units.
The application version being loaded is 7.05.
Thanks,
-Uplinger
Followed the post from branjo on the WCG board
My cache was dry so I just stopped boinc, renamed app_info (to xapp_info so it would not be used) did the edit, started boinc so it would download 7.05 wus, selected no new tasks, aborted CPU wus, stopped boinc, renamed back to app_info and restarted boinc.Quote:
1. copy app info file outside the current location (e.g. to the Desktop)
2. change all 6.56 to 7.05 and 656 to 705 in it
3. detach all 6.56 WU's
4. detach and re-attach WCG
4.a. detach all new WU's
5. close BOINC
6. copy updated app info file to its folder
7. start BOIN
And it worked
First post updated to include 7.05's.
I had to go back to no app info to get things started but at that time I had already had failed units as this was before I changed the app_info file
Old Chap,
Can you confirm that we are to use the full revised ATI GPU app_info.xml post? It seems to have bothh 6.56 (top) and 7.05 (bottom file numbers in it.
Thanks,
Jean-Guiy
I left it like that for today in case anyone has the older WU's (unlikely but still) I would say everybody can delete the first part (6.56) tomorrow but that is just my view.
Old Chap,
Thank you for the quick answer. Leaving in the 6.56 for a while seems like a good plan just in case 6.56 ATI hcc priority WUs reappear in the days to follow to complete the transition to 7.05. I will do the same as you. I have about 20 mins of 6.56 left to be processed. I have selected "no new tasks" and I am waiting for their completion. Once that is done I will go through the update procedures to switch to your revised app_ino.xml file for ATI 7.05 HCC.
Thanks again for quick reply,
Jean-Guy
I thought by just dropping the app_info file for my 2 computers that run both CPU and GPU (there were no GPU wus in cache) that it would continue running the 6.56 CPU and pickup 7.05 GPU but instead it discarded all 6.56 and only downloaded 7.05.
Someone pointed out that each of these new wu are claiming over 200 pts each.
Dang. Got home, no units for a while :( updating app-info...
soo, i have trouble with the app_info thing again. it tells me "resend lost task X....." and just after that "[error] no app version found for task X.... ver7.05" when i delete the app_info it works dls a lot of wu but does them only single threaded. once i close boinc, it deletes all wu and tries to start new with said messages.
Dont know what i am doing wrong....
Old Chap,
I have switched to your new combined 656/705 ATI app_inf file for GPUs only. Got quite a few unwanted non HCC WUs from other projects upon first restart without app_info besides HCC 7.05 WUs and a few ATI GPU 7.05 WUs. Unwanted WUs likely due to me ticking the "Send me other WUs if the one I have selected are not available. I have now removed that option form my WCG profile.
I aborted all non HCC WUs. I am waiting for 6 regular HCC WUs to complete before switching over to your revised App_Info.xml file. ATI GPUs WUs seem to complete in approximately the same time as their 6.56 predecessors.
Jean-Guy
They changed the app version for HCC. I tried to change that in the app_info file, but couldn't get it to work. I will try again tomorrow. Back to one WU now.
Here's how I got it to work.Quote:
Originally Posted by Rob_B
1. Let all of your version 6.56 finish. If they already have then do the following.
2. When the 656 task are done or you're already done with them shut down boinc.
3. Edit the app_info file by replacing 705 everywhere you see 656. Be careful of where to use and not use decimal points. Do all this without removing the file from the project folder.
4.Save the changes.
5. Rename the file app_info.txt
6.Restart boinc and let it download some new tasks. Let it finish one but suspend the rest then shut down boinc again
7. Go back and change the app_info file extension back to .xml
8. Restart boinc and resume 1 task at a time to make sure you're app_info file has the correct syntax. If they resume ok then resume them all. Let boinc download new tasks on its' own. No need to pound on the update button. Done
Got the app_info to work here with the new 7.05 WU's, running 8 on my 7970 now:up:
Yup mine is fixed now too. Didn't take too much to fix it but still a little depressing to come hope to an empty queue.
Hello Folks,
For those of you who decide to update to the new ATI GPU hcc 7.05 file format using Bun-Bun's revised ATI GPU only app_info file posted on the first page of this thread.
Remember that it contains both the 6.56 and 7.05 app_info file details. When you set the decimal count for number of WUs you want you per GPU, remember that you will have to do so in both the 6.56 and 7.05 count setting lines of the file.
I am now using the subject revised app_info file and it is working great crunching 8 ATI hcc 7.05 WUs concurrently with no problems.
Thanks for the good work Bun-Bun...!!!
FYI,
Jean-Guy
Two days in a row now...:shakes:Quote:
Originally Posted by Sparky
I had problems. I'm not sure what I was doing wrong but, this worked for me.Quote:
Originally Posted by nanoprobe