Posts by dem0707
log in
1) Message boards : Number crunching : Optimizing Collatz Sieve (Message 23694)
Posted 378 days ago by dem0707
I just want to throw in my 2 cents worth on this optimization thing. Sadly, I have no experience with the new 10XX GPUs, and have not tinkered around with my setups since the new 8x sized WUs have started. But I have had good luck with the following on my 5 GTX980s:
    <app_config>
    <app>
    <name>collatz_sieve</name>
    <max_concurrent>5</max_concurrent>
    <gpu_versions>
    <gpu_usage>1.0</gpu_usage>
    <cpu_usage>0.4</cpu_usage>
    </gpu_versions>
    </app>
    </app_config>


I have always found that for Collatz, running more than 1 WU per GPU will only slow things down.

    verbose=0
    kernels_per_reduction=48
    threads=8
    lut_size=17
    sleep=0
    cache_sieve=1
    reduce_cpu=0
    sieve_size=30


And don't be afraid to go for the faster time per WU even if the credit is less. I found that the credits per second was always better. I think that with the lut_size optimized a GPU can get a WU done in less time with less work (thus less credit) by not having to recalculate some of the same things over and over. I think this is even more important with new bigger WUs.

Happy New Year to everyone, and keep on Crunching(@EVGA)!!!

David

2) Message boards : Number crunching : Badges (Message 23483)
Posted 415 days ago by dem0707
I really like the old style badges and the new ones proposed by EG. But I feel that having so many spaces 5B apart is kind of odd. Yes, I know how long it will take to jump those gaps. I kind of feel that the badge spacing should be using the 1 - 2 - 5 scheme, so that each step approximately doubles. eg. ...1M 2M 5M 10M 20M 50M 100M 200M 500M 1B 2B 5B...
3) Message boards : News : SSL Certificate Replaced (Message 23303)
Posted 436 days ago by dem0707
It looks like we are back up and running. I was finally able to add the Collatz project again and it is running.
4) Message boards : Number crunching : Is "Best Result" gone? (Message 23121)
Posted 456 days ago by dem0707
Sorry it took so long to get it fixed. I had a minor emergency (water heater flooded the basement with 3" of water). Cleanup is still underway. The table in the database that stores the stats was totally corrupted. It's back up now but it will take me a while to pull the backups day by day and recalculate the top records from the missing days.


Oh, wow, sorry about the flood. It's a good thing your water heater is not in the attic. Thanks for getting this fixed.
5) Message boards : Number crunching : Is "Best Result" gone? (Message 23112)
Posted 458 days ago by dem0707
The "Best Result" on my page is empty. Is it gone or just broken?


It does looks rather sad and empty so I'll raise a glass to it tonight and check back tomorrow.


The whole "Today's High Steps" page is empty also. I looks like we need to raise more than one glass to fix this.
6) Message boards : Number crunching : Optimizing Collatz Sieve (Message 21962)
Posted 725 days ago by dem0707
32-bit OpenCL 1.2 is faster than 64 bit OpenCL 2.0? Interesting.

To force it to use the 32 bit version, I think you would need to use an app_info.xml file so that you can refer to the specific executable you want to run.


Could you be so kind to show me an Example.


Thanks Knobi, for posting your config file earlier, I have been using it and it is working great.

On my GTX980s, I am seeing the 32-bit (i686) version running about 10 seconds per WU faster that the 64-bit (X86_64). The WUs seem to mostly run using the 64-bit, then a batch of 32-bit seemingly at random.

What would we need to use for the app_info.xml file to force using the 32-bit version?




Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.