- Joined
- Aug 14, 2014
Had the issue since v2.43 (not always though), if the image doesn't starts loading in the 1st 10s or so restart the bench.
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
FEATURED Marathon Season V November: Realbench
- Thread starter Johan45
- Start date
Had the issue since v2.43 (not always though), if the image doesn't starts loading in the 1st 10s or so restart the bench.
- Joined
- Apr 3, 2015
- Location
- Planet Earth
So what happened to the Champion o' da People, and my favorite chicken farmer, Lochekey!!??
He started this whole mess anyway
Did you ever figure this out?
What does anyone else's OpenCL test look like?
Are the CPU's selected?? Mine aren't, and the caption at the bottom says CPU and GPUs.
I get the same score every time, individually, or entire test. Regardless of times run?
Luego
He started this whole mess anyway
Did you ever figure this out?
What does anyone else's OpenCL test look like?
Are the CPU's selected?? Mine aren't, and the caption at the bottom says CPU and GPUs.
I get the same score every time, individually, or entire test. Regardless of times run?
Luego
You can see everyone's Open CL scores in their posts/screenshots. 
Not sure what he found, but what I saw in the Open CL test is that the score doesn't change 1:1 with the K/Samples second (K/S /s moving forward) value. HOwever the K/S /s value DOES change run to run. What that tells us is the score its popping is associated with a range of K/S /s values. For example, you get 1000 points for getting 50-60 K/S /s and maybe 2000 points for getting 61-70 K/S /s... etc. So that if you run 52 one time and 58 the other, that would be why the actual score doesn't change. But if you pay attention to the actual value of K/S /s, it varies (a little) run to run. You simply haven't passed the top of your score range yet I think.
Not sure what he found, but what I saw in the Open CL test is that the score doesn't change 1:1 with the K/Samples second (K/S /s moving forward) value. HOwever the K/S /s value DOES change run to run. What that tells us is the score its popping is associated with a range of K/S /s values. For example, you get 1000 points for getting 50-60 K/S /s and maybe 2000 points for getting 61-70 K/S /s... etc. So that if you run 52 one time and 58 the other, that would be why the actual score doesn't change. But if you pay attention to the actual value of K/S /s, it varies (a little) run to run. You simply haven't passed the top of your score range yet I think.
Last edited:
- Joined
- Sep 13, 2015
Sorry I'm on the road with work, I got called away to help at another plant at the last minute late last week. I don't have all my data with me but I will peruse what I have and try to get everyone some awnsers today or tomorrow.
@ed, I'm not understanding this comment
What are you basing this off of. I don't remember posting my raw K/samples value, I thought I only parts muy overall opencl score. I'm not saying you are wrong because I never really paid attention to K/samples on each run just the overall score.
@ed, I'm not understanding this comment
What are you basing this off of. I don't remember posting my raw K/samples value, I thought I only parts muy overall opencl score. I'm not saying you are wrong because I never really paid attention to K/samples on each run just the overall score.
- Joined
- Apr 3, 2015
- Location
- Planet Earth
Yeah, but you can't see if the CPU option is selected in the actually bench. I just hit printscreen when it was open.
You also have to position the main window so it doesn't cover it up.
Buenas
I'm basing it off of the dozen or so total runs I made across my two systems. I think my typo (he was supposed to be THE..
) is what threw you there... my fault and fixed!@ swart... not sure there... I would bet those that have the intel open cl driver installed it would have the cpu checked as it then uses the cpu more. This does help scores a bit, but not nearly enough to make your open cl score close to what ours were with a 980ti.
Last edited:
- Joined
- Sep 13, 2015
When I installed the AMD sdk app to perform the opencl calculations on the cpu the bench did not "check" the box to show it was utilizing the cpu. You can tell it does based on an increased score and you should see 100% cpu usage during the test.
- Joined
- Aug 14, 2014
Was browsing the Seti@home forums (fold on and off since 2001) and caught this chart, a guy has been collating GPU performance for a few years and apparently it uses OpenCL. The 980ti rates above the 1070 and Titan X in raw performance :
https://setiathome.berkeley.edu/forum_thread.php?id=79874&postid=1820965#1820965
https://setiathome.berkeley.edu/forum_thread.php?id=79874&postid=1820965#1820965
Seems it peaks a little higher and starts a little lower in that type of work.
I also seemed to get more out of the 1070 in the open cl test (pre intel cl driver) on the 6950x system with slower cpu clock, than on my 6700k. I'm pretty sure that is an EarthDog thing though.
I also seemed to get more out of the 1070 in the open cl test (pre intel cl driver) on the 6950x system with slower cpu clock, than on my 6700k. I'm pretty sure that is an EarthDog thing though.
- Joined
- Aug 14, 2014
From the Realbench website : https://rog.asus.com/articles/overclocking/realbench-benchmarking-stress-test-insights/
- Joined
- Apr 3, 2015
- Location
- Planet Earth
So this is one biach of a bench. I kinda figured out how to get a consistent good score for #3, but then my Image Editing, #1, score took a shiat???!!!
I can't seem to reproduce the Image Editing score I posted. If I could get both #1 and #3 to fire at the same time, I would be golden boy!! But, alas still just a hack
Walking away from this for now, back at it before end-o-month!
Cheers
I can't seem to reproduce the Image Editing score I posted. If I could get both #1 and #3 to fire at the same time, I would be golden boy!! But, alas still just a hack
Walking away from this for now, back at it before end-o-month!
Cheers
You really need to run it back to back.. you can set the number of times to run it consecutively. As I said, my first one alway.s is way slow, but on any subsequent runs, the subsequent cores for all benchmarks are where they should be.So this is one biach of a bench. I kinda figured out how to get a consistent good score for #3, but then my Image Editing, #1, score took a shiat???!!!
I can't seem to reproduce the Image Editing score I posted. If I could get both #1 and #3 to fire at the same time, I would be golden boy!! But, alas still just a hack
Walking away from this for now, back at it before end-o-month!
Cheers
Just going to post this up as my last score (ran on 11/3). I put the 1070 with the 980Ti this morning, and though the Open CL score went up a holy ton, the score didn't budge... my best was with a single card...and below.
Earthdog / i7-6700K / Water / 158582 (GTX 1070 @ 2088 Actual Boost/2107 Mem)
Earthdog / i7-6700K / Water / 158582 (GTX 1070 @ 2088 Actual Boost/2107 Mem)
Last edited:
Similar threads
