PDA

View Full Version : anyone els getting low scores on v.54



LordZaros
04-02-2017, 02:07 AM
my old realbench version tests pretty high but this new one is really low like in th 90k's

Arne Saknussemm
04-02-2017, 07:16 AM
No point comparing scores between versions...the scoring formula has changed...especially in the last couple of betas...

LordZaros
04-02-2017, 11:49 AM
I am running an i7-7700k clocked at 5.2ghz and 16gb of dominator platinum running at 3200mhz and a gtx 980 overclocked to 1475mhz I really do feel like I should be scoring higher then I am. The previous version of Realbench that wasn't optimized for Kabylake was giving me 266k scores but I couldn't upload because of some 7th gen issue that was fixed on this later version*

Zka17
04-02-2017, 11:59 AM
It works...

63677

But it crashed my previously stable OC profile... - looking into it now!

LordZaros
04-02-2017, 12:06 PM
I think it's an issue with Kabylake or the ix hero motherboard. I've used every version of Realbench prior to this one and I really do think my scores should be much better and the bench itself should be running quite a bit faster. Look up scores on the leaderboard username is "flail manx"

Arne Saknussemm
04-02-2017, 01:25 PM
I'll repeat...different versions have different scoring formula...you cannot compare...

2.44 should have scored around 170000 more or less...anything else is a glitch...latest beta version should not be far off that...maybe a bit more.

2.53 had a much different formula and scores were way higher...up to 500000

Praz
04-02-2017, 01:39 PM
Hello

As you have this thread please refrain from posting the same thing in other threads. Thank you.

LordZaros
04-02-2017, 04:05 PM
so the leaderboard needs to be reworked?

Zka17
04-03-2017, 01:01 AM
I assume that it will be... soon or later...

Nate152
04-03-2017, 01:10 AM
Here's my Realbench score with the 7700k, ROG Strix 1070 and 16GB of Kingston HyperX Predator 15-15-15-35-2T.

I imagine with Hyperthreading off this lowered my score some.

63692

LordZaros
04-03-2017, 01:21 AM
i found it out by accident but gpu tweak does not ply well with the new relbench. i gained back 40k of my score by removing it.

Zka17
04-03-2017, 01:54 AM
Could you share more about that GPU Tweak thing?
What was exactly the problem? Which version of GPU Tweak? HOw were you using it?

LordZaros
04-03-2017, 02:19 AM
just as its supposed to be used i never noticed anythng wrong with it. i just uninstalled and bam, my pc zipped through the image editing. so i thought if i uninstalled
the AI suite it would improve somthing els but i was wrong ther it just slowed me right back to where i was. its really weird

Medico
04-07-2017, 06:28 PM
I was wondering about that. I was about to create a thread about that. I have that situation too. That version is getting lower scores, significantly lower. Version 2.44 was getting better scores. I had screenshots with evidence of that situation. Do I need to post it here? The Beta version 2.53b was getting the highest scores. Arne said we should not compare. I noticed version 2.54 getting worse scores in all four areas including OpenCL score. This score is lower by about 12,000. Video editing score is lower by about 80,000. Encoding is lower by 92,000. Heavy multitasking is lower by 87, 000. Total score is lower by about 50,000. I was hoping that the newest version would provide similar calculation of the scores as 2.44, specially when you try to improve what is on the RealBench leaderboard. With the new version, we will never beat the score we got in 2.44, unless you make an additional RealBench leaderboard for each version, one with version 2.44 scores and another with version 2.54 scores.

Arne Saknussemm
04-07-2017, 06:43 PM
Well, since the formula has changed it is pretty moot to try comparing scores....the table will have to be wiped or a new one made I guess...

However my top score on 2.44 was 225000 at 4.8ghz on a 5960x and on 2.54 it's 224000 at 4.6ghz on a 5960x so I'm not sure there should be that much difference?!

63766

Zka17
04-07-2017, 09:37 PM
Although I am still playing around with my settings, but comparable ones gives pretty close scores...

Medico
04-23-2017, 09:01 AM
Although I am still playing around with my settings, but comparable ones gives pretty close scores...

This is the evidence about the difference between RealBench 2.44 and 2.54. This is done one after the other. Same system as you can see in the information. As you can see, the difference is significant. The only area that I see improvement and similar scores are Image editing. OpenCL score went down, similar score that I had with Geforce 970. See the attachments please.

JustinThyme
04-23-2017, 04:02 PM
As Arne put it you cant compare one version against another. Test changed as did scoring method.
My rig went from 170K in 2.44 down to 125K in 2.54 just for comparison purposes.

JustinThyme
04-23-2017, 04:02 PM
As Arne put it you cant compare one version against another. Test changed as did scoring method.
My rig went from 170K in 2.44 down to 125K in 2.54 just for comparison purposes.

Caipirinha
06-08-2017, 09:01 AM
Hi guys Can you see my thread please >>> https://rog.asus.com/forum/showthread.php?93797-HELP!!!-RealBench-Encoding-amp-Heavy-multitasking-WEIRD-LOW-SCORE-!!!!!!WTF

oh now suddenly i saw this thread and you are talking about different formula and scores of different Versions of RealBench right??!!
2mins ago i posted a whole new thread about my issue with LOW SCORES of Encoding and Heavy Multitasking in my build COMPARED TO OTHERS IN Leaderboards page like im 134000 on encoding and others with exact same system (cpu&ram) with even lower ferequencies and higher ram timings HAVE WAY MORE BETTER SCORES on Encoding&Heavy Multitasking their encoding score is 220000 and their heavy multitasking is 192000 to 200000 BUT MINES ARE 130000 on Encoding and 153000 to 165000 on Heavy Multitasking score.

so is that Right??am i Right??!! my RealBench version is 2.54 So Now if i go back and benchmark on version 2.44 or 2.43 the scores are going to be in Right place??!!!

Zka17
06-08-2017, 12:22 PM
Caipirinha, see the answer in your thread...
Bottom line is that the valid scores are always the newest version ones...