Sign in to follow this  
Daedalus

low XTU score?

Recommended Posts

Hi all!

 

Need your overclocking expertise!..I set up a 6700K system lately (see my signature), and tried to see how it overclocks yesterday.

 

I got at 4.8Ghz at 1.4v with HT off. Max temp was 70C and I scored 1051 marks.

 

I have the latest BIOS (F20) and in the BIOS settings I just changed the base frequency, v-core, HT-off and set the XMP profile. Timings was automatically set to 15-15-15-35 on Channel B. My cooler is a corsair H100i.

 

I am wondering why I score so low, given the fact that the scores I see with the 6700K are from 1300-1500 marks.

 

Is there any settings that I am missing maybe?

 

Also does the mark reflects P3D performance? I mean, 20% more in XTU reflects +20% more fps?

 

Your help will be very much appreciated :)

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

4.7Ghz at 1.37v 1040 marks

 

Seems like 4.9Ghz is achieved around 1.43v.

 

However, the XTU marks are very low.

Share this post


Link to post
Share on other sites

Have you considered that you have HT off, and that the scores of others are probably HT on?

 

Have you tested with anything else other than XTU?

My ROG RealBench results for my 6700k, at 4.6 with HT on are...

 

Image Editing: 51791

Encoding: 183864

Open CL: 100032

Heavy Multitasking: 155068

 

System Score: 102682

 

CineBench: 1007

Share this post


Link to post
Share on other sites

Thank you Martin! Always helpful and spot on :)

 

Just run the ROG RealBench. at 4.8Ghz it was crashing on the encoding test. So I lowered to 4.7Ghz

 

The results I got are:
 

Image Editing :241698
Encoding: 147317
OpenCL: 93470
Heavy Multitasking: 143168

 

System Score: 137719 

 

Cinebench: 771 

 

Actually you are better on everything except the Image Editing, even at lower frequency.  Still my results are not the expected.

Share this post


Link to post
Share on other sites
Actually you are better on everything except the Image Editing, even at lower frequency.  Still my results are not the expected.

 

 

 

But do you still have HT off? 

 

You cant expect to match my score if you have HT off.  :smile:

 

 

Okay, ignore my previous score, I rushed and messed up.

 

Ran it again and...

 

 

Image Editing: 227559

Encoding: 184260

Open CL: 100032

Heavy Multitasking: 186328

 

System Score: 154538

 

Don't forget to run RealBench Stress Test... it's all I use these days. Run it for one to two hours. All you need.

Share this post


Link to post
Share on other sites

Thanks Martin. :smile:  Ok, I was running with HT off before. I now swithed to Ht-on at 4.6Ghz to compare fairly. What I got is:

RealBench:
 

Image Editing :233834
Encoding: 177125
OpenCL: 93470
Heavy Multitasking: 179262

 

System Score: 152228

 

Cinebench: 964

 

 

So, I suppose I should not worry? But don't you use HT-on for FSX-P3D? I'm coming from an i5 and little confused with the proper setting for HT.

Share this post


Link to post
Share on other sites

Yep, you are all good, don't worry. :smile:

 

Each run Of RealBench will generate a slightly different score so we will never match exactly.

 

 

So, I suppose I should not worry? But don't you use HT-on for FSX-P3D? I'm coming from an i5 and little confused with the proper setting for HT.

 

 

 

I don't sim any more to be honest. Got bored with it after all these years. Just BF4, BF1 and Elite Dangerous Horizons at the moment.

Share this post


Link to post
Share on other sites

Yep, you are all good, don't worry. :smile:

 

Each run Of RealBench will generate a slightly different score so we will never match exactly.

 

 

 

 

I don't sim any more to be honest. Got bored with it after all these years. Just BF4, BF1 and Elite Dangerous Horizons at the moment.

 

 

Feel better now :) Oh, ok! I really liked BF4 :) But good to have you here, your experience is very helpful again thank you :)

Share this post


Link to post
Share on other sites

XTU for best score , HT-on fast memory tight timings high uncore high CPU freq

Cinebench HT-on high uncore high CPU freq

Rog RealBench benchmark. CPU speed , 4 fast sticks ,980ti for open cl test

My score on RealBench is + 171000 at 5ghz with 980 Non ti watercooling ambient 20c.

I run 4x8 at 3600mhz cl14

 

Bench marks like XTU, Cinbench , Rog RealBench can be tricky to get high scores be aware that Its lot of tweaks.

Share this post


Link to post
Share on other sites

Thank you very much westman for the input :)

 

What I am interested in is best tuning my system to achieve maximum performance in P3D..I assume that a higher score would indicate a better tuned system, but I'm not 100% if its that the case.

Share this post


Link to post
Share on other sites

Ifikratis - do not underestimate the importance of not only the basic memory timings but the sub-timings as well. They put me from about 1200 to my current HT on XTU score of 1550.

Share this post


Link to post
Share on other sites

Hello Garrett, thank you :)

 

I upload images from my BIOS maybe, you or anyone can spot something. Also about sub-timings, that's important info ! Any suggestions? :)

 

With HT-on I find my max at 4.6Ghz and 1.41v for stability. I could possibly go at 4.7 but I will go higher on v-core which I don't feel very comfortable to.

 

So far, my results in P3D are astonishing. I can fly the NGX on Simwings EGLL on steady locked 30fps, while before I was ~20fps. Unlocked it was ~35-40 fps. All sliders to the right and all shadows on. However if a +200 mark points increase in XTU can be reflected in P3D, its something I'm certainly interested to get.

 

161118222949.jpg
161118223007.jpg
161118223031.jpg
161118223227.jpg
161118223255.jpg
161118223726.jpg

Share this post


Link to post
Share on other sites

Turbo Boost appears to be turned off. Turn it back on and ensure that the CPU is clocking to 4.6Ghz properly when benchmarking. And don't split hairs between 4.6Ghz and 4.7; we're worried about the huge deficit here and not the minor difference between 100mhz.

Share this post


Link to post
Share on other sites

Turbo Boost appears to be turned off. Turn it back on and ensure that the CPU is clocking to 4.6Ghz properly when benchmarking. And don't split hairs between 4.6Ghz and 4.7; we're worried about the huge deficit here and not the minor difference between 100mhz.

 

To my understanding the Turbo boost is boosting the frequency to the top value when needed. However I had the base clock set already to 4.6Ghz so it wouldn't make a difference.

 

However I did enabled Turbo Boost and set all 4 turbo to 4.6 too to test. I got 1059 marks.

Share this post


Link to post
Share on other sites

Turbo Boost appears to be turned off. Turn it back on and ensure that the CPU is clocking to 4.6Ghz properly when benchmarking. And don't split hairs between 4.6Ghz and 4.7; we're worried about the huge deficit here and not the minor difference between 100mhz.

What huge deficit? There isn't one.

 

He had HT off and was comparing to results with HT ON. You may not have read all of the posts.

 

His OC is fine. No more tweaking required.

 

Stress test with RealBench and he's good to go.

 

No point at all messing about for miniscule gains, just to get a slightly bettet score.

Any "fine tuning" will yield pretty much nothing in terms of frame rate increase.

 

My advice is don't stress, relax, enjoy your new system.

 

We build our systems to run the sim and other apps... not achieve a few more points in a bench mark.

Share this post


Link to post
Share on other sites

There isn't one.

 

Speak for yourself, Martin. With a 6700K (HT on) @ 4.7Ghz, I get an XTU score of 1560. Memory at 3333Mhz with timings of 14-14-14-34-2T.

 

Also, here's my realbench:

 

image editing: 228704

 

encoding: 189593

 

opencl: 103313

 

heavy multitasking: 181054

 

system score: 155003

Share this post


Link to post
Share on other sites

 

 

Also, here's my realbench:

 

system score: 155003

 

 

 

And he's getting 152228  in RealBench and beats you on Image Editing. Perfectly acceptable score. Especially When RealBench generates different results each time you run it. And that's at his 4.6 rather than your 4.7. 

 

As for XTU... as I said, he was running with HT off, and comparing to results with HT on. 

 

He has zero problems and any further "tweaking" in an attempt to match your score or my score would be pointless and would yield results that are minuscule in terms of frame rate.

 

As Hasse said...

 

 

Bench marks like XTU, Cinbench , Rog RealBench can be tricky to get high scores be aware that Its lot of tweaks. 

 

 

 

Worth remembering that just because you get high scores in synthetic stress tests and benchmarks, doesn't necessarily mean it will automatically translate into significantly improved performance in the applications you run. To ensure a new system is roughly in the ballpark it should be in, fine, tuning for fun, fine.

 

If you happen to be someone like Westman, a competitive overclocker then it's different, and tweaking to achieve max scores are for a different purpose. But for the rest of us... not necessary at all. He can tweak here and tweak there, adjust this and fiddle with that, and nudge his benchmark sores up... but he wont be suddenly throwing his hands in the air as a result, because he's achieved flight sim Nirvana.

 

We are all free to do what we like though, so if he fancies tweaking for the fun of it, fair enough.

Share this post


Link to post
Share on other sites

Thank you very much for the input and valuable help to this guys. :smile:  Of course I am not going to any overclocking competition so the actual score numbers themselves are not my goal. If the XTU score is not a realistic representation of performance in P3D or any other app then I don't mind. As martin said on the RealBench benchmark I get anticipated values so then I'm pretty happy with that. I would be concerned if on all tests I was getting far less. by the way when XTU is testing, one can see that the CPU is going from idle to max utilization all the time, so maybe that's why the ram speed and timings can make a difference. On flight sim the RAM is constantly feeding the CPU so maybe will not make a so big difference. My performace in P3D has sky-rocketed from my i5 3570K..and this tells me that the system did made a big difference.
I use 4.6Ghz@1.41v, HT-on, AM=87 currently. 

Share this post


Link to post
Share on other sites

What kind of voltage is that by the way. Fixed, offset or adaptive.

 

I'm at 1.36 @ 4.6, Adaptive.

 

Is that the lowest your chip goes at 4.6?

Share this post


Link to post
Share on other sites

It should be fixed voltage.The setting can be seen below. I didn't see any option for adaptive voltage. I revert to default BIOS settings when not flying.

Yes, XTU crashed once on 1.4v so I increased to 1.41 and it didn't. But haven't stressed tested yet, this is to be done.

 

161118223726.jpg

Share this post


Link to post
Share on other sites

Worth remembering that just because you get high scores in synthetic stress tests and benchmarks, doesn't necessarily mean it will automatically translate into significantly improved performance in the applications you run.

 

Yeah, no crap Martin. That's the first rule in benchmarking club.

 

Ah well, glad I could help. What do I know anyway?

Share this post


Link to post
Share on other sites

Great temp there Hasse. How did that do in comparison to the other coolers?

Share this post


Link to post
Share on other sites

Hello westman! 

 

4.7 Ghz @ 1.25v on air cooler and 55 C? Simply amazing! I need 1.41v for 4.6Ghz on a water cooler (H100i) ! 

I see 7 utilization cores, so I suppose its with HT-On! Is there any trick to decrease the vcc requirement or just silicon lottery? 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this