MSI GT80 3dMark Scores, test and tune

Everyone, I think this will be a great place to share our scores which can help track commonalities and differences between each GT80 model plus providing great information for those troubleshooting simply wondering what type of performance to expect before purchasing.

Instructions to follow
1) If you can when adding your score try to keep things in a proper format so things are easy to read, compare and to allow for proper thread searching if needed.
2) Please perform all testing without cooler boost running or any other supplemental cooling.
3) Please submit a screenshot of your score with hardware monitor temps being displayed.
4) Please leave useful comments in their appropriate sections.

Who knows if we get enough submissions we create "The List" for the 10 ten for each model.

Model (001/009/047/071/094)
Test (Fire Strike/Fire Strike Extreme/Fire Strike Ultra/Sky Diver/)
Thermal Compound (Stock/ICD/CLU/Antec7 or many more)
CPU/GPU tuning (Yes CPU -50 voltage offset, CPU Core +100, GPU +50, Memory +200 / No)
Shift Mode (Green, Comfort, Sport)
Room/Ambient Temperature 72
Comments (Stock Configuration first test ever!, Overclock testing CPU, EC Bios update)
Results with pic and or link, if you have already posted the image in another thread such as the "Owners Lounge" and which to post a link to it this should be fine too as long as other information is completed.

Members if I am missing anything please make suggestions via PM so the thread stays clean as possible, Thank you Everyone.

Please only submissions from the MSI GT80

 

Model 009
Test Fire Strike
Termal compund CLU on CPU and Antec7 on GPUs
CPU Tuning -60 dynamic core voltage offset (Stock Clocks CPU & GPU)
Shift Mode Sport
Room/Ambient Temperature 72
Comments This was my first run after switching to CLU and repeated testing.

 

Model 009
Test Sky Diver
Termal compund CLU on CPU and Antec7 on GPUs
CPU Tuning -60 dynamic core voltage offset (Stock Clocks CPU & GPU)
Shift Mode Sport
Room/Ambient Temperature 70
Comments Fresh Morning Sky Diver run after CLU application on CPU

 

Model 009
Test Fire Strike
Termal compund CLU on CPU and Antec7 on GPUs
CPU Tuning 4.3ghz x4x3x2x1 with -20 dynamic core voltage offset (Stock GPU)
Shift Mode Sport
Room/Ambient Temperature 72
Comments CPU overclock testing to see score impacts.

 

Model 009
Test Fire Strike
Termal compund CLU on CPU and Antec7 on GPUs
CPU/GPu Tuning 4.2ghz x4x3x2x1 with -60 dynamic core voltage offset GPU + 135 Memory +580 (Stock vbios)
Shift Mode Sport
Room/Ambient Temperature 72
Comments CPU and GPU overclock testing to see score impacts.

 

I will participate with my benchmarks when I return from work, in about a week, two tops. I would like to know the ASIC quality of everyone's GPUs, just to see if we can correlate anything

I have noticed that memory overclocking is a bit dodgy so far. I "used" to have stable 400mhz OC on memory (+200 *2) but in reality it was unstable when gaming on long periods of time. My current OC is at +135 core and +250mhz (+125*2) memory fully stable at any game.

 

Has ASIC quality provided useful with temps and or overclock abilities? In the past I have never paid much attention to it only because of my lack of truthful education on it. Please explain how it can help when possible, thx.

 

Model 009
Test Fire Strike
Termal compund Stock
CPU Tuning Stock Clocks
Shift Mode Sport
Room/Ambient Temperature 70
Comments Initial run, no modifications

 

Temps are pretty high. I definitely need some better thermal paste.

 

Same CPU temps as my unit, and my ambient room temp was 72F. Temps were lowered a few degrees with a reapplication of the thermal compound, but don't expect miracles.

 

Interesting indeed, great data and thanks for your submission. Let us know what compound you end up testing and your results.

 

Anyone have an reports of using Gelid thermal maternal on the GPU's or CPU?

 

Well a good friend of mine (Ken), bought a 009 from Ken at Gentech and got it a couple of days ago. He's pretty happy so far. I'll have to get him to show up on the forum.
He paid $35 for "IC Diamond 24 Carat Thermal Compound on both CPU/GPU" upgrade and interestingly his max CPU temps were exactly the same as mine for Fire Strike. We took steps to make sure we were comparing apples to apples (same geforce driver, same ambient temp, same at rest surface, so on)

I played around a bit with Intel Extreme Tuning Utility, lowered the Reference clock a bit and same with Turbo Boost Power and it didn't really help much with the CPU temp.
I think for now I'm going to skip opening this up and applying new paste. The CPU will allow 100C and so far I've never gone over 96C. I'd feel better if I could get this down into the 80's but right now that seems unrealistic.

 

I have said it many times that to pay for upgraded paste is a waste because the stock performs as good of not better. I would be interested in trying GELID extreme for a non conductive material otherwise of you want the best results go with Cool Laboratory liquid ultra as I did.

Thanks for posting your results, very much appreciated from all of us.

Also running a negative dynamic core offset between -50 to -90 will usually drop temps too.

 

It's hard to say with paste because differences are not usually huge and other factors can swamp out any advantages it may be giving such as paste job, heatsink tolerance margins, ambient temperatures, exact loading and thermal sensor calibration are a few of the large list of variables.

 

I would not say upgrading thermal compound is a waste. Some, such as myself, do this routinely as a matter of added assurance, and most of the time temps are indeed lowered a few degrees. I have seen times when factory paste applications are inadequate, or could be improved, and when purchasing a new notebook I like the extra set of eyes on this..... and knowing the job has been done right. The correct application of the thermal compound is of the upmost importance, more so than which paste is used.

Secondly, I would take care in endorsing a specific thermal compound, as even amongst credible sources, data is varied..... and good and bad can be found in the minefield of reviews, and web sources. As an example, Cool Laboratory Liquid Ultra should not be used with aluminum, is electrically conductive, great care should be used when applying the product, and many reports can be found suggesting it's thermal conductive property degrades, resulting in increased temps over a relatively short period of time..... it's hard to clean or completely remove without leaving electrically conductive residual...... and has fused to CPU or heat sink components. Now, this is not my experience, yet the reports exist nonetheless. Personally, I try to avoid any electrically conductive thermal compounds, as a matter of practice..... and I never recommend them for general use.

Thirdly, no doubt about it, GT80 CPU temps run a little warm, and I would not expect any miracles with different thermal compounds. Even amongst the best reported thermal compounds, reliable and independent testing generally only result in marginal differences..... of a few degrees. Heck, I have seen the same result, with different TIM applications to the same notebook, using the SAME thermal compound. Again, correct application of the product is most important.

Fourthly, GTVEVO's CPU temps seems to be the exception, not the rule, and I am reluctant to attribute this to any given thermal compound. I would like to see his CPU temps when compared with another unit, under the same testing parameters, same environment. This has been done by MSI with my notebook, and my CPU temp was only 2 degrees warmer, than a new unit pulled from their warehouse. Shortly thereafter, Ken received the unit, reapplied the TIM application, and only managed to reduce CPU temps a few degrees. Also, adjustments to the EC fan table were no help.

Bottom Line: CPU temps run higher in SLI notebooks, and the GT80 CPU temps are a bit warmer yet. I dont believe thermal compound will change this fact.

 

All due respect here and everyone has a right to their own opinion and beliefs. I agree that CPU temps will run higher in SLI notebooks because of the amount of components exhausting heat and the complexity of their cooling solutions. This is also why heat transfer is even more important and why better connectivity can help draw away heat from the components and through the cooling solution. But regardless some will try different thermal material to see what works and what doesn't and until we at least try's to learn we will never complete all variables of testing analysis. I thought this would be a great thread to learn and this was the proper platform in doing this. Hopefully others will keep posting their experience and testing results to help us all better understand.

 

Meh, no need to take this personally......

Anyhow, Meaker makes a good point in post #15 above, and I somewhat alluded the same in my previous post to you, HERE.

Bottom Line: CPU temps run higher in SLI notebooks, and the GT80 CPU temps are a bit warmer yet. I dont believe thermal compound will change this fact.

 

Ah nothing taken personally so don't worry about that but can the thread continue to collect data for educational purposes?

Meaker brings up a point that is valid and brought up many times but there is a margin of area with any public testing and data collection and there isn't any way around that.

At the very least I would think to collect and display results of performance and temp data is useful before making a 3k+ purchase.

Bottom Line: I think we know were you stand and I hope that you can be happy with your purchase and temps at some point in the future.

 

Sure, collect all the data you want, for what it's worth...... but for any integrity, there has to be control standards, which can be sorely lacking with 'public testing', amongst end-users posting screenshots in a forum thread. Even if you establish rules, there are no assurance data can be deemed reliable, or that the 'your instructions' were followed. Independent controlled testing is needed from a reliable source.

Oh, I am happy with my purchase, even with the higher than desired CPU temps. After all, this was MSI's first SLI release, and overall they did a fine job of it..... but CPU peak temps are high nonetheless. Where do I stand? Telling it like it is, let the chips fall where they may, in hopes CPU temps will be improved with MSI's next iteration...... which I am sure, it will be. Criticism should not be seen as a bad thing, but rather a pathway to progress, and I feel certain MSI is listening...... so don't be surprised when I am moving on to MSI's next, more powerful, more efficient, cooler running flagship.

Bottom Line: CPU temps run higher in SLI notebooks, and the GT80 CPU temps are a bit warmer yet. I don't believe thermal compound will change this fact.

 

Model 009
Test Fire Strike
Termal compund Stock
CPU/GPU Tuning -> CPU= -50 dynamic core voltage offset GPU= Core +135, Memory +350
Shift Mode Sport
Room/Ambient Temperature 70

 

Model 009
Test 3dMark11
Termal compund Stock
CPU/GPU Tuning -> CPU= -50 dynamic core voltage offset GPU= Core +135, Memory +350
Shift Mode Sport
Room/Ambient Temperature 70

 

Reached 90c after -50 dynamic voltage offshet

 

Temps look great on fire-strike and so does your score.! Thanks for your submission.

 

This thread of for the MSI GT80, sorry dude, those are nice numbers but it should be removed.

 

my paste will be in tomorrow and will be trying this out. But do remember ... I am not in SLI. lol

 

Sounds great, be sure to post up your shots before and after to help us see your improvements. thanks!

 

yes sir !!

 

Model: I7 4720HQ and GTX 980m sli
Test Fire Strike
Termal compund: Arctic Silver 5 on CPU and both GPU
CPU/GPu Tuning CPU at Stock GPU + 135 Memory +200(Stock vbios)
Shift Mode Sport
Room/Ambient Temperature 77
Comments only GPU OC, without Cooler Booster



What du you think about this score and the Temperature?

 

Taking into consideration your room temp at 77 and your machine temp peaking at 87 I would say it looks pretty good. It seems the 4720 at stock clocks average right around 1300-13300 and you are at 13916 with a moderate GPU overclock. Do you feel like its under performing?

 

no i think thats pretty well, i wos wondering at the physics score in 3D MArk 11! ( Link in the oder Thread)!

Is the copper Cooler you have pretty better as the Cooler at stock? What was your temperature before you had the copper Cooler? Can i order this cooler in Austria?

 

sorry for the post

 

no i think thats pretty well, i wos wondering at the physics score in 3D MArk 11! ( Link in the oder Thread)!

Is the copper Cooler you have pretty better as the Cooler at stock? What was your temperature before you had the copper Cooler? Can i order this cooler in Austria?

 

I have a stock cooler so no changes there but to compare I do have the 4980 so where you are running 3.4 ghz I am running 3.8ghz on stock clocks so this will make a difference especially in physics. If you want to improve your physics score you do have the ability to overclock the 4720 up 200mhz on each stock core setting, you can do this though Intel XTU. I would also recommend that you run a negative dynamic CPU offset around -50 mv to start and work it from there, this usually helps remove some of the over voltage that is not necessary when you are not using the 4720 built in graphics and in return it can reduce your CPU temps.
One other difference is I using Cool Laboratory Ultra thermal paste on my CPU to reduce temperatures, personally it has worked the best on haswell laptops for me but you must use it with caution as it is conductive. When used correctly its very safe and works very well even though the stock MSI thermal material seems to work very good too.

 

The physics score is certainly low. It should be over 10k when compared to GTVEVO's figures.

GT80-4720_FireStrike_Michael1983 by Cormogram posted Mar 31, 2015 at 1:14 PM

Try killing all unnecessary processes and/or disabling services that may compete for the CPU, such as Windows Update and anti-virus.
Note: I always run benchmarks off the network to avoid additional spurious interruptions.

 

How to put a negative dynamic CPU offset around-50 mv ?

 

I pulled some stock GT80 4720 scores and Physics for me was around 9100-9200 at 3.4ghz and around 9700-9800 at 3.6ghz with stock GPU and stock thermal material

 

Intel Extreme Tuning Utility

 

I did turn off all programs what is bad for CPU usage but i get no higher score in Physiks! What can i do to solve this problem?

 

Overclock the 4720 CPU and test, just monitor temps but I would say you are well within the safe zone.

The only times I can see that I was able to reach physics scores of this was when over clocking the CPU + 200mhz.

 

Ok but in the List from Cornogram you have higher Physics not overclocked?

 

I see, I am unsure where that list came from and I can't find a test either where I have that score so it could be just an oversight on his personal list or a previous post it was pulled from. Let us know if you hit your goal once you make the changes once you feel are needed.

 

i feel not needed to OC the Cpu because its enough power! only i was frustrated when member with the same setup get more score on Physics! i had belived thats something wrong with my CPU !

 

I understand and agree.

 

Your scores seems fine to me. Your combined score is very close to mine. I wouldn't worry about it.

 

Whats your Score ryzeki?

 

Ooops! My mistake. I've forgot to update the name after a copy & paste.
The values come from[1]:

It seems the CPU was tweaked but it's not mentioned what exactly was done.
@Silentlegion Can you tell us what was the CPU clock of that Fire Strike test?
@GTVEVO Do you have Fire Strike values for the stock settings?
So, here's the updated comparison with @Midou stock settings data:

GT80-4720_FireStrike by Cormogram posted Mar 31, 2015 at 6:59 PM

[1] http://forum.notebookreview.com/thr...tan-owners-lounge.769092/page-46#post-9948332

 

To get higher physics score you must assign the PhysX processor to the CPU. This is done in the Nvidia Control Panel in the same window where you enable SLI.
Note that not all applications may benefit of this setting. If you have one that depends heavily on CPU, such as Battlefield 4, the actual performance (FPS) will drop.
Worth trying though!

 

I remember my first run before installing any programs my 001 / 4720 GT80 scoring 13218 in firestrike, The GPUs showed 68c and 62c and the cpu peaked at 78c as I have it documented but I didn't save the 3dmark run due to my first machine the SSD crashed within just 2 days causing me to loose it. The runs after I had installed my virus scan which was causing my score to degrade about 300 points (don't use avast virus scan) I have attached below. Everything at stock clocks so its low but its a ballpark.



My 009/4980 is posted on page one.

 

Thank you!
I would like to ask you guys to include the link for the 3DMark reports, such as: http://www.3dmark.com/fs/3986089
Also, it would be interesting to include ASIC Quality values of your graphics cards.
You can get it with GPU-Z[1]. Just run it and click on the top left icon (on the window title bar) and choose "Read ASIC quality...".
It will only show the value of the current selected GPU. So, to get the other one you need to change the GPU at the bottom of the window and repeat the read.

[1] http://www.techpowerup.com/gpuz/