Clevo Overclocker's Lounge

So it seems to be really f*!*ed up in SLI. Can't start it either. Hopefully they will release a fix soon.

 

Well, after removing it, running a disk check and CCleaner then fresh installing 3DMark if won't run at all for me either. Crashes loading test #1 every time... 3DMark has stopped responding, then it goes to a screen with a valid score of zero, LOL.

Hopefully, it won't take them 2 to 3 years to release a stable version like it did for the Catzilla developers. That was what my smart remark to Brother @johnksss was poking fun at.

Edit: wait... it's been 2 years since Micro$haft started the Windows OS X Insider Preview and it's still not right. I'm seeing a trend here, guys.

 

Looks like the latest hot and steamy deep-dish colon loaf named Time Spy is probably going to be quite the hot topic in the gripe section at Futuremark Forums for sure. Hardly anyone posting about it yet. @CaerCadarn - there is a post about Avast screwing things up. But, I don't have ANY kind of antivirus software, not even Windows Defender, and have the same symptoms that are being blamed on Avast, so go figure.

https://community.futuremark.com/forum/forumdisplay.php?169-3DMark-Windows-Android-and-iOS

Edit #1: SO, I closed 3DMark and now none of the benchmarks show as installed and all of them are reinstalling themselves now. That's just too weird.

Edit #2: And, 6 minutes later only half are finished and it's taking forever for 3DMark to finish automatically reinstalling the benchmark apps. Seems like they really botched this up.

 

I did what you said and deleted the programdata folder and then all the benchmarks disappeared.

 

Wasn't your results saved online as well? Or do they have to be connected to your locally saved results to be shown?

 

Yeah, same here and it took about 10 minutes for them to install again.

So, after it finished I rebooted and with SystemInfo scan disabled it runs without crashing, but cannot validate the score. This was a stock GPU run. Going to re-enable SystemInfo scanning and see what happens now.

http://www.3dmark.com/3dm/13227458

 

http://www.3dmark.com/spy/24524

Going to have to try that and see if it will run.

 

I got it to finish with SystemInfo scanning turned on, but it says the driver is not approved and it is approved with other Futuremark benchmarks. Same driver I have been using for a month, so the guys at Futuremark really didn't give this the effort it deserved. I am really disappointed that an industry leader released such a half-baked piece of junk to their customers.

http://www.3dmark.com/spy/26088

 

I'm hating this release now patch later work ethic that's slowing getting accepted.
It's so cheap and unprofessional.

 

Well, setting aside all of its pathetic functionality and stability shortcomings, it looks like this is going to be just another fluffy Willie Wonka turdbook-friendly benchmark like Fire Strike. Your beast CPU with twice as many cores and threads literally mopped the floored with my 6700K (as we would fully expect it to) and scores a whopping 3454 more points in physics and yet the benchmark gives hardly any credit at all... less than 200 points... in the overall score. So, yeah, let's give a pass to the crapbooks so the kiddos can feel like big boys with their wimpy low-TDP BGA crackerjack trash and not be embarrassed about their anemic gamer toys. Ain't that cute. Score one for the Ooompa-Loompas.

This is what I am referring to: http://www.3dmark.com/compare/spy/24524/spy/28020#

 

60% difference in CPU and almost no difference at the end. Definetely won't like it!!!

 

I wish they added more combined tests. And yeah that everyone gets a trophy mentality on physics weighing is lame. We already got Firestrike from them last time round which basically isolates GPU benchmarking. I'd like to see an updated 3dmark11 or Vantage.

 

I really doubt they're intentionally trying to fluff people's egos. It's simply an artifact of the scoring formula Futuremark uses. A weighted harmonic mean will assign much more importance to components which deviate far below expected ratios determined by the weights. With this being a new benchmark, they're setting the expected GPU bar higher. If you were benching SLI 1080s, the proportionally lower physics scores would have a much larger impact on the final results.

 

im guessing that the older benchmarks get, the more of an impact the CPU physics score will have since the GPU bottleneck will not be existent anymore on more powerful graphics adapters...

 

dunno about you guys but is OC better with the modded drivers?

 

Not sure about the OC, but the scores definitely are

 

I'm in agreement with the "not working" nonsense surrounding this new benchmark. Hopefully it gets fixed sooner than later.

As to the age old argument about cpu and how much it affects the bench. Me and quite a few others have had this argument long long ago in the desktop forums. We would have higher gpu scores, but yet people on ln2 would come and beat us out because of how high their cpu was running. And that went for the same cpus we had plus others with more cores. This is why I have a 1680v2/5960X and not a 6700K. Overall scores win more battles than individual scores do. (And trust me. We did a LOT of arguing about that in the desktop forums.) At the end of the day it's the best score that wins.(Is what we were told. Over and over again) This is why benching is expensive.( A lot more goes with that but you get the picture) (Example: Just because your engine has 10k RPM@1000HP will not ensure a better over all time if your transmission is not of the same or higher quality build.) So to put it in a better perspective it looks like you are in line with "other" 6700K/980 setups if you overclock the GPU. We wont know for sure till that happens.
http://www.3dmark.com/compare/spy/26088/spy/7597#
I had to go well beyond normal overclocking just to even get anywhere near where you are, so a 980M is of course no match for a 980N, but then the 1680v2 kicks in and picks up the slack there.

So, when benching you can't keep looking at it as a machine, you have to look at it as parts. Gamers look at it as a machine. And in the end it will always be overall and not individual. That's why we gave up arguing it. We were flat out told to either go cold or go home, because air was just for testing. The first 3 months I went home. After that I came back and started on a mission.

 

Yup, all makes sense. I just don't understand their rationale for the overall score calculation. All I am saying is that it seems like simple math would have placed a whole lot more distance between our two scores based upon your massively more powerful CPU using just plain old common sense. Your physics score, in my mind, was largely discounted by whatever algorithm they are using for the overall score. I was actually disappointed that your overall score did not totally annihilate my score same as it did in the physics performance. That's what I wanted to see in the overall score as well. But, to your point the highest score wins... end of discussion.

Common sense doesn't always apply. It can easily give a false reading of system competence for people that don't know enough to look at the breakdown instead of the overall score. The overall score algorithm can make a piece of junk look a lot more robust to a noob than it deserves to look. The physics score if valid and has the appropriate different between the processors, but it is hardly reflected in the overall score. Just an observation, since I look at all the sub-scores, too. If nothing else guess the overall scores probably help them market turdbooks easier. The gamer-boys take one look at the overall score and are like " See! Look, my laptop runs just as good as a desktop." NOT, take another look, McFly. LOL.

I think there is a good chance the P870DM-G is possibly going to be the last high performance laptop I buy. It's an awesome machine, very glad I bought it in every way. But, when it's time to spend lots of money again I'm probably going to join you in desktop land with the most insane build I can do. I'm not liking the idea of anchoring a laptop to a desk with things like liquid cooling that render it totally worthless as a laptop, and to go any further than I am now it's going to take something more than liquid cooling to excite me. Probably will look into phase change cooling or something nuts like that. I need to stop procrastinating and just do it next go 'round.

 

Yeap, i am also going back to a desktop once i get confirmation that ill be in LA/US for at least another year or so. But will definitely hang onto the Phoenix unless absolutely necessary to let it go.

 

I'm going to try again this evening to see if the benchmark is stable enough to overclock. It was marginally stable stock and crashed 3 out of 4 runs without completing. Futuremark needs to fix whatever is broken with this public beta mess they released for end user testing far too early.

 

is it win 10 only?

 

How does the upgrade work ? If i already have advanced, and i buy an upgrade key, will i have to enter 2 keys every time i install 3D Mark ?

 

Yes, it is a DX12 test.

Yes, the 3DMark UI will have two keys. If they add more later we can probably expect to see more wallet pilfering and more keys showing in the UI down the road.

I recommend downloading a new 3DMark installation package, removing the old and doing a clean benchmark installation. Attempting to do the upgrade in place through the 3DMark UI failed several times for me. It would download but not install. I really hate the way they have changed everything. The new UI design truly sucks. It reminds me of Windows 10 for some reason. Maybe the integration of more benchmarks, including trash for phones and tablets combined with real computer benchmarks into a single "app" plus the stupid screen space hogging crap. I find it all very tacky and unprofessional, personally.

Kind of sad, but even if you have purchased the advanced version, the money grubbers charge extra for the "unlocked" DLC for Time Spy. I guess if game developers are fattening their wallets peddling DLC, Futuremark might as well feel as liberty to be leeches as well. It would be a lot cleaner if they just broke them up into individual programs and they might make more money that way as well. The fact that they are calling it DLC seems to indicate their target is gamer-boys more than overclocking enthusiasts. As their tag line suggests, it's a "gamer benchmark" by design.

The free version has to run the demo and has no access to run individual tests, looping, etc. The Time Spy unlock key code is only $4.99 so I bought it. Once they fix the bugs it will be easily worth $4.99. I go directly to the Futuermark/3DMark store, not through Steam. Running Steam in the background for benching seems like a silly idea to me... never understood the rationale for running something like a 3DMark benchmark through Steam with extra clock-stealing bloat running in the background. Maybe the Facebook gamers don't know any better. Some probably don't care about anything except playing games, which is fine if that is what floats their boats.

 

Yeah i actually grabbed the new installer and ran a test yesterday : http://www.3dmark.com/spy/21671

Thinking of purchasing the DLC today to run some OC benchmark and also going to grab 3DMark 11

The while DLC part is seriously annoying though.

Yeah i have it activated on steam as well, never used it though

 

Time Spy tested. How the calculation is done http://www.pcworld.com/article/3095...geforce-in-this-major-new-dx12-benchmark.html

 

Then it would be a physics benchmark and not a gpu benchmark. They do have tournaments for just physics from time to time. I had tried explaining this part many times before but no one seems to get it. It is not physics that matters the most it's the combined test that push you to new heights.
And in this benchmark You really need to have a high gpu over clock. They killed off physics running the benchmark after vantage. With physics enabled on the gpu a score went from 40 to 240 if you had an nvidia gpu. Amd user complained. So they killed using the gpu to enhance the score physics score. Then tessellation came out and it was vice versa. Amd was killing nvidia and all hell broke out once again. So they killed you from trying to enhance tess to get a better score. Then there was the AA drama. Amd was basically skipping along AA while nvidia was actaully forced to use AA as intended. So that also got changed over the years else AMD would still be crushing nvdia with old ass tech.

Here is a few scores i grabbed to compare.
http://www.3dmark.com/compare/spy/26088/spy/7597/spy/39782/spy/36525
Will a little finagling here and there the 6700k can pull off a win. Add ln2 against the other two and he could possible win cpu as well.

As to turd book seekers...They want everything based off the gpu as well. Who's wants the slowest component slowing you down to be calculated? And that's why you get a good idea using 3dmark11 because the speed of the cpu can help in test one (It has to be a base of about 3800 to 4200 & can can be only 1 core running. And it still wont affect the gpu score. Lower than that it will affect the score..), but the other 3 test are based off the gpu.

Benching is a crazy world to be in and simply put, one setup does not fit all. You need to have multiple setups of many different things. Hell, look how long it took people to beat a damn Celeron D cpu over clock of 8.543 ghz

All i can tell you is....You have to adapt and learn the rules of the game and how it's really played to be serious about it. We'll talk more on that when you get into that phase of things.

And as to a start up system....You will need a really good board/cpu/memory combo that can be used across many cards. As to cooling. Phase is going to run you around 1k for a new setup. And you will need to learn the prep work that goes into it and why. How to clean up for rma. How to bin or purchase a good cpu about 500 above normal cost. And so on.....All based on where you want to fit in at. And the main part about this....The only parts you are really going to keep is your cooling/cpu/motherboard/memory. And you will swap out cards each time the new ones come in. You will sell off or trade or send back to store if you didn't kill them. You will keep one decent card to tinker with games and what not and Keep One high end laptop for doing what you are doing now. And that is the process......That laptop you have now will bench quite a few old cards so If i were you i would start there....but we'll talk later though.

 

Definitely. This beast is not going anywhere. I love it and have no plans to sell it to help pay for a desktop. I still hate the idea of sitting at a desk for pleasure since I already do that 8 to 14 hours every day for work. I want to be out of my digital dungeon and into the rest of house at the end of the day, LOL.

When that day comes, the desktop will be for vanity only since overclocked benching is what I enjoy most about computers. The P870DM-G will be for that, plus everything else. So, the desktop will see very little use in comparison. That will have to be something I save and plan for because building a budget desktop system is not in the cards. It will every bit as or more expensive than any maxed out laptop would ever cost. I'm not interested in an adequate gamer desktop build with spare chump change for some of the same reasons I am not interested in a BGA turdbook. As Brother @johnksss mentioned, a phase change cooling setup will run at least a grand... a cherry-picked desktop Extreme or Xeon CPU is going to cost a whole lot more than a boxed CPU from a store shelf, and the list goes on. Having the best of everything is very expensive. The cost of high end desktop stuff pulls the rug out from under the notion that high end MXM cards are too expensive. The question is, compared to what? Anything worth buying is going to be expensive, whether desktop or mobile. Just the cost of admission if you want to play the game to win.

 

I'd be interested in a glass tubing water cooled uATX beast

 

I see what the deal is now. This benchmark doesn't play nice with crazy high GPU RAM overclocks. I can't use VRAM overclocks that are stable on everything else. Started testing different things and when I backed off a little on that it stopped crashing. Strange, since stable 3DMark 11 memory overclocks are still too high for it and that used to be the harder benchmark for high memory overclocks. Not any more I guess.

http://www.3dmark.com/spy/49380

 

Highest stable mem OC /= highest performance mem OC. GDDR5/GDDR5x error correction and/or timing changes sometimes kicks in. I get higher scores on +450mhz memory compared with +680 memory on 1080. Firestrike tends to complain about high "stable" mem as well.

As for benchmark not factoring CPUs that much, I do roughly understand their rationale. FS/3DM11/etc are all "gaming" benchmarks and CPUs more than quad cores really doesnt help that much with gaming. Pure gaming wise, I would definitely take your 6700k over my 5960x any time of the day unless for multigpu. Generally people only use those softwares to measure graphic/gaming performance anyways.

For CPU performance benchmarks, people tends to use other programs to benchmark anyways.

As for desktop, with the way intel has been pricing extreme CPUs(6950x cough cough), you probably are looking at 2k+ USD for decent chips and it looks like nvidia may exceed the 1k price point for their next titan as well. I dont want to be doom and gloom but the future for desktop looks grim as well.

OT:

@johnksss have you been playing with custom cards yet? I noticed on my FTW that I cant even close to take advantage of the full power envelope as I cant raise the voltage above 1.09v. It runs around 80% of the stock 215w limit on firestrike with voltage slider maxed out. The best I can do is +90/+450(around 2050/10900) from the ftw base clock.

And you mind if I PM you some questions about 5960x OC as well? I dont want to get too OT here .

Best I can do with FTW for now: http://www.3dmark.com/3dm/13279528?

 

Your thinking along the lines of a real desktop with case and fancy lights...Er....ah...no. That's a big 10/4 negative! What we are talking about is a benching station for a whole host of parts. That goes for any video card all the way back to 7950 gtx's and so forth. It's better to burn up cheap cards than to burn up 1k cards with in 30 minutes of ownership. So it wont matter what card is running at the end of the day. Could be just a 960 gtx, doesn't matter. Once benching is done then get back to doing other things. You will need to bench far more than current cards to catch up. So best to start with what ever comes your way. I think i found a gtx 260 laying around. That will get benched for hardware points then sent on it's way. . Also, you don't have to always buy top of the line to get points. Alot of time you get more points from hardware that has been around for a long while. This is why people keep rebenching that stuff with new cpus and motherboards. Scores seem to keep going up over there years.....

Sure, pm away.

 

Nope, none of that for me. Benching station all the way. Not sure where you got that idea since we talked about that already and you sold me on that already. That's the only way to fly, I agree with you. But, that was a long time ago. I can do phase change cooling with open air bench just fine, can't I? I'm not into the silly bling crap like fancy LEDs all over the place. I want a plain black keyboard with super bright but dimmable white backlight and that's all... not into the colored lights, and especially not if they're blinking, LOL.

 

The P870DM-G will be for that, plus everything else. So, the desktop will see very little use in comparison.

That's where I got that from. My bad

 

No problem. Sorry for the confusion.

 

No worries. All is always good with us.
It will be lots of fun to be able to do both, when time permits and all.

 

Totally agree. I'm going to have to start saving for it. Can't blow any more big wads of cash impulsively on stuff like that and nurture my retirement nest egg at the same time. Now that I don't travel 70% of the time and home 30%, having a beast laptop is not as important as it used to be. It would have been stupid to have a desktop (open air benching rig) back in those days. I'd never see it 70% of the year. It took me a couple of years to get used to being home 90% of the time, felt like I was going to get in trouble playing hookie from work or something. But, now that I am I don't want to travel for anything now, LOL. Jean or shorts, t-shirt, barefoot and 10 second commute to the office 5 days a week is pretty hard to beat.

 

That's just it. You dont have to start expensive. Get a board for the 6700K you already have. Grab some ram a cheap ssd and maybe a aio water cooler for now. Those things work wonders. And if you only have one cpu then grab a cheapy one to swap back and forth. That way you can get a feel for the desktop benching gain without going broke in the process.

 

how would one classify an overclock of +134/+300/+100?

mild, medium, or heavy?

i did see my laptop peak into 300 watts with doom running uncapped with the above OC where the power is set to max performance.

 

Light, with a slight tendency to medium!
Is this +100 mv you raised? If yes, it seems to be pretty much for your oc. Did you try lowering the mv in 12,5 steps and see how far you come with above mentioned settings?

 

core/mem/core-volt

EDIT: if you run 3dmark and see occasional flickering, what does that mean?

 

I'm really no expert, but that indicates that your oc is not rock stable imo. It should not flicker.

 

it looks like it was a memory OC. I tested +150/+495/+100 and saw occasional flicker. i then tested +150/+300/+100 and there's no flicker to be had in firestrike.

+150/+495/+100 - http://www.3dmark.com/3dm/13291634?
+150/+300/+100 - http://www.3dmark.com/3dm/13292110?

215 point difference....not sure how much of a big deal 215 points is.

 

in this case its a 1.7% deal ^^

Sent from my Huawei Mate 8 NXT-AL10

 

ah... so pushing the mem an extra 195 doesn't do much. will see if +150/+300/+100 will do as a mild-medium OC.

 

Thanks for this, Brother @Papusan . That explains a lot. 15% weight. As Brother @johnksss has pointed out, the best score wins... end of discussion. But, it's still kind of sad to see so little credit given to those rocking far more robust hardware that dwarfs the performance of anemic little turd CPUs. I'd like to see a new Futuremark product like an unpdated DX11/DX12 version of Vantage with truly brutal CPU testing, heavily weighted, that spotlights the shortcomings of pathetic little BGA turdbooks. What would be really nice is a Futuremark "Hall of Shame" listing highlighting the worst performance CPUs, where being #1 means you're an example of the worst the industry has to offer. Then the professional review shills would have a new tool that we could all sink our teeth into.

Sweet! Looks like SLI is scaling nicely.

Yes, severe flickering, artifacts and screen tearing can be an indication of an unstable overclock, or pushing the GPU hard enough to cause graphical errors. But, many of these benchmarks have some occasional flickering and minor artifacts even running stock. So, don't be surprised to see some minor evidence of that from time-to-time with no overclock. I've also seen it change for better or worse going from one driver version to another.

 

the flicker is like 1-2 frames....practically a blink and you'll miss it kind of thing.

 

You will find this varies a lot by benchmark. Some thrive on memory overclocking and others it makes very little difference.

Yeah, that's totally normal. Really no different that running a game and seeing an occasional graphical glitch.

 

i'm giving +150/+300/+100 a go as an OC and i have some games set to Max Perf. and the behaviour is different - doom uncapped on vulkan runs really high and peaks at 310 watts of power usage and stays around 300.

 

Vulkan does remove some bottlenecks you may have had in direct x.

 

Nice! I love Doom, too. Just be careful playing games overclocked. I run game benchmarks overclocked, but extended gaming like that can be hazardous to your GPU. Remember, we only have a sensor for core temps and you could easily be cooking other components without knowing about it. I play games running stock clocks on GPU for that reason.