M17x Benchmarks thread !!!

If it works it works, at least that is your obsticle and it can be overcome with pin mods too

the W90 issue is the locked bios, no ram divider or multi's for us to change.

I consider the W90 to have the best cooling I have ever seen though, the M17X is advertised with a very good cooling solution but you guys need to proove it.

Put all 4 cores of your cpu under a 100% load under a stress test (orthos, prime95, wprime) keep it there until temps go no higher.

Thats the first thing you need to do before you start increasing voltage, as a raise in volts is going to give a substantial increase in heat.

 

yup that's cause that's intel recommended max volts for these cpu's.

 

occt made my gpu go threw the roof. i hit 91c stock!!!!!.(i was in a very hot room)

my cpu also hit 70c!!!!!!!!!

i used it for 12 minutes each.

 

My hottest temp in any of the tests I've done for both gpu and cpu was 74c.

 

really. i'll check out mines when i can(i'm compressing a 8gb file right now)

in normal conditions. i was baking in the room i was in when i ran occt

 

Yeah dont use wprime i dont think, it may use 100% cpu but not heat the components fully.


OCCT is the deadliest from what i've seen, its great for identifing possible gaps in the cooling.

Although in terms of usage, WPRIME is most realistic

 

TBH I'm using OCCT for more than 3 years and i only saw one thing that demands more from your system than OCCT...

Compiling XORG under *nix system lol.

 

I think I said my cpu never hit 70c, my mistake lol it never hit 60c. I got confused with gpu temps.

I'll start another run now gonna do the Wprime large run and record the highest temp it hits before its finished.

Update so far 25% done max temp is 50c on core 3 & core 4 before going back down to 49c. This is on my desk at work with no cooler, overclocked to 2.7ghz.

Wonder if my ICD7 set in and got better as I remember it being a tad higher than this last time.

Ambient temp is 70F

Update 2

41% done with the large test.

Core1 46c
Core2 44c
Core3 49c
Core4 49c

Highest temp still 50c on core 3 & core 4 then drops right back down to 49c.

^^ thats good cooling.

 

OCCT is not reading the GPUs correctly. If I have the 9400M it reads that card. When I turn it off it reads just one of the 280M. The program says it gets it's info from CPU-z. Is there anyway to change that?

 

Just run hwmonitor prehaps mandrake, and leave occt run, or everest.

Did you ever run an OCCT test visc?

 

As long as it stresses the right parts I wouldnt worry about it.

Most programs call my gpu's Meow 1 (not kidding)

 

Yeah, adding the temp readings of the second card shows both cards getting hit. So far seems to be topping out at 72c.

Edit: Just hit 74c...75c, new record high for me. Time to shut down and go to work.

 

Nope but 100% load on the cpu core is 100% as far as I am concerned the Wprime run I am doing right now has all 4 cores under load and I like how it is timed

Its 67% done now and highest temp I have seen from the cpu is 51c for a split second before it drops back down to 49c.

I'll try OCCT I guess just to see if its different, I also have winamp and a full screen visualization running in the background, you will see that in my screen shots.

So when this wprime is done I'll go get OCCT and run that, then I need to work on my website or do some actual work.

 

have any of you guys done a bench marks with the new drivers if it improves anything?

 

Haha, yeah i should be working, but im sat on these damned addictive forums. As you say 100% is 100%, ive never seen anything run components like OCCT to be honest so it is a bit of a silly test.

Do you also have the temperatures of your northbridge?

 

No just cpu & hdd temps.

I still have the thermal pad on the NB, didnt have enough ICD to put it on there when I was redoing the TIM, also it was a VERY big pad.

 

We tried the new Dell drivers yesterday and they are an improvement over the older Dell drivers. It now gives you an option to turn off/on Pyhsx and performance is better with the 9400M on. For bechmarking the Dox drivers are still proving to be better.

 

Hmm, it will be interesting to see if you stutter during OCCT test visc, probably not but if so your northbridge is prob hitting around 90C under super load lol, northbridges sure get hot sometimes.

 

Ok OCCT going now 15 minute test medium data set.

took just over 30min for the large Wprime test to finish it never went above 51c, but also it reached 50c in less than a minute so if OCCT is going to be hotter I will know soon.

Edit: 4 min deep and yeah it hit 57c then dropped down. So it is hotter, but I also turned off EVERY program except OCCT to make sure it gets full priority.

Guess I'll try a 15min GPU test also, ambient is up to 72c now instead of the 70c when I did wprime.

Edit: 5 min deep 60c max, here is the amazing thing, no music or sound in this room and I barly hear the W90 at all, its still silent. That means it has not managed to even kick my fans into high mode.

Edit 8 min deep no higher than 60c still, drops down to 57ish often, this is core 3 & 4 the hot ones. Core 1 & 2 sit around 55c

Will OCCT give me a temp graph and test details I can post when it is done?

 

51C...Godly, the q9000 in my machine hits 66 peek hehe, thats ok for me but your temps are fantastic.

 

yeah my north bridges is hotter then my cpu but the funny thing is my northbridge and southbrigde has a dedicated fan. i may copper mod everything but it's harder to dissemble then my whitebook. my cooling i think is fine but i'm looking for ways to improve it and i can't find any.

have you seen any ways to improve cooling for the m17x or the w90 or the gt725?

sorry for going off topic but i would like to see how the cooling is made for your systems. or the insides of it.
mines is the oldest so here's a link with it striped
http://www.anandtech.com/Gallery.aspx?id=194

 

Ram heatsinks do the trick, i was thinking about north bridge settings, but i dont know where to start (as i have access to north bridge settings in the bios)

The single fan can keep the northbridge at 72C under 100% standard load which is ok, but under OCCT the last stage of MSI fan is revealed (i think over 2x the RPM ) and northbridge sits at 90C and WONT budge which im so thankful about.

Stock paste needs replacing, ICD7 i thinks officialy now the best and low risk too!

You could try a big enough pad to envelop the northbridge (unless thatl suffocate it....then i fail) or a copper sheet bigger then the northbridge so that it can disperse the heat a bit better.

I will take my system appart when i get in and give you a screen, the northbridge has a single heat pipe

 

@ 2.7ghz too

But I dont think clock speed has much effect on temps in my experience, its more about the voltage.

OCCT done in 1:20 its monitoring right now so I guess it produces a cool graph for me at the end.

M17X owners who want to match the test.

I did the stock stuff.

CPU >> OCCT >> Medium Data Set > Custom time 15 minutes.

This is on a flat desk, with no cooler.

From W90 Benchmarks

From W90 Benchmarks

From W90 Benchmarks

From W90 Benchmarks

 

Yeah alot of the time it is about voltages really, i run the power test (cant remember the name) that uses everything at the same time, tis a real burner.

Flippin eck visc great temps.

If you get time you should try the power test, apparently it identified power problems in the desktop 4870, have you ever opened the w90, ASUS cooling has surpassed everything ive ever seen with notebook cooling

 

Ok almost done did 15 min OCCT GPU test

> Full Screen
> 1920x1080

In my experience SLI or Crossfire do not work unless its full screen, and I used my max resolution to make sure the cards had a high load.

I could hear the fans on this test much more so than the cpu.

The cpu fan was almost virtually silent, the gpu fans made a nice "whoosh" of air noise but nothing distracting and nothing you would even hear over your game sounds.

I dont see my gpu on the monitoring tab though so I think maybe I ran the test for no reason and wont get a temp graph only program I know that shows my card temps is the ATI CCC.

 

I know it's a lot to ask but could anyone with an overclocked CPU run OCCT high priority, Prime 2004 priority 10 or CPU Stress MT (for linux) (priority makes a huge difference in temps) for at least 8 hours? 1h isn't enough to reach max temp. I know we almost never use our CPUs that much but at least we're sure it's stable at 100%. IMHO it's the only way to say the CPU is rock stable. I remember having to run these 24h to validate a stable OC WR.

If no one wants to do it I would understand since you probably won't be able to move the cursor for as long as the test runs but I doubt i'll be able to run it myself for a while since dell seems slow to ship this laptop to europe

 

Priority only makes a difference if your running other programs, also in all my tests I reached the max temp in under 5 min.

 

What about the most up to date hwmonitor? I had this issue with the 4850 but hwmonitor now sees it, if not i can get you the link to i think an older hwmonitor that reads GPU's easier.

 

occt will bring the heat out. cpustabtest i think is also good. prime95 is for someone who could wait. and i know that's not me.

my gpu was very hot using occt. i want to see your temps.

 

GPU gets hotter in furmark to be honest, but then again it may be just the type of test i did (the power one)

 

i never ran furmark. i will have to try that.

 

Its a very good test, and supplies monitoring data, it buuuurns GPU

 

literally in some cases

 

Haha, probably worse on a defective 8 serie nvidia GPU. (Nuclear particle generator lol)

 

Just got the new version, it works only shows GPU core, no NB temps still.

If it maps graphs or logs it will work for doing full screen stress test temp monitoring.

Both GPU 55c idle.

edit: oh I see only does min/max.

Well that will work. I'll do a test tomorrow night and see what the max is, for today I have to pack up and go home.

 

Look at these 2 graphs, 1st one at normal priority and second one high priority.
Note that the test only ran for 1 min. Priority definitely makes a difference, on a normal duration test the difference will be even higher.




I do not have screenshots of tests failing after 6+ hours and I don't have time to run tests and expect a failure but believe me I've seen plenty.

If we look at your first graph, we can see that, at lest the 1st core isn't at it's max temp, past some point the cpu can gain 1°C in 30min but it'll continue to rise.

PS: don't pay attention to the CPU

 

Depends if you are running tests after each other, they would both have to start off with the same ambient temp on all components to be truly accurate.

Everest may show your north bridge temps visc, there was a free everest ultimate offer going around for a while, not sure if its still there but who knows, will try dig up a link.

 

actually i have mine stable at 3.06 now.

 

I did wait for the components to be at the same temp you can see the temp rising faster on the 2nd graph anyway so even if I wouldn't have started at the same temps you would be able to see the difference. If you're still not convinced I could run a longer test which will show a bigger difference but... Anyway I'm not really concerned by the temps (I could fit my DOD on it ) but more the stability (since they are tightly linked I mentionned the temps too) and the only way to validate an OC is to run a test for more than 8 hours at high priority.

I know it's too late since I already ordered this laptop but I would like to see if it's a good overclocking platform before receiving it

 

after seeing some of these temp readings, i'm starting to worry about my temps, using occt my cpu core temps will reach 80c+. I idle around 46 to 50c. These temps make no difference whether i'm stock or oc'd really.

 

Those look VERY close to me, I dont see this BIG difference you speak of, infact there is a larger difference there when the cpu load is gone when the test has no bearing on temps at all.

As long as you have no other process taking cpu power from OCCT all priority modes will work the same.

 

I agree that on these graphs there is no BIG difference but it's because the burn only lasted one minute. The difference when monitoring after the test is because I was using another software to generate graphs, I didn't during the first test (on a crappy CPU like this one even the lightest program makes a difference. The reason why priority makes a difference is because while sharing the load, windows wastes some CPU power so you're not using it at 100%, ok it's at 99,9% so not much different but enough to change stability results. Anyway I'll test it myself when I receive it so it doesn't really matter

 

True, but bear in mind the surface area of the fan intake also warms up (in most cases) and that can heat it up also, i dont think priorities make any difference at all. My cpu boils with this test but i will try some runs when i get in on different locations of the same surface area at the same temp.

The cpu to temp ratio shows that its the usage that decides the temp, looking at the graph they would both be the same temp.

 

Mmm I didn't think of this variable, I never had an overclockable laptop but it is a very important point...

I found an old screenshot of a previous OCCT version which used to display an error when the OCCT process CPU usage was too low, I don't know if current versions have the same error but it says that if the CPU usage of OCCT is under 90% the OC can't be considered as stable even if the overall usage is at 100%.

 

The most accurate test would be the voltage one, as long as components are used and at full power, with the maximum voltage, i class that as stable, unfortunately for me stock cooling is not handling the temps well enough for me to feel safe with, so i think 7 carrot icd7 will be great

Airflows nice to have, without it your done for hehe, the main variable in your temp is the cpu usage times, they are not exact for each minute you took, you could wprime it with both priorities, run prime32 and then you can get even the miliseconds?? lol to see if high priority and standard priority is the same.

 

I've been trying to read what some of you guys have been doing over in the W90 thread but it's not easy reading similar to this thread. The current GPU Vantage scores are in the 11.4K range? What about the 3dmark06 scores? Has the W90 hit 17K yet? It would be nice to have some type of gaming system thread that shows the scores of multiple systems to see how they compare.

 

Someone told me on youtube that it broke 18k before but i cant be too sure of that, well maybe with qx prehaps and a really hard vbios flash.

I know ATI cards can be pushed...hard.

 

::Facepalm:: Again with 3dmark06?

It is just a cpu speed test.

You can get 17k+ easy with a dual core with high clocks.

Vantage is the only one to be concerned with. I have gotten just over 11,000 for gpu score, another user has like 11,400 but thats our limit. There are M17X scores higher in vantage. Though less than 300 points is not a large margin, I am also concerned a bit with the fact I think the higher scores from the M17X are with some modified drivers. Its very possible your just killing some of the IQ off for speed as many custom drivers do.

Not modified drivers, like just making the desktop driver set work on a laptop, thats ok. I mean modified where drivers magically give more performance. There is no "magic" in this really.

 

well the 4series can in your notebooks. my whitebook was driven by my quad. but I just pulled apart my computer and whao!!!!!!!

there was a lot of dust. I bought mines used but Inever knew so much dust can be in a laptop. I am cleaning it and reappling thermal paste.

btw what paste do you all recommend. I have as5 and mx-2 chilling next to me. but I know there's another one I just don't know the name of it. also I was reading the thicker the copper, the better it will take out the cooling. the m1330 copper mod saved 5 degrees on the gpu and the eee cpu copper mod saved 5-15 degree. so I am looking at what I can do to improve my cooling.

I just checked out pictures of the w90, and I see loads of ways to improve cooling if you want to know let me know and I can put it up on the w90 thread. but vic I love your temps a lot. I was going to buy the w90 also. I ws stuck between the np9262, w90, m17x, and the np9850. but I was afraid of the reliablity of the manchine. I was on newegg's website about 5 days on the add to cart because I wanted to stay with ati and I heard the cooling was great but I ddn't see anything I want to know about the manchine. lol.

as for the m17x I would like to see it opened and the different test to see if I like the cooling on it also.

I wll judge mines better when I clean it up.

 

Yeah there is no hiding the W90 had some reliability issues, and lost many customers due to it.

I myself had to RMA my first unit, but the end result was worth it.

Its a unit that looks great, performs great, and didnt cost too much. Also I have that outstanding warranty for 2 years to keep it that way.

Just hang out longer, word on the street is I am getting a 9850 for review, and once we have hands on testing for all 3 units you can pick the one you want.

I would been worse for Asus if the M17X was a flawless launch but it also had many quality issues, and I am not sure if they are all resolved yet.

Sager is the only one with a clean sleight, can they release a unit with none of the problems the other 2 have? If they can it will be good for there business.

W90 Issues

> Random 2D Clock GPU issues resulting in screen errors or crash
> Hard to update drivers

M17X Issues

> Screen backlight bleed
> Sub par overclocking
> Missing hdd brackets

So Sager needs a unit that wont have that stuff, already the panel wont be an issue, its probably using the same one the W90 has, and it wont have the gpu issues the W90 had because it uses the GPU's the M17X has.

On the M17X problem side, the 9850 holds 3 hard drives, I doubt they wont include the mounting hardware, so this really only leaves the overclocking and temps on the fense.

If the 9850 has really good overclocking & cooling, it could be the machine to get.

I will still prefer the W90 due to my budget, to get a M17X or 9580 it would cost me $1000 or so more, nearly 50% increase in cost, and no way am I getting 50% boost in performance, the units are basically dead even.