M17xR3 First Impressions and Review-DaneGRClose

I'm sure I will, the thing that makes me believe it's not a PhysX situation is this:
Mandrake and Dr. J both reported roughly the same scores, and even enabling PhysX to the default in the control panel didn't change anything. So it's either something within the settings or driver or I got a bad GPU really hoping it's the former.

 

The cpu effects the SM2 and SM3 scores in 06. Is you power plan different from theirs? Try it with the power plan on max performance

 

All of the runs were on "high performance" which I believe is the highest power option without tweaking registry's or advanced power plan options. The cpu is running amazing but the gpu seems to be the culprit, I though SM2 and SM3 were primarily GPU scores scook?

 

They are not so clearly separated in 06. They are the GPU scores but they are heavily influenced by the CPU power.

 

Ok, well once I have some 3dmark06 scores and vantage scores to compare to I think I should be able to narrow it down some more, until then I'll mess with the control panel settings and driver versions to see if I can figure it out. Thanks for the clarification scook +1 when I'm able to rep again.

 

Well it looks like my score was something to do with the Nvidia Control Panel settings, reran the test and pulled off 14.1k with PhysX disabled at stock:

 

@Dane - Congrats sir! Outstanding! Glad you got it sorted. Maybe it was an AA thing or something. What about the Tessellation settings?

 

Don't know what it was for sure, gonna have to do some slow tweaking now, strange part is at high overclocks it doesn't seem to make a difference which is even more confusing

 

Seems something's up with my settings too



Downloading 06 now

 

and here's my 06




I'm not really into benching, so I wasn't sure how to turn off physicx, but tried to make sure it wasn't using intel.
I noticed that my GPU-Z screen shows less gpu options than yours (CUDA, etc). sup with that?
EDIT: And my cpu is oc'd...? I'm pretty sure (read: 100% certain) that I didn't adjust my bus.

 

looks like phsyX is on, in your benches.

 

Here...



Edit: looks like you got it fixed.

 

do you see any signs of throttling, a few people posted problems but none has posted any usable info.

 

Looks like some decent OC headroom on the 460M.

Any low hanging fruit for motherboard OC'ing?

 

I am getting:
3DMark Score- 12756.0 3DMarks
SM2.0 Score- 4919.0
HDR/SM3.0 Score- 5322.0
CPU Score- 5001.0

I cant seem to break 13k either, What exactly did u do to fix it? It looks like my SM2.0 score and SM3.0 score are way lower then they should be.

 

Did you have PhysX enabled/disabled.

Dane, did you see any difference with PhysX on or off? Does 3DMark06 utilize it?

 

3DMark06 can utilize PhysX yes, if you set it to the GPU it will definitely screw up the score and make it much higher, if you set it to Auto-Select it may screw up the score. You have to have PhysX set to CPU in the Nvidia control panel for a true score. My 14.1k score was with PhysX set to CPU(off) so the score is off as well. What drivers are you running? Did you get the 3D version? Are all of your Nvidia control panel settings other than PhysX set at default settings?

 

EVERYTHING stock and at default. Did you try with PhysX on, off and default?

 

Yeah I did, only when PhysX was on(set manually to GPU) did it skew my score. I know what yours was Dr J I was asking MJOzlins.

 

FYI, I had to manually set the global default to nvidia GPU in the control panel (essentially disabling Optimus) to get proper performance. discreet kept powering down and switching over the IGP, which would then switch back to discreet when load was too high, which would then switch to IGP when operation was completed.....etc

 

Well that is not a surprise. You will have to choose between performance or battery life, you can't have both

 

How is it not surprising?

Optimus runs perfectly fine on my L501X. I've never had it switch between GPUs during a program's operation - with the exception of web browsers (when loading flash video). The claims of "seamless switching" are (usually) not over-exaggerated. It's not supposed to be doing what I described.

 

It is seamless switching, but is your performance on the discrete as high under optimus as when it is set to discrete only (optimus off)?

In the R3 it is still working and switching as needed, but the peak performance is not as good

 

There's no difference in the applications that I use @ xps between discreet only and optimus. The issue in the r3 is that it's switching to integrated when GPU load gets low, even with a specific optimus application profile and right click -> run with nvidia

 

I see. I have no experience with optimus first hand but have heard that it is far from a perfect technology at the moment. I have an M11x R1 and an Asus UL80Vt that both have Intel/Nvidia Switchable graphics (sort of optimus beta) and like it for the most part so far.

 

The big difference is the fact that the IGP in the R3 is a lot stronger than the one in your XPS so the power/performance level may be seen as high enough for some portions of the tests even though it won't net as high of a score. The whitelist is something you always want to tweak for each application so I don't see where the problem lies here.

Edit: And Scook while Optimus is a good idea on paper, it's still a buggy beast in real life, you're not missing a ton. I'd still take a good old fashioned manually switching setup over Optimus if I had the choice

 

yeah, I guess.
There's a power management mode that is set to adaptive (which i guess allows dynamic switching?) by default in global settings, and each program's profile is set to use global setting by default. The problem comes when you can't edit profiles in control panel - which is more common than it should be.

If anyone knows where nvidia CP stores the profiles then I could probably edit them manually. That'd save me some battery life and I would be greatful!

 

No idea on that portion kunekaden, 3 people I can think of off hand that may know though are Nautis, BatBoy, and NVPelly might be able to give you an answer on that

 

Dane, thx for the reply sry I had classes lol (maybe I should just bring it with me).

I have the same specs as u /w the 3D package. I tried all 3 physx setting and they all netted abt the same score. So I guess ill set all the nvidia cntrl panel settings to 'reset to default' and try again. Thats what your saying to do right?

Drivers are stock from Dell that shipped with it.

 

Yes just default everything in the Nvidia control panel to factory/global default on all pages, make sure that the 3D is not enabled, make sure that PhysX is set to CPU and try it again. Another thing I'd recommend if you know much about drivers is to lose the Dell driver and go with the 266.66 driver modded from laptopvideo2go. Let us know how the next runs go.

 

Stayed with the Dell driver but defaulted everything in the nvidia control panel and got:

3DMark Score
14071.0 3DMarks
SM2.0 Score 5797.0
HDR/SM3.0 Score 5701.0
CPU Score 5031.0

So I guess that fixed it lol (physx off also). Thanks.

So it would seem that the 3D settings or something screw with the 3D Mark 06 scores? Its odd I wonder if it actually effects in game performance also.

 

3D settings effect the performance of the card and therefore the 3DMark score by extension

 

Yeah I think it was the 3D settings that screwed mine up as well, that's about where you should be stock.

 

hows it going on the benching front? wish this sb thing gets sorted out fast, been kinda lame without a lappy.

 

Nice, thorough review Dane, thanks for all the effort and posts!

+1 Rep

 

inap- haven't had much time after the inital review to try to push it, I think I reached the roof on stock paste/cooling at about 850 core, I haven't reached the roof of the card on ram at 1500 but no OC program I've used will allow me to go higher. Haven't broke 76 degrees at those clocks in benchmarks or games so it runs nice and cool even on stock paste.

Docsteel- You're welcome, if there's anything else you'd like to see let me know.

 

There's something I'd like to see actually
PM me if interested, it's a long explanation.

 

Dane,

Is there any performance difference or advantage to correcting the bios drive setting from raid to ACTH for single drive users?

What are the easy OC settings in the bios. Saw one setting for the turbo time but could not adjust it.

Thx

 

Not that I'm aware of Dr. J, can't be certain as I didn't really use my HDD in either mode for anything other than creating a factory install restore drive. I would think that the only way RAID would make a difference is if you were actually running a 2 disk RAID 0 array. The OC settings for the CPU in the bios were for Turbo duration(both long and short), TDP wattage settings, and BCLK setting. SB CPU's aren't exactly known for BCLK overclocking so I haven't messed with it much and the turbo mode settings didn't show much of a change either so it may just be something dormant waiting for a later addition of the extreme SB CPU.

 

To quote Randy Jackson from American Idol, "Yo dawg," can you check to see how many fps you get on SC2 on ULTRA/Max everything on 1080p non-3d on the map "A Sinister Turn" of the campaign mode? It's mission #11 I think.

EDIT: Can you instead benchmark (get the average fps) the final mission where we fight against Kerrigan on ULTRA/MAX everything?

Thx bud!

 

Dookie11- I can definitely do that for you. It will probably be tomorrow(Thursday night) that I do it as I just noticed this and it's getting a bit late now. If you have any other requests (reasonable ones ) let me know

*UPDATE* I've added some info on the general 3D experience/settings to the OP along with battery life updates and temperature updates.

 

Goody! Can't wait! Reminder: just to the last mission where we fight vs. Kerrigan. If you can, can you post a video with ctrl+alt+F (to show the FPS in-game) also?

Thx for both the review, the video (hopefully), and for taking the time to kick Kerrigan's arsenic

 

thanks for the updates danegrclose! so you've managed to squeeze out 3:53 now - makes me wonder what they were doing to get claims over 4hrs on the 120Hz 3D model...

 

You bet! Yeah I have no idea vikingrinn, I had to really mess with things to get that amount of time. Over 4 hrs would be nothing short of impressive.

 

DangerClose, thanks for the review. Btw, could you please describe some more details on the 120Hz display in normal use (web, office, etc.) in comparison with the RGB on R2? I mean the sharpness, brightness, contrast, color, etc...

 

You're welcome. Never owned an RGB LED myself, I've owned an R2 but it was 900p, I have seen a few RGB's and own an RGB tv but never used the M17x RGB LED long term so can't do a very good comparison. I can give you my thoughts based on the screens I have used though if that works.

 

Just wondering, are u going to keep your R3?

So far I love mine (would like a 485m later if it happens) but am obviously concerned about the chipset. If theyre going to install new Mobos in-home for us do you think its much of a concern? Dont have much experience with Dell in-home repair techs / laptop mobo replacements.

thanks (Dont forget to do an OC guide later if your keeping it lol)

 

oc guide part 1 - cpu: you can't oc the cpu

oc guide part 2 - gpu: I ran a 4 hour render after accidentally cranking my gpu up to max settings in afterburner and the temps didn't go beyond 65 and there were no artifacts, so I'm assuming it's stable.

oc guide part 3 - ram: bios options don't work consistently (at all?) so...also can't oc ram. supports xmp though, which is nice. should eventually be able to adjust voltage and manually set timings.

 

Can't you adjust the turbo time in the bios? What was your 3DMark06 difference before and after your max GPU settings? thx

 

No, you can not adjust turbo (at least not in an upward direction)
You're supposed to be able to, along with many other features, but they just do not work...(yet?)

and i didn't run any benchmarks with the gpu oc'd; like I said, it was accidentally maxed out. I'm just glad it didn't melt.