*OFFICIAL* M17x Benchmark Thread - Part 3

still waiting for a pic where you apply the thermal paste on the gpus..

 

You have to remove the heat sink. It's pretty simple man just unscrew the heatsink from it, and you'll see the nVidia chip underneath it. A square shiny chip; it should be covered in ugly blotchy thermal paste stock from Dell Alienware factory. You clean that off with rubbing alcohol and stick your own paste on it. If using a ceramic or silver based paste, you don't need a copious amount. Just enough to fill in imperfections within the contact spots.

 

I have ATIs so no nvidia chip. still need a pic. dont want to screw up my gpus.

 

Sorry didn't read your sig...

 

Sorry for not responding. I just got a new phone and if you are a phone person you would understand. But I get out of work part tonight but I will see what I can do about a picture for you. Lease just give Me alittle bit to get over this phone issue.

I'm typing on this phone and ordain nice.
Android for life

 

Off topic, but Nexus One??? how is it?

 

no. i didnt get that. i got the htc hero from sprint with android 1.5. and omg. its much better then my palm pre or my mom's iphone 3g.

if you are techy, android is the phone for you. and if you like gui get it with htc sense......

http://www.htc.com/us/products/hero-sprint

i'm not getting the nexus one any more. i'm going to stay with my company and get this
http://www.engadget.com/2010/03/23/htc-evo-4g-is-sprints-android-powered-knight-in-superphone-armo/
http://www.phonescoop.com/news/item.php?n=5713

it comes out 3rd Q of 2010

 

What about the memory chips? Do you leave their pads or replace with a paste as well? You know, the paste on those can create a mess and leaving the pads will only lower the core GPU temps. But the mem temp is your most problematic point, since it tends to overheat more and dies faster.
I see a dilemma here.
Would probably just replace all the thermal pads with a better quality ones.

 

You can replace those as well. It's not going to make much difference. Just don't make the beginner mistake of thinking that because they're not metal, they don't diffuse heat properly.

 

Replace what with what? And heck, it's gonna be a huge difference if you just leave the stock pads and replace the paste on the core, or put some paste on the memory chips as well.
Correct me if I'm wrong but since you can't monitor the 280s' mem temps (unlike the 4870 that have 3 sensors). You will be oblivious to what's going on on your memory chips untill the GPUs die.
So be careful.

 

since the mem heatsinks is not snug on top of it like the gpu die, that gap will make a pretty high impedance for heat if you use paste - thus the use of the pads.....the stock pads work fine but if you really care, you can get new ones(unsure if actually better) online from ebay/china or fronzecpu

 

Replace the thermal pads with aftermarket ones. You wouldn't put paste on the memory chips because the thermal pads are durable and shape around them. Paste is just to fill in imperfections in the metallic contact points between chip and heatsink. It is NOT a glue. You want to maximize surface area contact potential so that the most amount of heat is dissipated, and with regards to the black memory chips, those have enough contact with the pads.

So I'm not really sure what you're saying.

 

Sorry I thought you meant replacing the pads with a paste.
Disregard my last post as I agree with you 100%.
On a side note: I see a general problem with the Nvidia GPUs, lack of the mem sensors. So the logic is fairly simple, if you lower the core temps, how much does the memory temp change?
AFAIK, replacing the thermal pads with higher quality pads may not provide more than a few degrees change. When running your benches and seeing the nice 75C on the core, we think - cool! My temps are low, but the mem maybe running at 90C+ and even more.

 

what harm could there be in flashing the 280m to a 285m?

 

One example:

Le-Gardrobiere wrote:
"
Hi !

I know it`s not the correct thread but i would'nt open a new one ...

I flashed my VGA GTX 280M to GTX285M

after that the DVI & HDMI port did'nt work

driver: original from sager for the gtx285m, laptopvideo2go, ...

could anybody held me with the correct GTX285M BIOS for the W870CU or the original GTX 280M for fixing my portproblem

thanks in advance


i cant receive board-messages > pls upload it "

 

Probably better to just edit my BIOS and re-flash that instead.

I gave up on the 4870s BTW. My in-game stuttering was driving me nuts - it seemed like thermal throttling. I never had this issue with the Nvidias. I put the 280ms back in and I don't have the issue any more.

Good thing Dell did not get on me to send the old cards back.

 

Wow, good for you, m8! I have some regrets but generally find the switch more or less equal.
BTW, Dell might still ask for the cards, so get ready.

 

Yeah, I know they will. I just mean that they did not rush me, which was nice.

 

Yeah, I know what you mean. Was holding on the R1 for too long and got a phone call from them. Embarrassing.

 

hm the first to swap back to GTX280s. regarding the absence of Dell support for ATIs crappy driver i guess i will do the step too. i also found some info finally showing that GTX280s are finally faster than 4870s like this

http://answers.yahoo.com/question/index?qid=20090923060349AAO9ZVG

http://www.notebookcheck.net/Nvidia-GeForce-GTX-280M-Review.14565.0.html

i dont know why it performs better in benchmarks (maybe due to the absence of physx) but in games the performance is the same or worse than gtx280s (rarely better), even though the price of 4870s is higher. I think Dell forgot that they are finally DDR3 and not DDR5.

however i dont have stuttering in games with stock clocks. everything more than that cause stuttering sooner or later.

 

My M17x has the following configuration:

280m GTX SLI cards
Q9100 quad core at 2.26Ghz
6GB of RAM

Nothing is overclocked.

I ran 3dMark06 with the most recent dell beta graphics driver (fix for the DCP problem) and with the most recent nvidia notebook beta driver. Both times my scores are in the 11000s.

Looking here, http://www.notebookcheck.net/NVIDIA-GeForce-GTX-280M-SLI.17191.0.html, it says the average score for my cards is a 15000.

Is my score abnormally low?

I tested 3 times:

(Reports my cards as 180s for some reason, definitely 280s though.)

Thanks

 

3dmark06 is very cpu based. so the answer is yes and no. for the qx9300 at 3.199 i got around 16200 the most on win7 if i remember good.

 

Windows xp gives the best scores for 06

 

In Far Cry 2 DX10 mode the 280ms are only returning a 35 fps score. I seem to remember getting high 40s with a dual core.

They get mid/low 50s in DX 9.

Radeons actually speed up in DX 10.

Anyone else with FC2 and 280ms want to try a DX 10 benchmark for me?

Settings:
DX 10, 1920*1200, No AA, Real Tree Very High, Fire, Very High, Physics Medium, Vegetation Very High, Shading Ultra High, Terrain Ultra High, Geometry Ultra High, Post High, Textures Ultra High, Shadows Ultra High, Ambient High, HDR On, Bloom On

 

There is something a bit off with your scores, Lewdvig.

For me, generally speaking, FarCry 2 performance between DX9 and DX10 is virtually the same. I might get a few--very few--extra frame rates with DX10, but since I am not demented about splitting hair over such a small difference, I prefer to remain rational and declare 'equal performance'.

You should be getting at least 45-50. I was getting better with my 9800m GTX SLi in the M1730. Also, with a quadcore, this performance only gets that much better. FarCry 2 thrives on a good CPU. I believe Don with an overclocked QX9300 got some very good results when he had his 280s. (Somewhere in the upper 60s for average with maximum rates going over a 100 FPS).

I'd try the benchmark, but at the moment, it's not possible.

 

Yeah, I had the same experience up until today! Nvidia scores were always the same in DX9 or 10. Now there is a huge gap - clearly something is borked but I am not going to worry too much about it.

Most the people posting impressive scores are with CPU and GPU overclocks. I have next to zero stability in games past 2.8 Ghz and I find that OCing GPUs results in problems unless you are doing short bench runs.

[update] I installed the Dell Beta driver and the installation fixed the problem. I did two clean installs of 195.62 and the problem persisted (never had this problem in the past with 195.62). Its a mystery, but its fixed now.

 

I agree with you on the overclocking. Even though I have done it in the past--and I do believe our Nvidia and ATi cards have plenty of head-room to survive overclocking with the M17x's cooler design--I am still reserving this tactic for when the majority of games become too demanding. Ditto for CPU overclocking. I don't employ such techniques just to gain a feeble increase in frame-rates.

Getting a bit off-track, however. What I had to add was that Don's score in FarCry2 was from only a CPU overclock with GPUs at stock. Ditto for my scores on the M1730. So yeah, it was probably a driver-related thing as you've already diagnosed.


I myself am having some driver issues while running Metro 2033. Whenever I switch to DX10, the game starts giving me trouble whilst starting up, but I am not in the mood to change drivers just for the sake of one title. It is already a chore switching PhysX drivers because the only ones that Metro is compatible with, don't work at all on the other games.

Anyone ATi and Nvidia users care to share their settings and corresponding frame-rates with Metro 2033?

 

so you finally see better performance in games with the gtx280s with the dell beta driver than with ATIs?

 

if you need anything just pm or email me.
qx9300@ 3.125ghz
280m gtx sli stock
heavily tweaked OS


here is my ati cpu clock dropped alittle but stock ati's.
http://i49.tinypic.com/n1qzqb.jpg

 

They are just about the same. In some games the RADEONs are better (Dirt 2) in others the Nvidias are better (GTA IV, NFS: Shift).

I really don't think there is anything between them. Choose the brand you like best and you will be happy.

 

I concur. Red or Green, there's no difference in the end.
As soon as your games run equally flawless, you're fine.
You were right from the beginning, but it was worth a try, right?

 

When benchmarking using Vantage, what type do you use, Entry or peformance?

 

Use the stock settings. Performance I believe. Standard 1280x1024 resolution. It should al be set when it opens. Also to speed things up go into options and unselect the "Features Tests" There should be six of them. They don't count towards your final P score. Also be sure to disable PhysX or it will inflate your cpu score.

 

wow, just ran Vantage and got P9955. I didnt overclock anything. Turbo is on, as well as crossfire. I got the following Gpu test 1: 32.95; gpu test 2: 33.47. and for the cpu test 1: 1140.73; and for the cpu test 2: i got 6.36 steps/s(overall Gpu score 11334, Cpu score 7292). I really think it is the cpu, because the last core is almost always throttling when put under pressure. The system is set to High Performance. I reformatted the System about 2-3 weeks ago. I had the dell drivers on and recently i installed the 10.3, and about 7 games. Should I try reformatting?

 

If I'm reading your sig right you have the 920 correct? Your CPU is throttling or being limited somehow. I have no answers since I havn't picked up the R2 yet, but you CPU should be pushing out around 16K. Are you sure it's running on all cores?

Here is Witchers stock vantage run with duel ATI's and 920xm
Screen Shot: http://img194.imageshack.us/img194/3203/vantage.png
Post: http://forum.notebookreview.com/showpost.php?p=5805304&postcount=8

Also here is Mandrakes stock 920/2xATI run.
Screen Shot: http://forum.notebookreview.com/attachment.php?attachmentid=44346&d=1265161009
Post: http://forum.notebookreview.com/showpost.php?p=5823681&postcount=208

 

I just ran the cpu test 1 with throttlestop and the lowest i saw was around 1900 steps/s and the highest was 2300, i didnt continue the test because i thought it might damage my cpu, the last core went up to 100C. any thoughts?

 

Sounds like your overheating issue was not resolved from your overheating thread. Have you taken the laptop apart yet? Perhaps a dell tech forgot to remove tape from the heatsink prior to assembly or somehow forgot to apply the thermal paste. What ever it is, it appears your system isn't dissipating heat properly. If it is running at stock clocks, and your fans are running at max speed, then it leads me to believe that there is a lack of heat transfer fromt the CPU die to the heatsink. I could be wrong, but that might be a start.

 

i used to get, 18.5K in 3dmark06, i just ran it now its 16.3K. i ran the diagnostic and it said ERROR Code 3700:011B The fan speed is incorrect. Thsi thing keeps on getting bettere and better. I opened up my system, with a tech that i know(not from dell), the heat sink screws were loose, so i wiped the old thermal paste and put some on. Not sure if it was a good brand.

 

And tightening the heatsink did nothing? Or did you just now do that? Have you tried reflashing your BIOS? Brand won't make much differance, maybe a few degrees, but name brand paste isn't essential.

 

Under system info in 3dmark06 and 3dvantage it lists this info about my processor


Internal Clock 1.1 GHz
Internal Clock Maximum 2.3 GHz


Why are they not the same? I checked others' results online and they are equal.

In my power settings the minimum clock should be 100%. I am plugged in. Using q9100 processor.

 

Thanx for all your help, DR650SE, reflashed the bios, didnt do anything, but i opened up my sytem again! and saw that the thermal paste was actaully ruined, don't ask me how, but i'm guessing it was either because of the brand i got or because of the overvolting or both. I went around the computer areas in kuwait, it was kinda tough finding a good brand, but i found the brand the tech used and it was worth around 1K.D.(~2.89 dollars). Searched around for a while until i found a good brand TG1, i think, the best i could find anyway. seems to be running smooth so far, Max temp on 3dmark06 was 80C. Thanx for your help, and you too Mandrake. +rep for both you.

Edit : Vcore was set to 75Mv. Going to run it normally and tell you what i get

 

Here is a benchmark of 3DMark06, for the top 5, 20,449
LINK:- http://img148.imageshack.us/img148/5390/20449.png
just in case the one above is not acceptable here is another one, 20,410
LINK:- http://img194.imageshack.us/img194/2422/20410.png

 

Here is my first benchmark with 5% oc in the bios
http://img130.imageshack.us/img130/4025/5ocinbios.png

 

I noticed your proc supports VT-x, but as far as I understood the M17x bios doesn't support VT-x. Where did that come from? Or am I mistaken about the support?

 

I dont know much about vt-x but I am in the same understanding as you.

 

Actually after I posted that I was researching a little bit and it appears that it's VT-d that isn't supported (The 'better' VT-x) as was discussed in this post:
http://forum.notebookreview.com/showthread.php?t=463228&page=127
Though VT-x is supported (It must be if it shows on yours), so I don't know why they were arguing. the -d just adds a few nice features but nothing that I've ever had to use and I've been running VMs for a long time now.

 

Maybe it is just showing that the CPU supports it although the bios is blocking the utilization of the feature?

 

Good job, moving up the ranks.

 

Thanks, Mandrake, you helped alot. Now for some Vantage runs.

 

If you guys keep this up, I might have to jump back in the game.