*OFFICIAL* M17x Benchmark Thread - Part 3

Yea, as far as things go, we're still using 3D Mark 06 and Vantage for official scores.. but good score never theless!

 

glad you made it with a 920xm theoretically i could reach this score too but i never ever overclocked the 5870s yet because in real life i never needed more gpu power on games

 

Nice! after seeing you post I said to give it a try for fun, no screenshot though
P5035
Result

GPU clocks 900/1200 @ 1.15V
throttlestop at 96/82
CPU 3% BIOS OC no extra voltage or anything

that's on my normal windows installation (not on the benchmarking partition), I will probably keep those settings for everyday usage

 

hey guys I wonder why did you guys could reach the 900/1200 with dual 5870 because df it runs at 700/1000
The key is the voltage?thanks

 

voltage to 1.15 is a must if you want to reach 900/1200. cooling is the other part. If you have stock dell pads and thermal paste, you won't reach 900/1200.

 

Got to move forward eventually.

 

Especially with DX11 being on the rise and all

 

I scored 14067. Is this pretty weak? It's just a 720QM, but I have 5870 Crossfire. : (

 

What benchmark? What settings? You told us very little...

 

what thermal pad you recommend for the 5870s and where, how to apply them?

 

Im thinking Vantage....with that score...thats about right for vantage with x/f 5870 and 720qm (was my 1st system) - oops , i mean 3dmar06 - im always gettin' them confoooosed...14k is the score i get with vantage on my current system...(16.5k 3dmark06 current system)

 

I have a 920xm incoming. Can't wait to push 100+ TDP and see if I can break 5100 points in 3dmark11.

 

and he's back...welcome Joker don't go around making intelligent posts now..keep all your secrets to yourself:wink: ....maybe throw a couple my way through use of PM...now beat my score...


CPU-Z Validator 3.1

 

here you go
CPU-Z Validator 3.1

 

You know, I downloaded 3DMark11, and have been meaning to benchmark it, but I just lost all motivation. I even have my benching SSD setup nd all, just havn' had the drive to swap the SSDs. Maybe I'll get to it sometime soon.

 

My bet 3dmark06. I scored 14500 with 740 qm, 8gb Xfire 5870. From what I know 3dm is highly CPU reflective.. A 920 would Make the score a few k higher

 

By a few you mean 5-6k higher haha

But ya, I figured was probably 06 as well

 

Yup...i got my 3d's mixed up lol....edit previous post. '06 for sure....

 

Not hard at all, trust me it can go further easily... that's on my new normal windows installation partition with the latest drivers... nothing disabled (yet )

P5229
Result


TDP 94 btw...
with only 3% BIOS OC

 

Awesome, I figured these new 11.1a's would make it easy. Did you mess with the tesselation settings?

 

no change at all, everything is default, I only had to setup the system to run dual screen (extended desktop on a lower resolution monitor) but that's it.
The only problem is that with any clock higher that 700/1000 there is flickering on the extended diplay (it's a known problem, with all the drivers) so even though I am stable I will have to decide if I want to keep the clock like that.
The problem disappears when I run let's say a youtube video that changes the clocks of the primary card to 500/1000 from 405/1000 (the default clocks when adding a second display)

 

Are we talking about these drivers http://forum.notebookreview.com/alienware-m15x/549218-catalyst-11-1a-beta-hotfix-drivers.html that were posted in the m15x forum? - do they work on 5870's? - anyone know of an official release for them if they do?

 

Wow, I'm still running on the official Dell drivers. Don't really have any issues with them so I never saw a reason to change.

 

yes they work but I don't know about a date of an official release

 

I have the 11.1a hotfix on ATM it's fine for me.

 

any sleep issues? - I had some with 10.12..using 10.10e atm...they seem fine.

 

Yeah I am still having problems with the sleep mode...

 

Yes...sorry for not stating before I benched with 3dmark06 with defaults. So...it sounds like this is a "normal" score for a 720 which I'm okay with. I just went through a ton of hassle finally getting my Crossfire working correctly, replacing cards, etc. So I wanted to bench this thing and give it a clean bill of health. And it's running great now with the replacement 5870's.

Thanks all for your input.

 

damn you ...now time for me to get back to benching if i get a chance to get away from the desktop...trying to aim for 5.0ghz on sandy bridge...but i will get back to 920xm ...In the mean time i think joker should quit being lazy and aim for 4.5ghz.

 

It's been there for a long time... I don't think it is possible to go any further without major changes to the cooling system (talking about -C)

 

Have you tried chest freezer? hehe

The ambient in a freezer is about -20c (around -5f to +5f)

 

how about an ac unit inside of a chess freezer thats also inside of a meat locker, omg the potential hahah

 

they all cap out around the same temperature though....lol

Chilled water though.....that has a HUGE heat capacity compared to heatpipes and air (Ask Johnksss, is what he has been using in his desktop for a year now)

 

New 11.1a drivers are an improvement. Black OPs and Bad Company 2 feels a bit more fluid. Also, now I get 15400ish for Vantage GPU with no OC (CPU has turbo boost though).

 

I wish, there is no space for such a freezer in the apparment LOL
I tried a couple of days ago to run a benchmark on my balcony when the temperature outside was -8C with the keyboard removed and a fan blowing cold air on top of the heatsink but still I couldn't drop the CPU temperature less than 16 C...

 

you are voltage limited, not temps

Raising the TDP gets you more current but you still need more voltage. More current adds stability at a given speed, but more voltage allows more speed!

 

You beat me to the punch, Scook. My laptop scores the same (score in sig), weather it is sitting on top of my AC unit, or just has the rear propped up with a couple of soda bottle caps.

 

Same problem I ran into with my M17x pre TS days. I could sit out on the front porch in 5f weather (which looked pretty funny haha - I was full Eskimo with my laptop outside) and I would score the same and overclock the same as at regular room temperature. Temps are not the limiting factor until 4GHz or so. By that point the problem is often not the need for more voltage but rather the need for cooling due to the voltage you have already added (I would idle at 60c with ambient temps with 1.5V vcore - needed to bench at 4.16 Ghz)

The GPUs are also at the same limit. I found that the cold resulted in less artifacting (meaning more stable vram) but my core overclock would not budge 5 MHz whether inside or out in the cold. This was true of both my GTX 280m's and 4870m's.

With laptops our cooling capabilities will always exceed the available voltage.

 

My scores are seriously affected by the ambient temperature... that day (-8 C) I could easily hit 100C+ with the build-in wprime of throttlestop even at 3.4GHz and by monitoring the multipliers, I could see them dropping from 25 to 21. If we could sustain those multipliers the scores would be much higher. So at those frequencies voltage is not an issue leaving us with the cooling.
For reaching anything above 4.335 yes the extra 150mV provided by the BIOS are not enough and an increase in the voltage will be necessary too (and we are not talking about benching at those clocks, just for simple CPU validation).

 

Ya that is true....the i7's run HOTTTTT (have seen this personally in the desktop space)

That seems a little too hot though, have you made sure that your paste job is good and you have good contact? Also is that with the back raised up and the bottom cover removed? Would also be a decent idea to drop the ram speed if you can to stress the memory controller less and lower temps that way a little maybe (would take some trial and error to see if this yields any gains)

 

yeah the paste was good, but I might reapply it and perform the retention mod as well. Without TS the temps are within acceptable limits. The problem is that when increasing the bclk I have to increase the voltage of the memory as well otherwise it crashes. It's unavoidable for me and memory runs on a dangerous voltage on top of that, 1.7V (it even crashes with 1.6).
So either I go with full increase in voltage and bclk or with stock voltage and only 3% oc of the bclk.

 

Have you looked into Thaiphoon burner? It could get your memory stable at 1.6V probably

 

I would if it was for free...

 

It is not THAT expensive....($20)

I figured, given how much time I spend overclocking and benching, I could spare $20 to aid the effort. Alot of people here have spent more on thermal paste and pads

 

true but i rarely bench, so I would probably use it once or twice and the results are not guaranteed so...

 

I'm thinking next week I might do some more benching. I'll have a few days off so it'll finally give me a chance to do things I've been meaning. Pin mod M11x, and run benches with HyperX RAM. Maybe it'll help out. Have yet to run 3DMark11.

 

Anyone running a core-i, I'm messing with CPU-Z. What I'm seeing now is it is reporting the speed at the lowest multplier, or the actual speed. Is there anyway to change it? I think it might make it hard for CPU validation It'll report a 9x multiplier, unless it turbo boosts up, where i'll get the 24x. I'm using CPU-Z 1.56. Just wondering if anyone else is seeing this too.

 

you can try to put some load on the cpu and see if you can maintain the multipliers high
Fossil Free Online CPU Load or Stress Test.
(if you need more load, open a second instance on separate window)

I think only the XM versions will maintain the multipliers high once overclocked from the BIOS and reported like that with CPU-Z, I don't have a regular cpu to try

 

P5343
http://3dmark.com/3dm11/478996
Result


and there is still headroom for better scores!

 

So close to the GHz GPU....GOGOGO lol