*OFFICIAL* Alienware M18xR1/R2/18 Benchmark Thread - Part 3

Really? I'm happy to hear that What do you see that I'm missing Bro John?
I'm thinking more voltage to get higher clocks. Thing is, that increasing voltage a little along with the clocks doesn't seem to improve scores in the same way it did running a single 780m.

 

for running firestrike.....you need windows 8.....
still working on a work around for it to run well in windows 7

 

Definitely run Fire Strike in Windows 8 as Brother John suggests... makes a big difference. It's optimized for Windows 8.

Check this out... HUGE difference in performance in benchmarks. I was shocked the "better" RAM wasn't...

 

Thanks. I remember reading your post about this a while back. Hmm.. So that makes a total of.. 1 reason to move to Windows 8 lol.

 

It's really not worth upgrading yet for just one benchmark. Fire Strike is one of the few that works better with Windows 8. Almost everything else works better with Windows 7. Fire Strike gives good overall scores to computers with relative wimpy CPUs and the Physics Test does not carry nearly enough weight in the overall score. This works out good for Windows 8 because it hinders CPU performance. I suspect this is partly because Intel is making more or more crappier CPUs and the new 3DMark suite is designed to be usable (Cloud Gate and Ice Storm, especially) with pathetically under-powered systems that have even more pathetic CPUs than Intel's worst.

I'm primarily running dual boot because I already paid for Windows 8, but in 20/20 hindsight is was mostly a waste of money. I would still do it for gaining experience with the OS, but not for the purpose of improving my computing experience because, overall, it's a step backward rather than an improvement. Micro$oft hasn't screwed up an OS this bad since WindowsME. I am confident that, like Windows Vista, it will get better in time and eventually we will be complaining about their next abortion and saying how great Windows 8.X is in comparison. The path they are on with being such control freaks, I have my doubts that future versions of Windows are going to be something to be happy about.

 

@Mr.Fox for the ram thing, I personally wouldn't count on cpuz for information as it is often wrong about many things but it does seem strange. unless the benchmark is heavily dependent on ram I don't see any difference. I up my 1866 to 2133 with same CAS and see no performance difference, and since I use ramdisk, HD Tune for sequential and random speed were the best to test any speed difference, and there were little better for 2133 at same CAS.

I can only think of benchmark software didn't like the ram for w/e reason lol I do not know fully how ram is made but I do know for a fact that Samsung SSD are targeting benchmark software and build it the way its better for benchmarks (in hardware and firmware) so looks better on review sites, not sure if same can be done to ram too.


aside all that heres storage mod TWO x 400gb SSD runs on PCI-E gen 2


performance was much better than my original mod with only 240/230 MB/s in read/write

 

I was not using CPU-Z. If you noticed from my screen shots I used several tests (AIDA64, WinSat, MaxxMem and good old 3DMark11 Physics Test) and all were pretty telling. The Ripjaws is just too slow to be acceptable. I was shocked at what a hit 3DMark11 took with the Ripjaws memory... didn't expect it to be so severe.

From what I hear the problem with the performance of the Ripjaws is that it has 4 chips per side instead of 8 (single rank versus dual rank). Anyhow, I returned it to NewEgg for refund and ordered a 16GB kit (2x8) 2133 CL11 Vengeance for testing against the Kingston.

 

let us know how that work out cause I am waiting for mine to arrive lol

 

What could be...

 

very impressive...!!!

 

Thanks Brother Bumbo2
Apparently when working, the 880M can actually run Fire Strike just as fast under Windows 7

 

Waiting to see those 880M SLI benches!

 

Those numbers are insane ! Well done !

Yeah id like to see what SLI looks like too

 

You can do the numbers by using 3d11. Since 3d11 can actually double it's performance. Literally. So going by those numbers. 23k gpu for 880M's or 106/106/140/70 FPS. This would be max. Daily driver would be 21k or so and about 2 fps off each test....Speculation of course.

 

Those benchmark numbers are going to look like all kinds of crazy once NVIDIA releases GTX 880M optimized drivers. Awesome stuff, Brother John.

Here's a new FFXIV benchmark...

 

Nice run!. I think I still have that. Let me re install it.

 

we need this, 16gb module SODIMM
AnandTech | I'M Intelligent Memory to release 16GB Unregistered DDR3 Modules

 

Do you think 880M SLI will be able to put out this kind of performance on a single PSU?

 

Fire Strike: NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2 - 11346 3Marks	Fire Strike Extreme: NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2 - 5817 3Marks

 

I don't know why I'm running this, just felt compelled with ChrisReffler's excellent run.

So here's my 3Dmark CPU run, or Ummm...Ah... sorry, 3Dmark 06 run

Totally stock GPU's. Run Indoors, no special cooling other than fan on high.

NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-4930K,Notebook P570WM

 

I have OC the 680m SLI to 900/2400 and CPU to level 2 in BIOS (cant remember exactly) and here's my result in 3dMark Firestrike

The temperature is almost perfect, which is around 70* for the main vga (with 3500rpm fan speed) and around 75* for the 2nd one (2500rpm), the vbios is from svl7 with default voltage
So can i push further?

 

Good job. Yes you can definitely push further.

 

Great job!
To bad this is a 680M's benchmark.. Fans high only.
NVIDIA GeForce GTX 680M video card benchmark result - Intel Core i7-3940XM,Alienware M18xR2

 

I might be able to break 38K if I tried. But here's a little better. It's now officially a 780M benchmark, lol. j/k

NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-4930K,Notebook P570WM

 

I wonder why the Ivy -e is doing well on that bench? '06 is a 4 core proc bench. Or maybe that's just desktop world. John, what kind of '06 CPU scores you get on your Haswell mobile? Or is it worthless to keep running this old bench, LOL

 

Nice run once again!

That's a big 10/4 negative. The 680M is running stock.
And my ivy cpu seems to down clock after running the cpu test. You and talked about that back when you had the first Clevo. Apparently it's not present in this version of the Clevo. It might be in the 4930xm version though, not sure.

 

Corsair Vengeance 2x8GB 2133 reprogrammed from CL11 to CL10 with Thaiphoon Burner...

 

cl10 with bios setting doesnt work and TB was needed? and theres a SPD profile for CL 11-13-13 at ~2400 mhz? lol

 

Is this running at 1.35v?

 

it should be 1.5 right? doesn't R2 auto set it to 1.5v?

 

Nice. Do you run your 2 sticks on the slots under the keyboard or under the back cover? Does it make a difference? And the CR 1T is that default or did you program that as well.

 

at TI some say you can go cl 9 at 2133

 

Not needed, just more convenient. I confirmed setting it manually in the BIOS worked. Abused the RAM with lots of stability and error checking torture tests for 2 days to confirm I would not regret doing it... turned out great, so then I committed the new tCL setting so I don't have to do it manually in the BIOS any more. (Sometimes the manual settings revert back to default for no reason, so now default is CL10.)

I don't know how to make it run at 2400. I think CPU-Z might be lying to us. Here is what AIDA64 shows for SPD profiles:

Memory Module Properties
Module Name Corsair CMSX16GX3M2B2133C1
Serial Number None
Manufacture Date Week 33 / 2013
Module Size 8 GB (2 ranks, 8 banks)
Module Type SO-DIMM
Memory Type DDR3 SDRAM
Memory Speed DDR3-2133 (1066 MHz)
Module Width 64 bit
Module Voltage 1.35 V / 1.5 V
Error Detection Method None
Refresh Rate Normal (7.8 us)

Memory Timings
@ 1066 MHz 11-11-11-31 (CL-RCD-RP-RAS) / 42-278-7-16-9-9-33 (RC-RFC-RRD-WR-WTR-RTP-FAW)
@ 1066 MHz 10-11-11-31 (CL-RCD-RP-RAS) / 42-278-7-16-9-9-33 (RC-RFC-RRD-WR-WTR-RTP-FAW)
@ 959 MHz 9-10-10-28 (CL-RCD-RP-RAS) / 38-251-7-15-8-9-30 (RC-RFC-RRD-WR-WTR-RTP-FAW)
@ 853 MHz 8-9-9-25 (CL-RCD-RP-RAS) / 34-223-6-13-7-8-27 (RC-RFC-RRD-WR-WTR-RTP-FAW)
@ 746 MHz 7-8-8-22 (CL-RCD-RP-RAS) / 30-195-5-12-6-7-24 (RC-RFC-RRD-WR-WTR-RTP-FAW)
@ 640 MHz 6-7-7-19 (CL-RCD-RP-RAS) / 26-167-5-10-5-6-20 (RC-RFC-RRD-WR-WTR-RTP-FAW)​

I don't think so. I am pretty sure it is running 1.50V. With the Kingston to run 2133 CL10 I had to crank it up to 1.65V so this should run cooler and more efficient. Maybe.

I am using the two slots on the bottom for ease of access. The HyperX, Vengeance and Ripjaws memory all ran CR 1T with two sticks. This is one of the main reasons I wanted 2x8 instead of 4x4. I can manually set it to 1T in the BIOS (the N-Mode setting in the BIOS is memory command rate) but it is not stable with 4 sticks and CR 1T is automatic plug-and-play with 2 sticks instead of 4.

Already tried. It will post, but it's not stable enough. The settings that Klem posted as a suggestion don't work at all. The machine turns on, pauses and turns off. Doesn't even beep, LOL. After three attempts the BIOS reset.

I also tested the Vengeance in the Alienware 18. It works great, but only 1866. It made a big difference in performance, but I guess maybe that's one more strike against and another feather in the cap for the M18xR2. No matter how slice it or dice it, the Avenger is a Viking slayer, LOL.

 

when i get mine i'll try 32gb at CL10, have you try changing tRAS to 27 instead of 31? newegg and corsair website advertise the ram at 11-11-11-27 so how is it 31?

 

btw this article is interesting on the new SSD performance at steady state worst case scenario, and another performance test with recovery time. I urge people whos into storage should check it out SSD Consistency Testing - 18 Drives Tested for Real-World Performance | TweakTown

 

I don't know. It absolutely does not run at 27 by default. It is automatically setting to 31. With CL10 it's not stable at tRAS 27 manually. I have not tried manually setting CL11 with tRAS 27, but I will.

Edit: OK, with 11-11-11-27 (tRAS manually set to 27) it is stable. But, the performance is less than 10-11-11-31. Compare the results in these two images. At the end of the day it is the results, not the settings, that matters.

2133 @ 11-11-11-27

2133 @ 10-11-11-31

 

thanks for the test and screen shot really help a lot. to think one day we'd have 32gb at 2133mhz cl10 in a laptop lol. the day when I can have over 20TB in SSD storage is the day I ditch everything else =), in a performance laptop of course.

 

NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-3920XM Processor Extreme Edition,Alienware M18xR2

 

This is fitting for this thread...

<iframe width='853' height="480" src="//www.youtube.com/embed/7CoeOq-lkdY?rel=0" frameborder='0' allowfullscreen></iframe>

 

Single GTX 780M... and this was run in Windows 7...

http://www.3dmark.com/3dm/2563004



...Windows 8, lower overclock...

http://www.3dmark.com/3dm/2563239

 

rofl I love that ending picture for your video classic stuff

 

Well darn it anyhow, one of my AC adapters on the dual 330W mod just crapped out on me this afternoon. I'm going to have to dismantle it to see if there is something I can fix internally or order a replacement. It was working fine when I put it away after making that video and running some benches last night, but now that one just won't turn back on anymore. Grrrrrrrr.

 

That really bites brother Fox. Hey, is it possible to somehow use a desktop PSU? I know some people have done this for the x51 which from factory uses the same 330w PSU...

These sellers selling PSU's for $100+ on eBay are dreaming. No way I'm paying that for an artificially scarce $30 part.

 

I am hoping it has some kind of replaceable fuse on the PCB, a broken wire or something that I can fix and continue using it. It's odd that it was working fine when I unplugged it and put it in the canvas bag that I store it in. I first though maybe the green LED burned out, but when I disconnect both and plug them in one at a time the one without the LED lit up does not do anything. The system still runs fine and charges with the other one.

I just purchased two brand new ones on eBay for under $50 each. There are 5 more left. Genuine Dell Alienware M18x 330W AC Adapter XM3C3 | eBay

I also purchased some of these with the idea of possibly doing a variation of my current mod so the AC adapters are plugged into the box instead of permanently hard-wired. New Dell Alienware M17x R1 M17x M14x DC Power Jack Socket in Plug Connector | eBay That way if one goes bad I can just plug it in another one and run. I will also replace or splice the cord on the good one left on the current dual mod, so I'll actually have four to work with by the time everything is said and done.

 

Maybe the wire just came loose inside the box. I would start there. These things can take alot and you are not even coming close to pushing it on it's own anymore.

Edit: My first real run since getting my system back up and running with my (Outdated and old tech no match for maxwell) 780m's

Spoiler

 

I almost committed to getting a couple and then I saw they wanted $50 to ship it here a piece. Ahhggrr. I really have try Dell here and see if they'll sell me one (actually only need 1 more).

Those adaptors were a good find on eBay. I'm hoping that will work
Since I have 2 PSU's already it means I can get the eBay seller to cut the cord and send me it in a big envelope lol
Then I'd be able to make your (proposed) box that interfaces with the AW yeh?

 

One does not turn on. Plugging that one in does nothing before it gets to the box... no green LED or anything. Like it's not connected to the wall outlet. I'll look inside of the box just to be sure nothing is out of place, but with the LED light never coming on when it is plugged into the wall outlet I think it's something inside of that one AC adapter.

 

Hopefully it's not a fried prong, but if you got a new one, then that's half of your new double sided adapter plug.

 

Just something I am curious to know, on average how many years do these 330W adapter last before they go out?

 

@mr. fox probably capacitors going bad stuff on pcb burns all the time in a power supply. that 50 dollar PSU is darn cheap

@johnksss hows that 880m working out?