Aw m18x R2 Dual 980m SLI upgrade!!

When you see that Dellienware is so perverse that they manage to delivers a 180w power supply on their high end laptop; which right now is the new Alienware 17r2 then you can wonder what else strange they might have done with bios (power setting / throttled TDP of the CPU and GPU) ..

 

The only real problem with the new machines (and it is a huge deal breaker problem) is BGA crap. BGA is the death blow that makes it an unacceptable product. That BGA CPU (and no MX CPU option) is what makes the MSI turd equally unacceptable. Having MXM slots does not cure the cancer for MSI... it's still cancer, just hasn't spread as far yet. It's not 100% about performance. It's mostly the fact that the expensive machine is a disposable pile of junk that can never be upgraded or easily fixed when a part fails with BGA.

Performance is what it should be with the components it has as long as you are not trying to use the 180W "trickle charger" to run it. It runs as well as the M18xR2 with a single 980M, doesn't have the GPU throttling problems, as long as you don't expect too much from the 4980HQ CPU... but, as lame as that 4980HQ CPU is compared to an Extreme CPU, even that is more potent than 47XXMQ and 48XXMQ.

NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7,Alienware Alienware 17 R2

 

Did they seriously mismatch memory in that laptop... Also that CPU physics score outdoes my 4910MQ.

 

Yup, sure did, LOL... 4+8=12.

But, it's typical consumer grade 1600 RAM so it matters little in the grand scheme of things. Fixing it with matching sticks and 1866 or 2133 modules might net a couple hundred more 3DMarks in the best case scenario, no benefit for normal use and games.

 

Not only is it asynchronous dual channel, but each stick is a diff manufacturer.

I like matched sticks because I like to flash the memory if needed. I'm currently working on a profile for my 32GB DDR3L.

 

I agree with you, but the average console jockey that just got his life in order, converted to PC gaming and lives to ride the Steam wagon isn't going to notice or care. That is their target audience. With only two RAM slots, 32GB won't be happening unless they start making 16GB DDR3L modules and I think the Lake of Fire might freeze over before that happens. This is a expensive child's toy with a nice chassis, not a serious high performance machine.

 

They say that the New Alienware 17r2 running xtu stress test with temperature of 90 degrees Celsius; But with how high wattage? A thin laptop normally has a minor heatsink ..

 

Just trying to get the 970M working in the M17x R3 atm :thumbsup:

http://forum.notebookreview.com/ali...70m-6gb-testing-see-if-works.html#post9907275

So far no luck.
Brother Fox - Was just wondering what symptoms you saw with your M18x R1 last time with the single 980M?

 

Depend what you prefer.

GT80 will be MUCH better in gaming than M18xR2. The GPUs in the GT80 are running unhindered without stupid UEFI and BIOS getting in the way. CPU wise its not bad either. Sure its a HQ processor but it got unlocked multiplier up to 44x. Stock clocks is 40x, which is respectable as well.
If you plan to run CPU benchmarks, sure, M18xR2 with Extreme processor will be better. But for gaming, overall score, GT80 is beating M18xR2 with 780M SLI severely.

The cooling of the GT80 is a different story though. Alienware machines, AW18 or M18xR2 got it beat bigtime. The GPUs there are cooled properly with super beefy fans. But the problem is poor CPU cooling. The layout for heatpipes in this notebook are just crazy in regards for the CPU. 2 tiny heatpipes that goes all the way to the CPU which is in the middle without any fans.
Absolutely horrible: http://www.hardwareheaven.com/wp-content/uploads/2015/01/msi-gt80-titan-sli-review-inside-base-300x190.jpg
http://www.hardwareheaven.com/content/reviews/laptops/52438/msi-gt80-titan-sli-review

So although the 44x multiplier are there, I think the CPU will just run way too hot

 

That 4980hq run is toughly 100 points more in physics for 3dm11. Both systems running w8.1.

 

Hidevolution says; The M17x R3 currently does NOT support the GTX 900M series. I do not know what to think about this .. HIDevolution NVIDIA GTX 970M 6GB Upgrade Kit for Alienware 17 and M17x R4

 

The total score and the graphic score is still far ahead of M18X R2, even when it runs the crappy 4720HQ
http://forum.notebookreview.com/attachments/msi/119866d1421312321-official-msi-gt80-titan-owner-s-lounge-msi-gt80-3dmark11.jpg

 

But have a hideous physics score .. :cry:

 

True that - Just thought I'd give it a shot anyway. :thumbsup:

 

But why? Is it due to no throttling, proper utilization, better drivers? I really don't know. I'm too poor to afford a single 980M to test anyway

 

Since the HQ chips are TDP locked, you'd be running the OC at 47W TDP and 72A current... not gonna even BEGIN to hold those clocks.

Also, the chips is only 3.6GHz stock at 4 cores. 2 cores it's 3.8 and 1 core it's 4GHz. Its OC can hit 4.2GHz on 4 cores like the 4900MQ and possibly 4.6GHz on a single core, but if it can't add more power to itself, it's not even as good as a 4710MQ at 3.5GHz when you REALLY stress it, and a 4800MQ at so much as 3.8GHz trounces it thoroughly. I've been looking for a 4980HQ owner to help me test with power limits unlocked in XTU to confirm, as I already have confirmation from MANY 47x0HQ owners that their chips don't draw above stock no matter what ya do in XTU/Throttlestop/BIOS/etc, but I can't seem to actually FIND anyone with one.

 

Eh, seriously, not gonna lie. I'm jealous of that physics score in WINDOWS 8.1 (AW 17R2)!!! Almost 11K on a BGA CPU. That's 900 points higher than my system... when running Windows 7!!! Something is wrong with my 4910MQ :S.

 

Brother Reborn... I tested the 980M in the R1 with dual 980M and PEG mode it gets 8 beeps the same as trying to use 980M SLI in the R2 in PEG mode without pure UEFI. Setting SG mode (Optimus) and a single 980M it completes loading the Intel Option ROM. The alien head logo appears for a split second and the machine restarts before the BIOS complete the POST operation. It was impossible to press F2 to enter setup. It would loop this way indefinitely until I power off the system and remove the single 980M.

 

Thanks for that Fox! :thumbsup:
That is exactly what I am getting atm. Exact same symptoms. Only thing I am noticing is that I tried reverting back to an earlier version system bios and was able to load the Alien head POST bar to about half way on A08 where as on A12 it only loaded like 2cm in or about 1/10th of the bar. I'm hoping to find some much earlier versions of the M17x R3 system bios to flash too and conduct more testing later on. Just a stab in the dark really.

 

Where have you seen a 4980HQ or 4720HQ get 11k physics score? I've not seen any recently linked benches showing that.

 

Right here... [ LINK] I posted an example last night in a couple of other threads.

NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7,Alienware Alienware 17 R2

Most jokebook owners are not going to be able to extract this kind of performance from the Iris Pro turd CPU. I used a 330W AC adapter and cranked every setting I could with XTU and strong-armed it a bit with ThrottleStop. It turned out better than expected, but with the CPU welded to the mobo, that might not be a very good long-term plan for the 4980HQ. If it doesn't hold up to the pressure it will be a mobo rather than a simple CPU replacement. I don't really lose much sleep over my Extreme CPUs getting worn out from too much extreme overclocking since this is their primary reason for existing and the solution is a really simple one.

 

So the 4980HQ is able to go above stock power draw? And you sent it to 4GHz and it stuck? Or was it simply able to handle 4GHz using throttlestop at the stock TDP/Current due to the nature of 3Dmark 11's bench?

I've been hunting for a 4980HQ owner to see if settings in XTU etc would stick. I've already determined from MSI/ASUS/Clevo owners that 4710HQ/4720HQ/etc can't go above their short power boost of ~56W on package, and frequently downclock if under any REAL load (hell, one GT72 owner couldn't get over 44W on the package on his 4710HQ). I do remember however that my TDP at 3.8GHz running 3DMark Firestrike never actually went close to 60W, so that may explain it. But I would still like to do more testing to find out their full limits.

 

No, not 4.0GHz... 4. 1GHz. And, yup, it stuck as you can see from the excellent Physics score. It never budged an inch off the 4.1GHz with the 330W AC adapter. Using a 180W and 240W adapter it falls flat on its face and plummets to about 3.2GHz under load. That's all I tried so far. I am going to see if I can get it to 4.4GHz without catching it on fire or desoldering the CPU from the motherboard, LOL.

 

Ohhh what? ? I need to use throttlestop? >_<

 

Can you run it through Linpack at 4GHz or an XTU stress test while checking the max power draw in HWiNFO for me? If that chip can actually take extra TDP, then it's at LEAST half a saving grace for it. The 4720HQs can't even hold STOCK when playing some intensive games. I mean... stock. How you no hold stock? Wth?

 

Omg I don't know what they did to this site but it's aweful almost completely unusable change it back ! It wasn't broken so why screw with it ....

 

Can you run a Cinebench 11.5 and Wprime 1024 test with maximum overclocking + 4.1ghz on Alienware 17r2 as well check wattage with Hwinfo ?

 

And still far behind the Clevo. At least for now anyways.

 

Yeahh I think single 980m on R2 can perform better sometimes if it wants too 3DMark 11 Graphics score 16321 (Single 980m Alienware m18x R2)

http://www.3dmark.com/3dm11/9356686



Fire Strike score 10295, Graphics score 12061 (Single 980m Alienware m18x R2)

http://www.3dmark.com/fs/3932054

 

Well, considering the fact that NVIDIA changed something that makes 980M not function correctly with Alienware laptops released before 980M and the MSI jokebook was built on purpose to work with 980M that's probably understandable. That's more of a negative against NVIDIA than Alienware and nothing positive about MSI working as they designed it to work. The MSI is still a jokebook with a POS BGA CPU and it always will be... crippled is never cool and the M18xR2 is vastly superior to the MSI that works correctly with the components it was intended to work with. If 980M SLI worked correctly with the M18xR2 and AW18 they would both obliterate the MSI jokebook.

Alienware machines released before Maxwell even existed are not the only ones that don't work correctly with Maxwell. We have NVIDIA to thank for that. It's not a slam against any Alienware or Clevo system that doesn't function correctly with Maxwell. Hopefully, NVIDIA will get their act together and stop being so sloppy at what they do.

The new Alienware BGA jokebooks also perform better with a single 980M than the superior machines that came before them. So, I don't give Alienware or MSI credit for building machines that work right with the components they were designed to work with. This is the basic price of admission, not an act worthy of special recognition. If they had done anything less than that it would be worthy of ridicule.

 

so, i am too lazy to read ALL the comments, but for alienware 18(the haswell), I can upgrade to 980m sli with just pure uefi and it will not have throttle issues if i have the 100w nvidia heatsinks and the modeded drivers? or will there still be throttling even if my cpu is at stock speed.

 

It's possible... but we can't really guarantee it to you. It's up to you to make the plunge and you *WILL* have problems making it work. It's not going to work as well as 780Ms would, at least not 24/7. You'll also have to do something about your fans, etc.

 

kinda unrelated but here goes:

 

i7-4710bga can not keep up Maximum turbo core clock in 3DMark Sky Diver bencmark test. Clocks down to 3 ghz .. http://www.3dmark.com/sd/2760703

 

Are you sure there's not something wrong with the Physics score there though, because my 2630QM scores 6511 points - i7 4710HQ should & does score more than 8% better than my CPU right? The Combined Score (Physics & GPU test) is pathetic too, it's actually less than the 15000 that my laptop scores! The poor combined test could be due to the 180W power brick, but not the poor score of the Physics test - something up with that particular A17 R2 example you picked I think.

EDIT: (below was a mistake on my part, CPU test is only CPU, hardly any GPU.)
That's the new A17 R2, it's probably due to the measly 180W power brick. The Physics test in Skydiver uses a large amount of GPU & CPU power (not just CPU power alone), so that's probably why it's throttling in that particular test - 180W power brick not enough.

 

The Sky Diver benchmark test challenge the processor greater in relation to 3DMark 11 and fire strike in 3DMark. Sky Diver is a very nice test for bench test of the processor.. . I guess Sky Diver test uses approximately 20 to 25 more watts than the Fire Strike test

 

Fire Strike is mostly a GPU test. You can get nice Fire Strike score with a great GPU(s) and lousy CPU. That's true about Sky Diver. I may runs some test with the UPS meter to validate the power draw.

 

Check the cpu temp too. Should i7-4710bga be throttled under Sky Diver bench test ? Temp or power problem ?

 

Maybe this weekend I can look into this with one or two of the other systems I have sitting around here that do not have an Extreme CPU to see how they behave. I have a lot going on, but if I have time I will. The nice thing about having a mobile Extreme CPU, desktop K or X CPU is that you can use force when you need to. Can't do that as easily with the other crap, especially the BGA crap.

 

I think it's impressive that the i7-4710bga in Alienware 17r2 not throttle down more than to 3GHz under Sky Diver benchmark test..HaHa

 

This might be a shot in the dark, but has anyone tried to check for throttling issues on linux? (Using both, Nouveau, and Proprietary Drivers?) The most gpu intensive application on linux, that I can think of is heaven benchmark.

 

Throttling issues is on the processor .. Use Wprime 1024 or Xtu bench

 

Has anyone noticed some sets of drivers affect the physics score in 3d Mark 11?
I've tested this time and time again consistently.
Some drivers seem to totally kick my CPU in the balls (pun intended) at the physics test.
I'm trying the Gsync modded 347.25 drivers. They overclock and graphics scores are good, but physics and CPU is affected.
If I switch back to 344.91.. things go back to normal but the GPU score isn't as good.

Has anyone witnessed this kind of beharior affect the CPU ( beyond the token error or margin) after changing drivers?

 

Yes the drivers 347.09 and newer do create a CPU performance hit during benchmarking.

 

I get an ok score in physics (3dmark 11) @ 4.2ghz with 11250p. This with windows 8.1 and the standard 347.09.(11870 in 3mark Fire strike With 347.25). I do not see major change from earlier drivers ..

 

This is something I've noticed only with 980m.. and only happens on some drivers when SLI is activated.

 

How big is the difference in physics scores, between this driver and the old ?

 

High 10k's to early 11k's vs 9k's

That's quite a lot. Strange thing is if I run in single GPU mode on the bad drivers, the physics score does goes back up to 11k's. Switch back to SLI it gets kicked down..

I can't figure out why.

 

The same result of physics ? if you do not overclocking the hardware ?

 

For 3DMark11 I've only noticed a difference of only 30-40 Physics score points or so between drivers, which is for the most part within margin of error.

EDIT: Ah ok, just read your later posts, it's only when you have sli enabled that you get the low Physics score - in single card mode the 'bad' drivers give you the same good Physics score. Mine's obviously a single card machine!