MSI 1070 in Alienware 17 R1 Ranger SUCCESS!!!!

very good input, i operate on the same principle: cover every single component larger than a resistor with thermal pads and triple check for proper coverage u can never use too much thermal padding, but too little is always a risk!

Sent from my Huawei Mate 8 NXT-AL10

 

The mosfets above the odd placed coil should be cooled. The ones in between or circled in the picture, your free to make contact and cool them too, shouldn't be much of a deal breaker TBH.

 

Now that you have to dremel your chassis, what about to design a single heatsink that cover the Processor and GPU made of cooper and replace the original ones (GPU+CPU).

I know there is a lot of work, but is anyone here capable of doing this?

I believe that we are a lot of users of the AW17, and before buy the 1060 we highly preffer to go for the 1070.
If we have that custom heatsink and Lee can sell it to us in Combo with the 1070 would be amazing!!!

 

They already designed that single crap in AW15R3 and you need to repaste everything even if you want to repaste only one component. And you need to take care of the thermopads too

 

+ The heatsink must be nearly 100% perfect. A small unevenly on a unified heatsink, will leading to problems with correct assembly. Always been that way a unwanted design!!

 

Just out of curiosity, from your pictures I don't see an SLI connector on the board.

I thought only the 1060 did not support SLI?

 

No sli possible with this card.

Sent from my SM-G930W8 using Tapatalk

 

Ah man! There goes my idea of possibly upgrading my M18x R2...that and that little bit that sticks out on the 1070 which looks like it will impact the heatsink for the primary card (see http://forum.notebookreview.com/attachments/image-jpeg.128954/) - I think I'll just stick to my 970M SLI upgrade plan which is roughly the same performance as a single 1070.

 

1060/1070/1080 Won't work in M18xR2

 

sorry I just want to take a moment and say:

HOLY ****KK!!!!! WOOOOOOOO

 

yes, I have a very very nice Dremel set, AND I also have a new AW17 to work on

 

if you are looking for a custom/modded vbios for that msi mxm card, check out the msi forums. there is a forum guy there named Svet who used to create custom bios and vbios for msi. not sure if he still makes custom vbios (system bios seems like all the rage nowadays), but it maybe worth a shot
https://forum-en.msi.com/index.php?action=profile;u=29469

 

I'm pretty sure that he could pull this off actually

 

UPDATE 2: Made a bit of progress with the 100W heatsink. The unfortunate result was 1 broken and one unusable heatsink. I will have to wait for arctic silver thermal adhesive to complete, which will likely be monday. In order to make the power component area of the heatsink flat enough to clear the components by using the dremel, you are left with not enough material in the one spot. (easy to see in my pics) For now I have updated the photos on my onedrive link so others can see the heatsink. I have also added thermal pads to the 75w heatsink in the 2 locations I pictured before, and it has not solved anything as far as my strange crashing upon cpu/gpu combined loads. Stability after the first crash is completely fine but the first crash takes about 10 seconds or so to recover from. Im starting to think its due to a limitation of power from the laptop bios itself. Everyone I have seen with the 980m in this machine had throttling issues as well until Svl7 and prema created their modded vbioses. I feel this will likely be a similar issue.
The driver issues are still persistent, but I have noticed that If I power cycle the laptop with the mini display port disconnected I have a 50% chance of the laptop booting to a black screen. Once it has booted, sleep mode appears to work perfectly fine and my display will power back on each time I wake it. The issue only appears to happen upon cold boot. Im going to try my spare edp display cable hopefully tonight to try to eliminate that from my possible list of variables.
I have created a thread in the MSI forums in hopes of finding a vbios modder capable of the msi vbios mods I will likely need.
I still cant complete a 3dmark11 test, but I have yet to fail a 3dmark FS or FS extreme test.
Here's more results since the thermal pads were changed on the 75w heatsink.

http://www.3dmark.com/fs/10694986

http://www.3dmark.com/fs/10725665

 

Have you underclocked it to minimise power draw to see if you can get it stable?

Is there a program that can read GPU power draw? To see if it's spiking just before crash.

Total guess, are your Intel GPU drivers up to date?

The blank screen with mDP connected is a weird one because the iGPU should be handling all the outputs. But good call on trying the cable, but don't AWs beep and fail POST without an internal panel connected?

+Rep to Svet, he is like the Prema of ungimping MSI stuff

 

Basically it under clocks itself after the first crash. It runs around 1350-1450 mhz core and bounces there endlessly. I played 4 or 5 hours of wow tonight on it and it was 100% stable. Never dropped below 1350 mhz. The Intel gpu is disabled as it can't handle 120hz. I have the 120hz edp display so Intel gpu is completely disabled.

Sent from my SM-G930W8 using Tapatalk

 

righto, so like the m17xR4 the LVDS is connected to the iGPU and eDP is connected to GPU for 120Hz/3D?

Please upload a screenshot of GPUz sensor tab so we can see if it's GPU driven throttling

E.g. this is how I found the 980M's in this P870DM power throttle down from max boost in the low 70's (at least 15C away from when they thermal throttle)... SLI vBIOS flash on the todo list

 

Its gpu power throttling for sure. Green bars all through the power limit reason sensor. Im hoping is just lack of cooling on the power components and if im wrong then I'mm move on to vbios testing.

Side note, I havent had a hard crash in over 10 hours of gaming. Turned up the graphics settings too so I am testing more of the memory as well. Thermal adhesive is on its way... if im super lucky I'll see it tomorrow.

 

Since now you have quite the collection of Heatsinks, I would say you can weld or thermal adhesive 2 pipes from one of the broken ones over the working one to give yourself more cooling capabilities.

 

Everyone, It's OK - I have the solution

 

Yes you should monitor the TDP wattage pulling when running the benches I always do it. From HWInfo you can read it on the OSD, Better first to have a look at that...
And did you try the MSI AF to check the voltages too, maybe some underclocking would get it stable ?

 

possibly a GPU failure?

 

Any updates?

 

What we need tested is one of those new standard MXM 3.0B sized 1070N cards to test in the 17 R1! Then you can use the 100W heatsink and chop nothing away

 

Do you have a link to that new card King? Iwas just about to buy a 1060 to throw into my 17r1 but I am willing to buy a 1070 standard form card if there is one available

 

see this post and the ones after: http://forum.notebookreview.com/index.php?posts/10384104

basically, just get urself a zotac zbox en1070, take out the gpu and slap it into your machine

Sent from my Huawei Mate 8 NXT-AL10

 

That is one way. I am hoping Woodz or someone can source this particular card from somewhere as a bare card sale or upgrade kit. Then I may just get a ranger like loafer!

1070 is quite alot better so worth waiting for perhaps. Also E9550 coming soon at 95W tdp it will also be better perf/watt than 1060 and come quite close to 1070 indeed

Sent from my SM-A500FU using Tapatalk

 

m8, don't waste your money on a 17 ranger

 

So what should I waste it on?

Sent from my SM-A500FU using Tapatalk

 

M17x R4 with the 120Hz display. That way you don't deal with the retarded NVRAM and reduction in ports the Alienware 17 has, not to mention the locked BIOS. Just put it into full UEFI mode just like the Alienware 17, no reason why it shouldn't work. You may have to alter the installation though.

 

Over eDP right?

 

Nope, just not a very stock BIOS (stock BIOS hangs at post).

EDIT: Ooops, posted this in the wrong thread...sorry for hijacking!

 

I thought the R4 had hard limits on cpu power consumption?

Sent from my SM-A500FU using Tapatalk

 

You got a pascal card working over LVDS??

I had a 3940XM running at 41-41-41-41 no issues.

 

the EM series has switchable graphics, so its via iGPU...

 

Personally I like both of them, The last true Alienwares...

The R4's lid is not metal though, and the Ranger doesn't have the base as the old gen MG/AL alloy but chassis and lid are, Next the R4 has the easy access battery bay, I don't know how many SATAIII and USB3.0 ports does the R4 support, M18x R2 had 1 or 2 SATAIII only iirc.

Yeah the Unlocked BIOS is super cool with the R4 but her twin Ranger has NVRAM secure flash locked sBIOS ( @svl7 unlocked the sBIOS but with a hard mod BIOS chip, only his AW17 Ranger machine had the Unlocked sBIOS in the world, unfortunately he's MIA, Hope he's doing well ) & the Fans are crippled on the Ranger as well (Max RPM lock from Dell, can't use custom table with 980M on Compal EC only need to use Dell EC to max RPM on fans while on load), both the machines can be modded to have the M14x R2 High 14CFM CPU fan modded with lil dremel on the fan, the GPU fan and CPU fan on the Ranger are turds, I use both fans from the R4, pnp.

R4 and Ranger have some issues, Here's the Info on the CPU restrictions.

 

Didn't @D2 Ultima say that the M17x R4 is hard locked to 68w?

 

On the CPU? It is.

 

Then @Raidriar 's XM chip is a lucky one.

 

I also had a custom BIOS mod I did myself, but this was 2-3 years ago at this point, and I don't have any of my old M17x R4 files since they went in a hard drive crash

 

It doesn't prevent the CPU from clocking up... but voltage and load type obviously will be the deciding factor under load. If he was doing 4.1GHz on 1.15v or something like that then he probably never hit the 68W limit in any game or most benchmarks.

 

minor update before bed. I think nvidia has software running in the background that is causing the graphics hangs. If I exit the nvidia tray I can complete the 3dmark 11 run without crashing. Haven't tried gaming for a longer period of time yet but I would only assume the same results.
CPU @ 4ghz 1.134v cache at 4 ghz GPU still stock for testing.
http://www.3dmark.com/3dm11/11746704

Another note, not sure whats going on but now with 2 different 100 watt heatsinks, I am consistently getting lower scores AND higher gpu temps. Not sure if this is due to the copper on the power circuts heat soaking the gpu pipes or not but very interesting results regardless.
100w heatsink is the LOWER score. 75W heatsink HIGHER score and better temps (by about 4*C on average)
http://www.3dmark.com/compare/fs/10800655/fs/10725665

3dmark 11 1070 on 100w heatsink vs 980m best OC I attempted.
http://www.3dmark.com/compare/3dm11/11746704/3dm11/11261867

 

Is it this? What drivers are you using??

http://www.majorgeeks.com/news/stor...o_latest_drivers_heres_how_to_disable_it.html

 

Do you have the whole Nvidia software package installed?

Try uninstalling Geforce Experience, 3D Video, Wireless Video, and anything installed other than the Display Driver and Physx, then reboot - if you need the HD Audio then keep that too.

 

Did you actually unlock that 68w limit?

 

Tried both ways, with and without the software packages installed. I only caught the issue now (with full software packages installed) but I was not able to complete a 3dmark 11 pass before until I tried exiting the nvidia software in the tray. i never tried exiting the tray software when I only had the bare driver installation. The error I would usually get is "3dmark has lost priority over the gpu and/or keyboard" and it would crash on the final test (combined load) every single time I attempted it. I even tried underclocking the cpu to 3ghz with less voltage. It made no difference.

Drivers are 375.70 using J95's modded inf from tech inferno (has has added the 1070 strings now to make it easy for other upgraders.)

 

@j95, our driver wiz

 

I think so, but it could be that 41 x 4 is not enough to push past 68W even under full load (for core tdp, not package TDP)

 

Damn, that's a nice chip.

Even my "golden sample" 6820HK as indicated by @D2 Ultima gobbles 75w of power on 4.1 and most 6820HK that gets sent takes 90-95w for 4.2GHz.

 

It was Package TDP that was the lock as far as I know.

You had near 1.2v though, and were using OCCT. Wasn't TSBench around 60W for you or something? Most benches don't pull as much as P95/OCCT/etc, which is why I figure that many people managed 4.3GHz or so on their 39x0XM chips for gaming/benching. For me with streaming I would have hit a power limit real fast, but that's got little to do with most gaming scenarios.