*OFFICIAL* Alienware 17 R5 Owner's Lounge

Captn.ko first of all do you use bios 1.2.1? The low frequencies maybe because of power throttle? Because, if it is power throttle I do not know what to do other than undervolt which could also reduce highest frequencies by its own? Because it is not thermal throttle. But it could also be thermal throttle because of samsung SSD high temps while gaming.
That's why I am looking for another NVME heatsink.
I think I have it on performance but I will double check when I go home. In addition, when I run prime95 test all frequencies drop and stay at around 3.3GHz stable for all cores... If this is not the bios or power throttle I wonder what it is

 

nope 1.1.6

 

well, this is the start point. I even have to reinstall 1.1.6 or you have to update? I do not see the point of comparing with different bios versions.

I suspect that dell has done changes to CPU handling. High performance is selected on windows. Frequencies drop to 800MHz even during idle. Of course no thermal throttle with temps of 40s.

However, before I apply LM, farcry 5 temps were going to 96C or higher resulting to thermal throttle. Now as stated it does not go over 83C. Of course, this temp was at 4.3Ghz.

I have to resolve the samsung NVME high temps during gaming. I will try EK waterblock ones next even though pricey. I have seen reviews of those and looks like they are good.

 

never change a running System

idle is not so important...but clocks should never drop below 2.9ghz under load

 

Why not? You may get better temps or not...
Then why do frequencies drop to 800MHz (for almost all of the cores) during idle with high performance selected on windows? This is not never.. I checked with Intel XTU during idle. No kind of throttle thermal or power or limit throttle during idle. Frequencies still drop to 800MHz, But this is bios 1.2.1.

You gonna make me downgrade. But never change a working system

 

i said never below 2.9 UNDER LOAD! idle is the most unimportant Situation for me. And no, i will not update. Normaly it shouldnt drop below 4.3ghz... especially with LM repaste

doesn't look like a working system with load clocks 2.4ghz and below...

because of me you do not have to downgrade ... my cpu clocks are fine. I just wanted to help you...

 

I know you try to help. I never said the opposite. Do you recommend to run cinebench? I will downgrade to 1.1.6 because bios settings on different versions are different. We can compare only with same bios versions and same components(But the second cannot be done anyways).

But I am still waiting for an answer. If it is not the bios and the thermal throttle, what can reduce frequencies under gaming If under high performance in windows?

 

Cinebench @ 4.3 should be around 1380- 1420

 

It didn't trip the PL1/PL2 limit.

With prime95 you have a constant 100% load, so the CPU doesn't downclock because of the lower usage, instead you are probably limited by something else.

It is possible that your limitations (in prime) are similar to those encountered by doofus99 in Prime 95, see post:

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-136#post-10760367

You could run the same test to figure out if there is a similar pattern:

-set your power limits in BIOS to 110w,

-open throttlestop and untick everything in the main TS window, except speed shift epp and set the value to 0, default is probably 128),

-make sure you force max fan speed before you run the stress test,

-open hwinfo sensors and make sure that all the relevant info is displayed (VID, core clocks, core usage, core temps, cpu package power, PL1 and PL2 values, the limit reasons: core #x thermal throttling yes/no, core #x power limit exceeded yes/no, ia package-level rapl/pbm PL1 and PL2,PL3 yes/no) and expand the window with the arrow icon <--> so all the relevant data is readable,

-start Prime95 small FFTs test AND click the clock icon in the hwinfo sensors window,

-capture a pic of the hwinfo sensors right after the test started and one after ~8 minutes BEFORE you stop the test.

 

With the high performance plan it isn't normal to downclock to 800mhz even in idle.

It is possible that you power/performance settings are not managed by the win power plan.

Test again, but this time open TS, untick everything in the main window except speed shift epp. Change the value from the default one (128) to 0.

 

1) You can try to clamp this register by setting the clamp bit to 1 with something like '00018370' if the 1 is not there and see if that prevents changes to his register.
If this is not the clamp bit, ask @Vistar Shook what bit is the clamp bit.

2) You could try to disable Intel Thermal Velocity Boost to see if this prevents this from happening. But no one has been able to unlock the Bios to access this setting. It's possible on MSI Titan GT75 easily with AMIBCP 5.02 and the newest FPTW64 on the Intel management engine thread on win-raid.com forums. But Alienware seems to have some sort of Bios Guard (NOT the same as Bios Lock, which is just a bios option that can be disabled via dumping the UEFI bios strings and using RU to disable bios lock), which means attempting to forcibly unlock menus or flashing a modified APTIO capsule will give you a permanent brick. So you can't disable Thermal Velocity Boost.

You shouldn't have bought a BGAbook.

3) You could spoof the CPU into reporting less than 30W at full 12 thread load, by setting IMON SLOPE to 50 or lower and IMON OFFSET to -31999, but you can't even unlock your bios menus to try that. *IF* UEFITOOL could read the main alienware bios somehow and you could somehow find the "SETUP" capsule, extract it, and then open it with Universal IFR extractor successfully, you could get the hexadecimal offsets for IMON SLOPE and then boot into RU and forcibly set IMON SLOPE to "32" in hexadecimal, which is 50 in decimal, but good luck with that.

You shouldn't have bought a BGAbook. :/

4) The PECI power limits can override anything it wants, anytime it wants. You shouldn't have bought a BGA cancerbook. A 15" Clevo LGA TM model with an 8700K and Prema Bios from HIDevolution would have solved all your problems.


@Papusan

 

I think AW R5 BIOS is in more locked down than mine.
I am guessing IMON slope tweak might work using RU but unsure about its success on AW BGA Books.

 

How to disable Intel TERMINAL VELOCITY boost on a BGA Cancerbook
(MSI GT75 example):


Of course you can just avoid all this by not buying a BGAbook with Thermal Velocity Boost

 

Raz8020 where is the BIOS option to set the power limits to 110w? I cannot find it

 

Here is my cinebench results. Hwinfo started after started the benchmark. Score of 1109 but cores average speed seem to be at 4.1GHz. Something still limits frequencies? Throttlestop is removed.

https://imgur.com/a/v4EOmRe

 

I don't have an AW so I don't know the BIOS menu for this model, but I'm sure that @doofus99 or @captn.ko can tell you where to set the power limits in the BIOS.

In the previous posts, doofus99 mentioned that he has two options in the BIOS: one for 45w PL1 with 90W PL2 and another one for 110w for both limits:

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-136#post-10760362

 

Another Cinebench test without running hwinfo at the background. Score now at 1210. I did not change any CPU frequencies. I wonder if Hwinfo drops frequencies by its own. But I do not know any other way to see temps

https://imgur.com/a/lP0U8zI

 

I just finished repasting and repadding my 5th replacement with Kryonaut and fujipoly pads and still getting 100 degrees max on cores 0,2,4 after 15 min occt with .140 under volt -- insane. I'll be requesting a permanent refund tomorrow. I give up. I'm soo sad.

 

I'm really praying I get the rare unit that works perfect out of the box-they said they would especially test it to make sure it doesn't overheat but I doubt it. It's infuriating because when it works, the machine is absolutely sublime...but more often than not, you get 100C and stuttering

 

im using 1 half of those alphacool heatsinks, put a piece of thermal arctic pad under it and used scotch 33+ tape to keep the whole thing together, worked quite nicely for me, went from hitting 98c on the memory controlller down to 80s or so iirc when writing large amounts of data

 

way to low. Can You post CPU package power the next time... I think there is a power limit. CPU Performance mode on?

That's much better but still a bit low. What clocks do you had? again 4.1?

maybe you can post your results in this form:


then we don't have to ask for additional information so often and can help you better

once you activate the CPU Performance Mode the limits are set to 110W automatically. Make sure you haven't activated the Fan silent mode in the AW Command center (it limits the cpu to 45w again).

out of the box the very most of them run hot under high load. With a bit undervolting and limiting the clocks to ~3.5 - 4ghz the R5 is very usable without cooking the hardware.

 

You can wait for 2-3 years since octa cores and beyond from Intel camp will definitely throttle the moment you turn ON the PC or even worse BSOD on first Boot.

 

Lmao, you joke now, but watch this actually happen...

 

Yes, second score with same frequencies (I guess) because I did not change anything and the previous one was at 4.1 average. Only difference that I did not run Hwinfo at the background.

Ok, I will post results like that. I will make sure CPU performance is selected in bios. Fans can be in performance mode or have to be in full speed (I never have them at full speed but performance) ? How to do you show those diagrams with cores info?


By the way, Alienware command center comes with error very often, like no overclock profiles (but cannot download because it says I am not connected to internet but I am, or some of the menus missing (so have to reinstall). I also saw two ACC drivers (yes two of the same) in device manager and one having problems. So, I will first try without Alienware command center installed.

 

full speed or performance is fine.

Use HWinfo to create a log (the butten with the green plus). open this log with generic log viewer.
https://www.hwinfo.com/forum/Thread-LogViewer-for-HWINFO-is-available

 

Mine run (with a noname amazon NVME heatsink) about 79C max while gaming but I have seen them running up to 88C while, benchmarking disks. I already ordered EK-WB heatsink. While idle or doing normal things temps are below 40s but problem is throttle that could be due to high SSD temps during gaming.

 

Thats ok. how can you fit all those sensor readings in the screen? Did you change resolution or something? I cannot

 

haven't changed anything... 1440p panel, 100% res scale. Do you run 150%?

 

It was automatically set at 250% (recommended) for my screen from windows. Ok, if I change to lower I can fit more readings.

 

Ok here is another run. CPU runs at 4.1 again. I know I did not include log viewer but included more sensor readings from hwinfo

https://imgur.com/a/naEobGg

and here are my BIOS settings. I do not know if I have to enable overclock profiles. It seems that with overclock disabled it runs at 4.1GHz?

https://imgur.com/a/WUHm3SJ

 

just try if it improves it.

Can you run OCCT with same sensor settings?

 

This is running at 3.1GHz... No need to run longer. It does not go higher.
https://imgur.com/a/4HYAi06

1)BIOS settings are posted above
2)Alienware command center is not installed
3)throttlestop removed
4)Intel XTU installed but did not touch anything in it.

By the way temps are without any unvervolt as you can see

Frequencies continue to drop to 800MHz during idle with windows high performance. I am getting crazy here or I am doing something wrong?

So, is there a power limit option in bios or not?

 

have you installed an older bios version?

 

No, but this what I am going to try next. I will try your version

 

its strange that you have only 3.1ghz with 65w while your temps are ok. Powerlimit is fine so the CPU should boost up to 110w / or thermal throttle

 

It doesn't show any power limitations in this run (it does on the CB run when you reach 95w, but that's only the max and since all your minimum and current values are with no load, we don't know what happened during most of the test, that is why I mentioned the steps for testing in one of the previous posts, so all the relevant data is displayed).

Your power limits appear to be at 110w (as you can see in the pic you posted), but it tripped the power limit in the CB run (this shouldn't have happened if the 110w was used, but it seems you are also limited elsewhere).

Disable XTU (make sure it doesn't appear in the running services) and and USE TS to manage the speed shift settings (your win high perf power plan settings are not being used since it down-clocks to 800mhz in idle).

You appear to have a similar conflict for power control to the one from here: (read all the posts from that page, you can also try to display the hidden power settings by editing the registry):

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-813

Just part of the limit info is displayed on you pics, those that I mentioned in my previous post are missing.

If the info doesn't fit in the window, expand it further with the arrow icon <--> (you don't have to capture the score in the same pic).

The pic should be taken during the test, just before you stop it (so the values from the current category are relevant).

You don't have to start hwinfo after you started the test, but you have to press the clock icon right after the test begins.

 

I installed previous bios 1.1.6 to start from there again. OCCT test runs now at 3.5GHz (3.1GHz with latest bios). If I format, is it going to remove whatever is causing the restrictions? I really want to avoid it but it looks like we dont know what it is causing the restricitons

I already tried unticking everything from throttlestop main screen and ticked only speedshift with value of 0. Nothing changed

By the way here are my throttlestop options and the high performance advanced settings.

https://imgur.com/a/Bon5uO7

 

-Bios set at CPU performance which according to captn.ko sets the power limit to 110W. There is no other power limit setting in R5 bios
- I have done all throttlestop settings as described
- cpu performance selected in bios (no difference with full speed according to captn.ko)
- all others done

Check my screenshot just after starting the test

https://imgur.com/a/3YW05ei

Then Screenshot just after finishing the test

https://imgur.com/a/m5ZRcVA

I attach my Hwinfo log too. Stopped before stopping the test

 

Here is the prime95 hwinfo log (started logging after starting and stopped before stopping the test). Any help is appreciated thank you both captn.ko and raz8020

 

So, I didn't really do a good job applying the thermal paste and my temps escalated quickly, since I was too excited for this new Alien.

Did a repaste and opened up, here are a couple of pics, requested by @alexnvidia and a few others. The heat sink is made by a manufacturer called ROBIN, instead of CCI and SUNON. Not sure why Dell decided to include that manufacturer. Also, if you noticed, the tension arm causing the uneven heat sink temperatures look different to the R4.


Looking good. You can see they use bigger heat pipes as compared to the 17 R4's 1070 heatsink which were smaller.


Really Dell/Alienware?! TOOTHPASTE AGAIN?! That's nonsense!!! Stop using it!!!
That aside, I noticed that the 0.1mm thermal pad was removed. I think so as to allow the heat sink to sit properly on the CPU die, unlike the R4.


Dumping the Kyronaut on the CPU and GPU dies, as well as the heatsink (Don't mind me using quite a bit of thermal compound here, HAHA


And as a last minute inclusion, I added a tiny copper heat sink on the PCH as well as some electrical tape to prevent the heat sink from shorting the PCH. Thanks @iunlock !

Also, if you noticed, I used different screws I found from my laptop stash that are longer, so I could screw the heat sink down tighter and more evenly.


And here are the CPU temps. Running Furmark GPU Stress Test and Furmark 12 Thread CPU Stress at the same time. Don't see any throttling other than power throttling due to the 240W power brick. Also, look at the CPU temps, pretty even, despite the tripod CPU heatsink design eh and I am using a Kyronaut Paste.

I live in Singapore and its a pretty warm country here. So my room temperature was about 32 C when I was testing the system. I believe anyone else here living in countries that are cooler or having an Air Conditioned room will get better results. The laptop is sitting on top of a Cooler Master U3, just so that it gets more breathing room. I also did an undervolt of -100Mv to reduce unnecessary heat (without undervolt, its 5 C warmer) which also doesn't affect performance. CPU TDP around 70-75W


GPU Temps. Not bad eh? TDP is 130W now. Unlike 120W before (A 1070 GPU for laptop is rated around 115-120W, according to Nvidia's site, not 150W)

Ran test for about 30 minutes so far and still stable.


Cinebench scores.

And, no, I didn't want to use LM paste on this Alien, since this Alien regulars heads outdoors. Also because Dell Singapore is pretty stringent on warranty, I didn't want to get my warranty voided. Even so, while adding LM will most likely bring lower temps, it won't make any difference for my model. But looking at people here who got the i9 1080 model, yikes, looks like lots of issues.

 

doesnt look too bad for me...



...mhm

 

@Aristotelhs2060

The situation isn't better but we have more info (so we have a better base to work with).

Your PL limits don't actively change like they did in doofus99's test.

The only limitation that is displayed is EDP Other, which means it reached the current limit (in Amps).

Before you raise the current limit, you should undervolt with TS (in the FIVR window) to see if you still trip the limit with a lower voltage.

You should also edit that registry position (from the throttle stop guide) so your hidden power settings are displayed, after that you can manage them according to your preferences (you can disable windows control and use TS speed shift or you can disable speed shift in TS and manage them in the win profile).

After you undervolt, post a pic from the FIVR window to see if a different sst values doesn't appear there.

Try -100mV for both cpu and cache and test again with prime.

 

Now I have another issue with alienware command center. It cannot download the predefined overclock profiles (they were there before but suddenly got this error). Download in fusion does not work and of course I have internet connection.. Uninstalling-Reinstalling both ACC and OCC did not work. If I do not fix that I cannot go on because I think ACC interferes with bios settings.

https://imgur.com/a/kAeZtfG

 

Ok, I run cinebench again with -100mv for cpu and cpu cache using throttlestop. Alienware command center fusion does not work. So, maximum GHz at about 4.3GHz. Without HWinfo running at the background, I get higher scores. Have a look below..

https://imgur.com/a/O5R6sQN

 

as long as we dont see any data its a useless info for us... HWinfo please. what have you changed to get higher score?

 

only undervolt -100mv for CPU and -100mv for CPU cache.

here is the log for prime95 with undervolt

 

here is log for cinebench. score of 1292 when hwinfo is running at the background (-100mv cpu and -100mv cpu cache undervolt)

 

The score is better, but still low for 4.3Ghz, but you're not too far (compared to the previous instances) from the "average" score.

ACC is alien command center, but I don't know what OCC is for AW!

Do you really need those OC profiles (since you can do everything better with TS) or even the acc?

I'll check those logs in a min (I can only view them on a mobile device right now).

 

In prime, you were limited by EDP again, but this time you managed to exceed 100w with a constant 3.8Ghz, you were also PL2 limited.

 

Ok, what can I do for that then?

OCC stands for overclock control center. It is a needed component for alien fusion to run. Alien fusion is the overclock menu of ACC. If there are no default profiles, it does not allow you to use it or create your profiles. It is just disabled even though you can still see live frequencies for CPU and GPU. Seems like I have to manually remove ACC files. ACC has been a pain. I have no idea why it is giving me all those issues.