The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
Problems? See this thread at archive.org.

    m17x R3 and 580m - Keeping it cool with heatpipes!

    Discussion in 'Alienware 17 and M17x' started by charlie22911, May 16, 2012.

  1. charlie22911

    charlie22911 Notebook Enthusiast

    Reputations:
    0
    Messages:
    32
    Likes Received:
    2
    Trophy Points:
    16
    As I stated in the conference call thread, I believe Dell is content with doing nothing substantial to remedy our throttling issues.
    It is my belief (which I will try to prove best I can) that the design of the heatsink is NOT sufficient to cool the 580m, and that the situation cannot be resolved by shipping the exact same cooler with some new thermal interface material. Only by issuing a redesigned heatsink assembly will the problems delt with.

    It is my intention to make clear to this community and to Dell that their current "solution" is NOT working should no longer be pursued, and instead a newly designed heatsink should be demanded.

    I will keep this thread updated as time allows (I'm Military) with results. I welcome recommendations and constructive criticism.

    To start testing this theory I've added small heatpipes (due to clearance issues) to the stock cooler. I am using NO bios mods, and NO P-state clocking what so ever. Everything is 100% stock as it came from Dell (minus added heat pipes).

    These are heatpipes that were scavenged from an old laptop (passively cooled Pentium M).
    The heatpipes are using old thermal pads that were scavenged from various sources. No effort was made to clean them and they are in very poor shape and quality, this is only for testing; better pads will be used as a permanent solution.

    (Click thumbnails for larger image)

    Heat pipes were placed over various hotspots of this card (VRMs\FETs, chokes etc) and dump heat into the metal chassis of the m17x, while one unloads into the existing stock heatpipes.
    [​IMG]

    You can see two of the heatpipes through the middle vent.
    [​IMG]

    Space is tight, but there is no discernible bulge of the back cover. It took quite a bit of trial and error to achieve this.
    [​IMG]


    The first test to be run was a short run of MSI Kombuster, which showed a very nice improvement and most importantly; zero throttling!You can see in the graph that temps flattened out, I estimate a max of 1-2 degrees rise due to heatsoak of the chassis and heatsink assembly should this test have been left active for a longer duration (~15-20mins).
    (Click thumbnail for larger image)

    [​IMG]

    This is all I have done for now. I will try some overclocking tests at a later time to see just how robust this cheap (free for me) and mundane mod has made the existing solution.

    Again I welcome questions, recommendations, and constructive criticism!
    I also apologize for any grammatical errors, this was a bit rushed and could stand to be better formatted :eek:
     
  2. widezu69

    widezu69 Goodbye Alienware

    Reputations:
    3,079
    Messages:
    4,207
    Likes Received:
    168
    Trophy Points:
    131
    Actually, I reckon Dell's solution works fine. I challenge anyone to tell me other wise. With a good repaste, I max out @70C in over 30C ambient temps using Kombustor and a overvolt. The people we see with high temps are still rocking Dell's long life stuff that's sturdy but not amazing. PK-1, IC Diamond, MX-4 will reduce temps dramatically especially when applied right.

    I wouldn't want to dump any heat onto the chassis as it isn't metal. It is plastic. Your concept however is decent. The guys over at the M18x forums are working on a CPU heatsink with more heatpipes.
     
  3. charlie22911

    charlie22911 Notebook Enthusiast

    Reputations:
    0
    Messages:
    32
    Likes Received:
    2
    Trophy Points:
    16
    I assume you have the throttle fix done (Pstate\bios method)?

    This is my second heatsink, and I still throttled with it and an Arctic Alumina repaste with fur mark. If I were to remove these heat pipes it just throttles again.
    I do find it a little hard to believe you are dissipating near desktop levels of heat while overclocked with nothing more than a 3 pipe heatsink and a repaste. Or maybe you have a different revision cooler?
    A stock 580m has a TDP of 100w, even this measly 55w TDP core2quad HSF barely did its job and it dwarfs our cooler (old cooler\card pictured):

    [​IMG]

    Thermal conductance and surface area are the name of the game when it comes to coolers, and the stock cooler just does not have enough to deal with a 100W+ TDP.

    Also the interior chassis is in fact metal, while the exterior panels are rubberized plastic. There has been a slight rise in temperature above the touch pad, and temps are still down, I'm idling at 53-54c with the laptop in my lap.
     
  4. widezu69

    widezu69 Goodbye Alienware

    Reputations:
    3,079
    Messages:
    4,207
    Likes Received:
    168
    Trophy Points:
    131
    Thanks for the info, I hadn't realised that the interiors were actually metal.

    Oh I forgot to add, I have the back of the laptop elevated to allow better airflow. People say that elevating the back isn't a proper fix as it should work with it flat on the table. I agree to an extent but when I game, I elevate it for better ergonomics anyways so it doesn't really matter.

    I have the p-state fix, I helped create it. Furmark will probably cause me to throttle in hot ambient temps but I don't want to use it, it has been the culprit of many 580m deaths.

    The 580m although rated for up to 100w, does not emit that much heat. I've owned all six GPUs during the time I've had my M17x R3 - 6970m, 6990m, 460m, 560m, 580m and the 7970m. The 6900m cards are the hottest peaking over 80C, I've never peaked over 70C on any of the nvidia cards yet they are rated at higher wattages. Never go by numbers, always experience.

    I idle around 52C same as you :)
     
  5. charlie22911

    charlie22911 Notebook Enthusiast

    Reputations:
    0
    Messages:
    32
    Likes Received:
    2
    Trophy Points:
    16
    I will admit this is my first performance laptop, but I'm no stranger to overclocking\tweaking.

    Furmark\Kombustor are known to actually push cards past their TDP, prompting a bit of controversy when AMD implemented targeted throttling when it detected one of these two stress tests loaded in memory.
    It's also what started throttling in the desktop GPUs since many people were actually killing the VRMs on there desktop cards (mostly nVidia IIRC).

    With this in mind, I think it's safe to assume that with furmark\kombustor our cards are at least at there 100W TDP; this also explains the dead cards (I bet money its killing VRMs). This is why I used it in my "test", and I saw a marked improvement.

    Since you did have those different cards, did you get coolers with them or did you use the same one? I originally used the cooler that came with my 460m on my 580 before getting the new cooler (with extra copper), and its only slightly better but still throttled in furmark.

    EDIT:

    Its also possible that most of the benefit that I am seeing is coming from the heat pipes I've attached to the VRM area along the back of the card. That area gets very hot!