New P71 owner mini review and rant

Discussion in 'Lenovo' started by PsyberEMT, Jun 16, 2017.

  1. ibmthink

    ibmthink Notebookcheck Deity

    Reputations:
    893
    Messages:
    1,887
    Likes Received:
    355
    Trophy Points:
    101
    I would expect it to perform much better than the Celsius, because the CPU cooling system of the Celsius is much smaller. The Celsius also has two separate cooling systems for GPU and CPU, while the P71 has a shared one, the CPU heatpipe is connected to the GPU cooling-system. This means that with pure CPU load, the GPU fan should also help to cool the CPU.

    The Celsius H970 showed abysmal CPU performance from the very first test, it was unable to achieve its performance-potential from the very beginning. The ThinkPad P71 Cinebench-score is visible in this review and it easily beats the Celsius H970 (despite having a nominally slower CPU).
     
    Last edited: Aug 29, 2017
  2. Mobius 1

    Mobius 1 what is quality control?

    Reputations:
    2,723
    Messages:
    7,791
    Likes Received:
    4,940
    Trophy Points:
    431
    Both are sad machines imho.
     
  3. ForestForTrees

    ForestForTrees Newbie

    Reputations:
    0
    Messages:
    1
    Likes Received:
    1
    Trophy Points:
    6
    Hello. I just purchased a P71 and came across this thread. Thank you to the OP and the other posters for such a well informed and interesting thread.

    I'd like to reassure potential purchasers of lower end P71s with some data of my own. As I write this, I am running a 4 minute CPU stress test in Intel's Extreme Testing Utility. I believe this is the same software program that the OP referred to in the original post and that he relied upon to find Thermal Throttling.

    For my laptop, I'm not getting any thermal throttling at all despite running the testing utility's CPU stress test.

    Presumably this is because I have a far less powerful unit, with only the base i7 7700HQ @2.8GH, 16GB RAM, etc. By usual standards, it's of course quite powerful, but it's no high end Xeon workstation.

    It's whirring along quite comfortably and the air coming out of the right fan is coming rapidly but isn't too hot. More precisely, XTU reports "Package Temperature" as 73 degrees Celcius. I've run the test several times, and haven't noticed any throttling at all.

    Results were similar when I ran the RAM test.

    Obviously, Lenovo should still be blamed for the disappointing performance of high end P71 workstations. Kudos to the posters here for pointing that out. However, purchasers of less high end P71s like mine will hopefully be reassured.

    Overall, I'm quite happy with it. Some windows updates failed to run and the start menu has exhibited some strange behavior, but that could be as much Microsoft's problem as Lenovo's. Other than that, it's been exactly what I'd hoped for. ...And huge.... definitely huge.

    Hope this helps others. It's not hard to install or use XTU if others want to try with their own machines.
     
    huntnyc likes this.
  4. Mobius 1

    Mobius 1 what is quality control?

    Reputations:
    2,723
    Messages:
    7,791
    Likes Received:
    4,940
    Trophy Points:
    431

    Run AIDA64 FPU only for at least 30 minutes and see if the CPU throttles.

    Try also with OCCT, Realbench, etc.
     
  5. Quicklite

    Quicklite Notebook Deity

    Reputations:
    149
    Messages:
    1,513
    Likes Received:
    6
    Trophy Points:
    56
    I'm not sure if Dell/Lenovo use those poor stick on "paste" still? Maybe the CPU will improve a bit with a repaste. Though it's awkward to have to repaste something nearly new / hasn't yet degraded in thermal performance. Ironically in the Lenovo ads I remember the presenters making a big deal of the dual fan system.

    Just a quick question. 7700HQ has a high TurboBoost frequency of 3.8Ghz too - apart from the lack of ECC support / smaller cache - it's not really different from the Xeon - is it?

    For some reason, I keep wondering - why the Xeon chips have a TDP of just 45W? (looking at the frequency - it's a pretty high TurboBoost for a laptop - reminds me of the older Extreme Edition Core i7 Quads - which had a TDP of 57W).
     
  6. PsyberEMT

    PsyberEMT Notebook Enthusiast

    Reputations:
    10
    Messages:
    19
    Likes Received:
    20
    Trophy Points:
    6
    As an update....copied from my Reddit post...

    The 2nd replacement unit that I got seemed to do better. Once I applied an undervolt it wouldn't throttle during an XTU stress test. A Prime95 stress would cause it to thermal throttle. I was relatively happy until my consulting workload picked up and I noticed I was getting thermal throttling again under extended geoprocessing loads and just recently under extended SQL loads. I decided to try and correct the issue myself by repasting.

    During the repaste process I discovered the cause of the thermal issues. There are 2 design flaws present. One in the heatsink screw and standoff assembly and the other in the thermal pads used.

    The screws they used to attach the heat sink assembly to the system board are designed in such a way as to prevent over tightening during assembly. They will only allow the heatsink to get within a specific distance to the board. That gap is too wide. When I removed my heatsink I noticed that the gap was so wide that there was an obvious leftover air gap where the TIM covering the GPU heatsink didn't make contact with the die at all. There is also a noticeable slant to the impression in the TIM from the GPU die indicating that it isn't mounting flush.

    Photos:
    https://imgur.com/N4J8mYr
    https://imgur.com/pTD5Ou1


    The thermal pads cooling the GPU voltage delivery components are too thin. Many aren't making contact with the heatsink at all and are obviously 1/2 the thickness they need to be to actually make contact. As you can see in the image above there are no component impressions in many of the thermal pads.

    The gap between pad and component is obvious when mounted:
    https://imgur.com/qgBGWj3

    The CPU die appeared to make somewhat better contact, but the amount of TIM left on the die and on the heatsink indicated that there was a much larger gap there than desired:
    https://imgur.com/H2m7xFs
    https://imgur.com/5nEFKgS


    The thermal pads for the CPU voltage components appeared to make some contact but were very sloppily installed and overlapped on some of the edges creating small gaps:
    https://imgur.com/35iGXOF

    I blasted Lenovo twitters until someone replied and escalated me to a case manager. Again. I am going to see what they'll offer to do. The reality is this could be easily fixed by removing the screws they used and replacing them with ones that would allow a careful, manual torque down onto the board to insure proper mating of the heatsink, TIM and dies. It would also require replacement of the thermal pads with pads of the appropriate thickness.

    I will likely just make my complaint with the case manager and, unless they offer to actually fix the design, I will fix my PC myself. When I had to wait on the 3rd unit for almost a month they upgraded my GPU from a P4000 to a P5000 at no charge and gave me a 10% refund on the entire order. I feel like they've done what they could monetarily to make things right. I could probably get a refund but at this point the unit is setup the way I want and with the exception of the heat issue it works the way I want.

    The tricky part will be finding the proper screws. If I go that route I will update here and my original thread on the Lenovo forums with the screws I used and pad thicknesses used.

    Would I suggest someone buy one of these brand new today? No.

    I have serious concerns about the longevity of the GPU and CPU dies as well as the voltage components due to overheating. Sure you can extend the warranty and not worry about it but I don't find that an acceptable solution to what is an obvious set of design flaws coupled with questionable workmanship.

    For those that Lenovo has told it's "normal" for a laptop CPU to constantly thermal throttle...just no. Not in a million years. I wouldn't expect a consumer level laptop to be able to deal with the heat generated by something like Prime95 but I do have that level of expectation on a workstation specifically designed for high-end computation that is in this price bracket. I also have the expectation that the workstation will properly cool the voltage delivery components to ensure my investment doesn't prematurely fail.
     
    Last edited: Nov 6, 2017
    Mobius 1 likes this.
  7. Mobius 1

    Mobius 1 what is quality control?

    Reputations:
    2,723
    Messages:
    7,791
    Likes Received:
    4,940
    Trophy Points:
    431


    Pic broken
     
  8. PsyberEMT

    PsyberEMT Notebook Enthusiast

    Reputations:
    10
    Messages:
    19
    Likes Received:
    20
    Trophy Points:
    6
    Had to convert to links, IMGUR just wasn't cooperating today.
     
  9. Mobius 1

    Mobius 1 what is quality control?

    Reputations:
    2,723
    Messages:
    7,791
    Likes Received:
    4,940
    Trophy Points:
    431
    That's even worse than I expected.


    2nd pic: seems like a part of the copper slug is extending towards the vram area, causing the thermal pad to have an uneven contact surface with the heatsink. Could be remedied with fitting two piece of different thickness thermal pad to compensate for it?

    Looks like the aluminum top left part of the heatsink is bent. Two right vram makes an impression but the left two doesn't. If that area is flat then the thermal pad would have the same imprint on all 4 vram...

    And the two large rectangular box you made is for the chokes I think, that's not too important to cool but still sad to see negligence.

    On a sidenote, did you check that the two FETs above the R47 chip and near the screwhole are covered with thermal pad?




    3rd pic: I see what you mean by the VRM thermal pad being too thin.





    4/5: I wouldn't say that there's too much gap, maybe there is, but you can't judge by the amount of thermal paste remaining as the "stamp" thermal paste for the CPU is already way too much from the beginning.

    What I'm curious about is if the mylar sticker on the CPU edges are too thick and causes poor contact between the heatsink and CPU.



    6: so they use the rollcage as the CPU vrm heatsink huh... Would be interesting to repad this one.





    @PsyberEMT I actually have a P71, 1505v6 and P4000. Thanks for the info about the internals. I'm planning a repaste of the machine and your post give a lot of insight towards planning.

    Ps: can you PM me how to get the P5000 upgrade? :D
     
  10. Mobius 1

    Mobius 1 what is quality control?

    Reputations:
    2,723
    Messages:
    7,791
    Likes Received:
    4,940
    Trophy Points:
    431
    @D2 Ultima @Mr. Fox @Papusan @Dackzy

    Might want to check above. After I'm done posting the review for my X1 Carbon today the next one would be the P71 (laptop being discussed in this thread).
     
Loading...

Share This Page