NCASE M1 v5 Build

Discussion in 'Desktop Hardware' started by TBoneSan, Jul 16, 2017.

  1. TBoneSan

    TBoneSan Laptop Fiend

    Reputations:
    4,290
    Messages:
    5,520
    Likes Received:
    5,645
    Trophy Points:
    681
    Lol sorry I haven't gotten back to you guys. I've been busy with work and all of my have been from my phone. I haven't forgotten about y'all. I just wanted to wait to reply to you with more detail from a keyboard. Will do as soon as I get some home time :)
     
    Robbo99999 likes this.
  2. TBoneSan

    TBoneSan Laptop Fiend

    Reputations:
    4,290
    Messages:
    5,520
    Likes Received:
    5,645
    Trophy Points:
    681
    Hey bud. Yeah so just to clarify. It's really hard to be specific with the gains I had from Metal on the GPU core since I never measured it accurately or at all to begin with. Mainly because GPU Boost 3.0 always has clocks and fans bouncing around based on temp intervals. What I did notice though was that at stock clock on long gaming sessions the fans were around 75% prior. Now they are around 49%. However I can't say emphatically that was due to the Metal because I didn't do due diligence in testing before and after with controlled variables. My hunch says it was down to the metal though.

    Regarding getting the ball of liquid to break off from the syringe. I usually aim it at the underside of the IHS in case I overshoot or goof up - I've had CLPro and TG squirt out of the syringe before from accidentally applying too much force. So I usually get acquainted and spread on the bottom of the IHS first.

    You can push the ball off with a Qtip too. I personally like the black ones as I find it easier to see the metal. The spread doesn't have to be as thin as possible - you can and I recommend to be a tad more liberal, but it has to be thin enough so that gravity won't allow it to gather and create a pool big enough to roll off. I usually hold the machine up on it's side for a minute or so to get an idea of how it could behave in the worst possible scenario.
    If you're really worried about it dripping off on the motherboard, just cover the surrounding areas with heat resistant tape and you'll have piece of mind too. Don't worry about damage too much. Unless you're a like THIS clown (nice find Paps @papasan), which I'm sure you're not, you'll be fine on all your Metal ventures :D

    Nothing worth writing home about yet my friend. Here is a bone stock GPU run with Powersettings maxed. At the CPU at 5Ghz and the Ring at 4.7 and RAM shaved back just under 4000Mhz.
    Room for improvement here. But you can also check out the temps.
    I also highlighted the MB temps for you @Robbo99999

    [​IMG]
     
    Last edited: Jul 28, 2017
    Robbo99999 likes this.
  3. Robbo99999

    Robbo99999 Notebook Prophet

    Reputations:
    2,880
    Messages:
    4,881
    Likes Received:
    3,038
    Trophy Points:
    331
    Thanks for the Liquid Metal tips, I'm fairly confident I can use it on the GPU without ruining my GPU, but I do have reservations about it staining & adhering to the copper heatsink - although my curiosity to see how much it would lower temps might get the better of me, I don't know yet!

    Your motherboard temps are good!

    Have you got any tips on RAM overclocking? I've historically been running with the RAM timings tightened to their max on my XMP 3200Mhz RAM, and recently I tried trading some of the tight timings for an increase in Mhz, but I couldn't really get very far - I only ended up with 3333Mhz and at looser timings (with no performance improvements, some slight negative), I couldn't seem to get any higher in Mhz. I tried increase DRAM voltage in various stages all the way through to 1.5V, but DRAM voltage seemed to make no difference. How far have you overclocked yours from stock, and what specific variables did you find were the key to hitting higher clocks on the RAM?
     
  4. TBoneSan

    TBoneSan Laptop Fiend

    Reputations:
    4,290
    Messages:
    5,520
    Likes Received:
    5,645
    Trophy Points:
    681
    Yeah the temps are surprisingly good all round. The ram I have is 3600. I used only an XMP to get to 4004mhz. But I had to up the voltage to 1.37 to get stability. The next stepping up was aroun 4137 and I couldn't even get the machine to boot with 1.45v. I should have tried 1.5v for science but I probably wouldn't have kept them there because of the diminishing returns.

    I'm an absolute RAM timing bimbo though.I might try and tighten the RAM up at 4000Mhz and call it a day.
     
    Last edited: Jul 30, 2017
  5. Robbo99999

    Robbo99999 Notebook Prophet

    Reputations:
    2,880
    Messages:
    4,881
    Likes Received:
    3,038
    Trophy Points:
    331
    I've spent a lot of time since my last post trying to overclock my RAM further. I have learned that if you want to increase frequency you may need to increase CPU SA and CPU IO volts (which I found to be the case) - I tried to get 3433Mhz stable with this knowledge but it helped to get it only borderline stable. I've now backed it back down to my previous 3200Mhz and manually reduced CPU SA & CPU IO voltage below the levels that the XMP profile auto sets on my motherboard (why use more voltage than necessary), and I've also tightened timings from CL15 down to CL14 by increasing DRAM voltage from 1.36V to 1.4V - currently testing stability with HCI Memtest, looks stable now. I've read a lot about RAM overclocking in the last day and it does seem that 3200Mhz CL14 RAM is the sweet spot in terms of performance and how little comparative CPU SA and CPU IO voltage you need - so I think is probably a finalised RAM overclock for me until I get really really bored & have way way too much time on my hands - RAM overclocking takes ages!! I've now got my RAM at 3200Mhz 14-15-15-34-250-1T @ 1.4V, CPU SA @ 1.232V, CPU IO @ 1.16V. (Previous XMP profile auto voltages for CPU SA were 1.288V and CPU IO at 1.248V)

    I learned quite a lot from this thread:
    http://www.overclock.net/t/1569364/official-intel-ddr4-24-7-memory-stability-thread

    EDIT: Argghh, the CL14 setting wasn't stable, failed with just a frozen screen on a Prime95 Blend test after about 45mins. Gone back to my previous RAM settings in sig - but have left it with the reduced CPU SA & reduced CPU IO voltages. Will do further testing with Prime95 to see if stable, because this reduced SA IO voltage setting was HCI Memtest stable to 400% on 8 threads.

    EDIT 2: Weird, saw some strange artifacting playing Mass Effect Andromeda, never seen that before with my GPU, might be because I've lowered the voltages for CPU SA and CPU IO maybe. Humph! Back to my very initial setting of auto controlled voltage for CPU SA & IO! To be fair I've tried tweaking RAM overclocks at least twice before on my system, but I always end up at the same 3200Mhz & slightly tightened timings - think I've probably wasted enough time on RAM overclocking now - mental note: don't tweak the RAM in my system anymore!

    EDIT 3: Possibly the final edit! Wasn't the lowered CPU SA & IO causing the artifacts, even happened at previous settings, and removing GPU overclock didn't remove artifacts - must be a bug in the game, only happens in one specific spot in the game. Ah well, will lower voltages for CPU SA & IO again then - praps I therefore managed at least one positive from my recent RAM overclocking adventures then!
     
    Last edited: Jul 30, 2017
  6. TBoneSan

    TBoneSan Laptop Fiend

    Reputations:
    4,290
    Messages:
    5,520
    Likes Received:
    5,645
    Trophy Points:
    681
    Thanks for the link my friend. Yeah I've had my Windows install get botched up a few times recently.. I'm not 100% it's RAM but my suspicions is it's due to fiddling around with settings / OCing - although I never get errors using TM5. It hasn't been the end of the world using a Macrium Backup to another SSD but it's still somewhat troubling.
    Yeah RAM takes a lot of time to get right, it's a PITA but it's always hard saying 'no' to free performance on the table :D
     
    Robbo99999 likes this.
  7. Robbo99999

    Robbo99999 Notebook Prophet

    Reputations:
    2,880
    Messages:
    4,881
    Likes Received:
    3,038
    Trophy Points:
    331
    I was gonna say use Macrium Reflect but then I kept reading & saw you use it! Yeah, if you don't mind taking the time to overclock your RAM and then test it thoroughly then it's worth doing, but only if you realise that it can be frustrating & all for nought! It's easy to do an initial rough tweak to get a little chunk of extra performance, but I wouldn't bother going any further to try and fine tune a RAM overclock - I gained nothing subsequent to doing the first sucessful rough tweak that was thoroughly tested.

    Hey, like you did, I put some liquid metal on my GPU a couple of days ago. I got a 3 degC temperature drop in comparison to Kryonaut the first time I tested, but on subsequent runs it was only 1 degC temperature drop in comparison to Kryonaut - so somewhere between 1-3 degC temperature drop by using Coollaboratory Liquid Ultra. Since then I swapped out the 2 stock Phantek case exhaust fans and replaced them with Arctic F12 fans that I had lying around - that gave me a further 2 degC drop in GPU temperature and 1 degC drop in case air temperatures. I then created some extra rear ventilation near the GPU by removing the PCI slot covers at the back of the case and replacing it with a 'fine' mesh - this mod enabled the GPU to stabalise at the next boost up (+13Mhz, lol!!) while staying at the same GPU temperatures - sucks in a little cold air from outside the rear of the case (GPU fan is only an inch or so from the mesh). So, when combining all these little gains over the last couple of days (liquid metal, more efficient case exhaust fans, PCI slot cover mesh mod), I lowered GPU temperatures by 4 degC while at the same time allowing the GPU to stabalise at 1 extra notch of boost and decreased case air temps by 1 degC - this is not to mention a previous GPU back plate cooling mod & repaste with Kryonaut that I did a while back now that had dropped temps by 6 degC in comparison to stock. If I add it up all these little changes since I first built my rig back in November they've resulted in a 10 degC lower GPU temperature while at the same time boosting a little higher - ha, it's a bit like the F1 philosophy of hundreds of miniscule improvements that add up to a big sum!

    @TBoneSan , because I'm editing my post so you might not see it if you've already read my post, did you ever get round to overclocking your GPU? I know you've got the Founders 1080Ti, so maybe there's not much thermal headroom to make practical use of it, but if you overclock without adding voltage it will still make your GPU more efficient for any given performance level, because you're shifting the Mhz up at all the pre-existing voltage points on the curve - so theoretically you're getting more performance per Watt.
     
    Last edited: Aug 2, 2017
  8. Robbo99999

    Robbo99999 Notebook Prophet

    Reputations:
    2,880
    Messages:
    4,881
    Likes Received:
    3,038
    Trophy Points:
    331
    TBoneSan likes this.
  9. TBoneSan

    TBoneSan Laptop Fiend

    Reputations:
    4,290
    Messages:
    5,520
    Likes Received:
    5,645
    Trophy Points:
    681
    Yeah there's not much of a difference between all the metal pastes. I'm not too fussed which one I get hold of, even Pro is all good in my experience. I probably prefer Kyronaut right now though.
     
  10. Robbo99999

    Robbo99999 Notebook Prophet

    Reputations:
    2,880
    Messages:
    4,881
    Likes Received:
    3,038
    Trophy Points:
    331
    Kryonaut is not a metal paste, your post makes it sound like you think it's a metal paste - my point was that a non-conductive non-metal paste like Kryonaut is only 1 degC hotter than a metal paste like Coollaboratory Liquid Ultra.
     
Loading...

Share This Page