Jump to content

WereCat

Member
  • Posts

    17,188
  • Joined

  • Last visited

Awards

This user doesn't have any awards

Contact Methods

  • Discord
    WereCat #0437
  • Steam
    http://steamcommunity.com/profiles/76561197999269752
  • Twitch.tv
    https://www.twitch.tv/themightywerecat

Profile Information

  • Gender
    Male
  • Location
    Slovakia
  • Interests
    PC, Music, Books
  • Member title
    Bad Apple

System

  • CPU
    Ryzen 7 5800X3D
  • Motherboard
    ASUS X570 Crosshair VIII Hero Wi-Fi
  • RAM
    64GB (4x16GB Dual Rank) Micron E-Die 3200MHz CL16 (stock)
  • GPU
    RX 6800 XT Sapphire Pulse
  • Case
    Lian-Li O11 XL
  • Storage
    Crucial MX500 500GB, 2x XPG SX-8500 Pro 1TB M.2 NVMe, Seagate Barracuda 2TB Advanced Format, Samsung 850 EVO 500GB 2.5" SSD
  • PSU
    EVGA G3 750W
  • Display(s)
    Oculus Rift, AOC AG271QG (1440p 165Hz AH-IPS G-Sync), 2x ViewSonix VX3211-2K-MHD (1440p 60Hz IPS)
  • Cooling
    Arctic Liquid Freezer II 360mm CPU cooler + 7x Arctic P12 PWM case fans
  • Keyboard
    CM Storm Ultimate with Cherry MX Blue, Steam Controller
  • Mouse
    Razer Basillisk Ultimate (wireless) + dock, Logitech G502 Spectrum
  • Sound
    Sennheiser HD650, S. M. S. L. M6 Mini DAC, SONY WH1000XM3, BlueYeti (black) microphone
  • Operating System
    Windows 11 PRO x64-bit, Linux - Pop!_OS x64

Recent Profile Visitors

45,587 profile views
  1. Can't really compare that as both systems are wildly different. I get what you're saying with the avg FPS being higher on your friends PC but that can be a run to run variance as I mentioned the game performance varies wildly depending at which stage of match youre in, where on map you're located and how many people are nearby. You would have to test the identical scenario every single time to make a more fair comparison as you can have like 50FPS+ run to run variance in PUBG depending on how you test it. His GPU is slower so he will be pushing it harder and thus will have higher GPU usage most of the time than you. His CPU is basically the same as yours. The extra cores you have are not really relevant for PUBG and you will hardly notice a 100MHz difference in boost performance.
  2. Holy shit, this is so stupid it's genius 😄

     

     

     

  3. PUBG is a CPU and memory heavy game so CPUs like Ryzen 7 5800X3D or 7800X3D will work best due to the large cache pool... almost all battleroyale games behave the same way and performance will always vary wildly depending on which stage of the match you're in and how many players are nearby. There is not much you can do besides overclocking your CPU and memory to get some more performance.
  4. At worst few weeks until the temp starts to creep up due to pumpout after thermal cycling at best 1y-2y before you have to repaste.
  5. For all none of you! 

     

    Fedora 40 has been released! 

  6. It may not be much... but it's nice to have

     

    image.png.a302334f465c6fccb2164dd718f7a699.png

    1. Poinkachu

      Poinkachu

      8 minutes ago, WereCat said:

      but it's nice to have

      A megaphone icon ? 😜

    2. WereCat

      WereCat

      1 minute ago, Poinkachu said:

      A megaphone icon ? 😜

      The notification icon... duh! 😄 

    3. Senzelian
  7. It's a bad contact because there is a glue and also because there is a lot of paste spread thick on the die. You want as thin spread as possible otherwise you are just using the paste as insulator instead of thermal conductor. No, knife is not fine... unless you are sure you wont do a single accidental bad move and either scrath the PCB or yourself.
  8. No. TERRIBLE contact, too much space between IHS and CPU. Remove the glue from CPU... Use plastic credit card or somethingike that to remove it.
  9. No, the whole point of deliding is to create as small difference in Z height between CPU die and IHS and thermal paste vuscosity is not good enough and as mentioned, will pump out over time due to thermal cycling. I have experience with LM, graphene pad and PTM 7950. I say go with PTM 7950, almost all the benefits of LM without the hassle and risks. Also read my edit on previous comment.
  10. Yes that will work. But have in mind that KryoSheet is also electrically conductive and is NOT reusable and it's not really easy to apply it as it can easily move around. Alternatively you can use PTM 7950. Btw the 4770k I delided back in like 2015/2016 still uns to this day in it's overclocked state with the original application of LiquidMetal (now in friends PC). - I just removed the glue from both IHS and CPU. - Applied nail polish for the small SMD components around the CPU die. - Applied LM on the CPU die AND on the IHS. - Put the CPU without IHS carefully into the MB socket THEN placed the IHS on the CPU. - Then carefully clamped the CPU in the socket while pressing on the IHS with a finger so that it wont slide - Mount cooler and done (used thermal paste on cooler)
  11. Offset mounting for LF II 360 results are in!

     

    Testing:

     

    Linpack Xtreme 10G for 1h 30min to reach steady state then immediately disable all logging and run Linpack Xtreme 8GB 5runs and check performance then immediately run CB 2024 10min and check performance.

     

    Can't control room temp but in both tests the room temp was anywhere between 20C to 22C with it getting hotter towards the end of testing. Fans were set to 100%.[

     

    Baseline was with Arctic MX4 that was applied last time around 2y ago without the offset mount.

    New test is with PTM 7950 which was applied today with an offset mount.

     

    I'm using 5800X3D with  Curve Optimizer offset of negative 25 with LLC lvl 3 on ASUS X570 Crosshair VIII Hero Wi-Fi.

     

    PTM7950 application (didn't have enough for full IHS coverage):

     

    Spoiler

    image.thumb.jpeg.cccdcc45637a94842c7d89bae719a31e.jpeg

     

     

    Arctic MX4 benchmark results:

     

    Spoiler

    8GBMX4.png.04377442ea9f8f1cea35ad52f84bfc4e.pngMX4.thumb.png.da6079687555f63690430843a5eaf46c.png

     

    PTM 7950 benchmark results:

     

    Spoiler

    8GBPTM7950.png.9bfee3f9a734aaf56bcbb7b2553158c2.pngPTM7950.thumb.png.c82f4bf962348200a743eba8dbe58422.png

     

    Temperature and clock comparison of the last 30min of Linpack Xtreme 10GB:

    Red = MX4 without offset mount

    Green = PTM7950 with offset mount

     

    Spoiler

    Snmkaobrazovky2024-04-21163205.thumb.png.2a464e879189a16423b178cd2a3e376a.png

    Looks like the max temp is unaffected but CPU get's cooled much faster in between runs by around 10C.

    The avg clock speeds is also on avg 50MHz higher with PTM7950 and the offset mount.

     

    Overall the performance increased in every test which is nice and confirms that the avg increased clock is not just a result of clock stretching. I don't think this will have any effect on actual in-game performance as games don't stress CPU to this extent ever and for day to day use this makes basically no difference as well.

     

    For chasing extra scores for benching... worth it!

  12. I've just applied the Arctic offset mounting kit for my LF II 360 and running tests. I have also used the remaining PTM 7950 instead of the MX-6 paste they've sent me. I didn't have enough for full IHS coverage but it should be fine.

    1. PDifolco
    2. WereCat

      WereCat

      8 minutes ago, PDifolco said:

      Aaaaand ? :)

      to be continued ... 😄

  13. That is a very old video. You most likely have Curve Optimizer settings in the BIOS now. Just set a negative offset of anything between 0 to 30 and make sure to test for stability. 30 will be the best if you can actually run it. I had to settle for 25 with LLC lvl 3 on my ASUS motherboard as default LLC was causing voltage droops and preventing me from going above CO 20.
  14. Yes but also the clocks are temperature sensitive so more power = more heat and in case of X3D you're most likely to run into higher heat load and thus get clock regression or clock stretching so it's not worth it unless you have really good cooling. And also it will do nothing for gaming as these chips sip power when playing games. It's in loads like Linpack Xtreme or OCCT where this will make the difference. Heck even Cinebench isn't that heavy for it to matter too much though it may affect the scores with weaker cooler.
  15. Just use Curve Optimizer. These CPUs are both power limited amd multiplier locked so there is no point trying to use PBO. You can get around power limit with skewing the CPU current Telemetry readings if you have option for that in the BIOS but that will not really help for gaming as you will never run into power limit when playing games.
×