Jump to content
NotebookTalk

electrosoft

Member
  • Posts

    2,602
  • Joined

  • Last visited

  • Days Won

    111

Everything posted by electrosoft

  1. Future playtime (yet another project 😁) aka Project Dragonflight Hopefully your memory arrived @Mr. Fox
  2. I agree. Instead of having 3-4 models you then have numerous sub models within the primary models based on power. This GL66 with a 3070 clocks in at ~9400 gpu stock clocks on Time Spy with no overclocks so you're looking at best a ~10% boost over your 3060 which is not exactly an exciting boost.
  3. This generation with such massive power discrepancies on desktop vs laptop, the 3060 is the closest representation of its desktop counterpart. Throw in Nvidia playing all types of PL games within the same model and tons of cross over and things are all over the map. Some laptops don't even match their given specs. The GL66 in my sig that is supposed to have a 105w 3070 pulls 140w under load and gaming. Depending on your Time Spy score, you may need to jump to at least a 140w+ 3080 laptop variant. What is your Time Spy score stock no OCs?
  4. Yup, two deliveries now rescheduled for tomorrow too from an original Wednesday delivery date. On the other hand, a FedEx delivery for a new display I ordered due tomorrow showed up today..... Outside of World of Warcraft (which I usually pre-order the epic edition), I always wait for prices to drop for games. I just picked up Fallout 76' deluxe edition with all the goodies during Steams massive sale for $15.
  5. Luckily my crusted LM was on top of the IHS for the stock 10900k but it stopped the CPU from pairing correctly with the heatsink.
  6. eBay listing with racks of KPE 3090's mining for sale. I've never seen that config with all those 360mm AIOs in action top and bottom:
  7. Ditto for a few of my UPS orders too. A late trailer is causing the delay. Ah well.....
  8. lol, 145w nice! The PL on this MSI is actually 200w (!) but it hits thermal wall around 121w so far. Going to try to get a uv going but also adjust the PL down to 80w and start scaling by 5-10w and see where it hits the perfect equilibrium point so it doesn't throttle and drop so much then try to climb again, throttle rinse/repeat. The TG contact frame might be a major game changer if the socket is subject to the odd bending forces like desktops. If the retention system exerts the same amount of force as a desktop board on the laptop board which might not be as robust you might even experience even more from the TG than the desktop even.
  9. If the X15 had a beefier heatsink and cooling so much more performance could be extracted. Hopefully something pops up on aliexpress sooner than later. I ended up having to clean up a lot of garbage mistakes under the hood including several mis-aligned and double stacked thermal pads, incomplete thermal coverage and tape covering a memory chip for starters. After a few thermal paste replacements I ended up doing a shim mod and replaced the stock screws with another set. I was then able to switch to kingpin thermal paste. Opened up the extended menus in MSI BIOS but the H chipset is pretty limited. Next stop is Throttlestop. This is all without an undervolt yet. Those full fat 8+8 12th gen chips do add some extra oomph. 🙂
  10. Nice, make sure to post a screen shot to see the temps, pull, Vcore under load and score! 👍 Still tempted to give the aliexpress universal HS a try. It supports both 3000 and 2000 series cards and looks decently solid.
  11. Yep, always seeking cool hand lukes which usually (but not always) means a better binned chip overall to get better pull and thermals at stock. My 10900k isn't delidded and I'm using regular past too in that run (nanogrease extreme). For stock usage, Prema doesn't come into play as only a uv has been applied (~-82uv) So you those parameters for comparison. I had no idea you had a X170KM-G and I'm not sure why I didn't know. 😁
  12. Oh then we are in agreement. I use LM between the die and internal IHS. I usually tape off the area internally to match the die perimeter so it is exacting and put a thin coating of LM on both sides so run off is close to zero. I never use LM on my GPU dies either. I *will* experiment with kingpin and TGx versus nano if I have a good enough pairing but when in doubt it is always Phobya nanogrease too. I seal my IHS lids and I've never had the problem internally but externally IHS->Heatsink? It's happened a few times ugh.
  13. I don't overclock my laptops. My objective is to make sure they don't throttle and run at stock (or close to it) as best as possible especially DTRs. I want my memory optimized as possible and my desktop CPUs running as they should like a real desktop. I center everything around CB23 single, triple and 10 min stress test as I will never run anything that will push it harder than that. If I experience any type of power or thermal throttling I adjust accordingly to get it where it needs to be. 👍 I also try to get the fans to run silent as possible while doing it. Somewhere in this thread is a post of my optimized 4.9 all core with memory tweaked runs for Aida and CB23. I have it getting exactly what I got on the desktop with a few Z590 boards.
  14. KPE 3090ti's are sold out but it was a very limited run so it was to be expected. It would have sold out sooner if queued up members had picked them up. All other EVGA 3090ti models are $1499.99 including the hybrid: https://www.evga.com/products/productlist.aspx?type=0&family=GeForce+30+Series+Family&chipset=RTX+3090+Ti EVGA has pulled all 3090s and 3080s from their combo deal integrator and only 3070ti and below are available. Only three 3090 models are now available to buy for $1299 (FTW3 Black, FTW3 Ultra, XC). All the others are out of stock including the original KPE 3090. All 3080ti's left are 1099.99. 5 models in stock. All 3080 12gb are 799.99 gobs of those 🙂 Clearly they want to push the 3080 12gb as all the 3080 10gb are priced above that.
  15. This along with determining what are your objectives since games really benefit from optimized memory but if your daily driver frequency is 3.5ghz then you can have one set of memory timings fully optimized for D2D at that frequency then set it back to SPD and go ham on overclocking.
  16. Whew, I know your heart jumped when you caught those pins! I've been there before. 🙄 Nice, did you do a bin stepping analysis to see if 3.5ghz was optimal from let's say 3ghz up to 4.5ghz? I usually let mine run stock for gaming so it sits at 4.9ghz for playing WoW which is CPU hungry. Temps top out around 80c after hours of playing. You can probably bring that tRFC down and tREFI up depending on quality of memory along possible tuning the primaries. With good pairing pressure, liquid metal is superior in every way, shape and form. Next big project is my SL isn't delidded but I'm starting to get the itch. I think for now I'm going to clean up the original LTX IHS and recap the LTX so I can free up my BartX Nickel Plated IHS for the SL. The SL does equal if not better with its original lid as the BartX delidded and capped LTX. Im going to rebuild out my Z590 MSI for testing it with both caps and eventually the SL along with install 4x 128GB 2.5" SSD I have laying around to install Win10, Win11, Unbuntu and Kali on their own individual, isolated SSDs.
  17. Even downgrading the BIOS gives the same results? Luckily ($$) in these situations comparative test hardware usually gives clarity. 🙂
  18. I've had LM harden and cluster on top of a few Intel stock IHS and had to scrape/sand it off too. True about the soak. I just always knew as soon as temps went all cattywampus it has started to soak and would get in there a few days later and apply some extra. For laptops, I'm content with standard thermal compounds on top both the GPU and CPU even if it costs me some performance and degrees at this point.
  19. Having used LM on numerous laptops and some desktops, I'm only using it when one (or both) parts are nickel coated. It just isn't worth the long term hassle of soak and when too much air gets in over time the hardening and having to scrape or sand that mess off. I'll also use it for very short term benching to get the best results possible within a limited time frame before cleaning it off and switching back to traditional TIM.
  20. There's only so much you can do in the limited VRMs and cooling of the X170KM-G. Luckily our units are still under warranty but you will don't want to get slap happy. How did you pick 3.5ghz all core as your D2D frequency? I haven't had mine out in quite some time still. I was too busy modding and wrestling with the absolute crap show that is the stock hardware and firmware that is the MSI Pulse GL66 12700h. Initially the X170SM-G was blowing it out the water, but lots of work and internal tinkering has it now beating the X170SM-G (at its tweaked stock settings) finally. Fortunately the X170SM-G you can tweak the memory settings and lock in frequencies better and coupled with the 2080s, it smashes the 3070 mobile in gaming still.
  21. Your old 10900k actually looks like a decent bin too. Not as good as your new boy but I know the original ZtecPC binned chip I had couldn't do anywhere near those numbers. 5.3 all core and it ballooned to 340w+ on the desktop and bsod immediately on the X170SM-G. It was binned clearly for optimal 4.9-5.1 range. Your old 10900k also looks like it was working better with the memory slots and channels. Did you re-check to make sure all the pads are clean on the new 10900k and more importantly check your pins to make sure you didn't compromise one in the slightest? It sounds like a contact issue where lower frequency can get by but when pushing the edge of the PCB of the X170SM-G it borks out. I know on my X170SM-G, I can run the sticks both bottom or both top but I had to clean a tiny bit of thermal compound on one of the corner pins left there from ZtecPC I'll assume. I could push stock 4.9 -120uv but for CB23 stability I backed off to -92uv. See how your system handles a 4-5 looped CB23 run. All my testing is centered around CB23 for consistency. R15 will pass some things that CB23 borks out on so that's why I use it. It is all I use as my baseline for comparative analysis between systems and posted testing. Since it is @Mr. Fox 10900k I'll assume it is already delidded? 😁 If you can, do a single CB23 run at stock -90uv and report the temps, package power and Vcore under load. Every 10900k I've tested that managed to make it to 5.3 all core CB23 eventually fell off the rails at 5.4 and needed decent cooling. There isn't enough magick in the world that will suddenly super empower the X170 Heatsink to handle the serious heat and pull 5.3 and 5.4 all core put out under meaningful loads. I keep looking at that aliexpress heatsink to give it a whirl since they are being sold dirt cheap now. What memory sticks are you running and Vmem and freq/timings?
  22. Good to hear you just opened a case and wash your hands of that mess. The more I think of it, unless the box/package was absolutely smashed to oblivion, there is no way the IHS would be removed AND smd's sheered off. I sense shenanigans on the seller's part. As for cache, yep and it can get real finicky >43x in many situations, shares power and can increase thermals. The fact it defaulted to 47x in a constrained environment is perplexing. Oh Yes!
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. Terms of Use