Jump to content

VeganJoy

Member
  • Posts

    1,347
  • Joined

  • Last visited

Awards

This user doesn't have any awards

3 Followers

Profile Information

  • Gender
    Male
  • Location
    bumfuck nowhere, arkansas

System

  • CPU
    Ryzen 5 3600 (placeholder until zen 3)
  • Motherboard
    Asus ROG Strix X570-E Gaming
  • RAM
    2x16gb GSkill Trident Z RGB 3600C16, 14-15-30@1.45v
  • GPU
    EVGA 1080ti FTW3, max power/voltage stock bios: +50 core +600 mem 
  • Case
    Fractal Design Meshify C Dark
  • Storage
    OS Drive: Samsung PM981 1tb (OEM 970 Evo)
    Storage Drive: XPG SX8200 Pro 2tb
    Backup Storage: Seagate Barracuda Compute 4TB
  • PSU
    Seasonic Prime Ultra Titanium 750W with black/white Cablemod extensions
  • Display(s)
    Asus VP249QGR 1080p 144hz IPS
  • Cooling
    Noctua NH-D15S w/ white chromax bling
  • Keyboard
    Corsair K95 Platinum
  • Mouse
    Logitech G502 Hero SE, hard mouse pad not one of those crap soft ones
  • Sound
    Sennheiser Game One, Sennheiser 6XX, Sennheiser GSX1000
  • Operating System
    Windows 10 v1909

Recent Profile Visitors

800 profile views
  1. i am now realizing that i'm a bit dumb and forgot to change the load line setting. so it's been on auto for all of today lmfao so thats interesting, i thought LLC on auto was supposed to try matching load voltage to whats set in bios... time to try what i meant to try in the first place
  2. yeah once i have the cooling solution in place i'll be running 0 avx offset... this d15s was really not designed for such a monster chip lmao. but i don't like wasting power at idle so i usually leave the big cpu power saving options on. unless some of them affect stability like the ram power down stuff? or was that amd only?
  3. i have no idea, i'm running memtest86 on a bootable stick which i'm pretty sure isn't really intensive, so nothing should be getting too hot. even the sticks are basically cool to the touch, maybe 30C lol. i was running all my memory overclocking at stock cpu settings anyways. so i'm running the silicon lottery settings and it's working pretty good, load voltage is 1.2v for 5.1ghz. although they have you put in power limits so that avx doesnt catch everything on fire lol, so it's throttling to ~4.5ghz avx as a result. and idk if their "per core usage" way of sorting the clockspeeds works right cause setting p95 to run on just a few threads doesnt run at the proper positive clock offset. once i get all my custom loop parts in stock, i'll make sure to go even more crazy lol. actually now that i'm paying more attention to it, it seems that vdroop varies depending on how many cores are being used? at LLC 3 and 1.375v set in bios, 20 threads of p95 non avx results in 1.2v load and 1 thread results in 1.35v load (running between 5.1 and 5.2ghz, from what i can tell in hwinfo). i wonder if this is intended by the low LLC setting?
  4. honestly i've been having problems with just a basic memory overclock lol. i had 2x8gb of those patriot 4400c19 b die sticks that werent stable at xmp (was able to get 4000c15 stable with shit subtimings lol) with this 10900k+hero and now i cant get 2x16gb of trident 3200c14 running at 4000mhz without errors. i've gone through the memtesthelper guide on ddr4 overclocking and all of the "safe" settings throw errors, violating the guidelines/rules makes it more stable (3600 14-14-30 throws errors, 3600 14-15-30 works, 3600 14-15-34 throws errors??). 4000mhz 18-19-19-39 at 1.5v throws errors but i know the parts will run 4000mhz ram. and presumably, comet lake is supposed to have really good support for dual rank overclocking compared to previous gens. been running 1.3v VCCIO/SA too, higher is more unstable and lower is more unstable lmao. i set those to auto with 4000mhz to see what silly numbers it'd spit out and i stopped being able to boot 4000mhz with any settings, let alone memtesting it..... just so confused lmao oh btw, i forgot that i always use the silicon lottery's historical binning stats to help my overclocks lol. in their 10900k overclock instructions they say to run LLC 3, making the load voltage (die sense, which seems to be ~0.1v below socket sense) around 1.19v... would it be better to have the LLC higher and set the bios voltage lower, such that the load voltage is about the same but the level of vdroop is lower? they do run a -1 avx offset and i've read that the voltage jump between avx offsets and normal operation can cause instability, maybe that has something to do with it?
  5. your gpu will automatically utilize system ram when it runs out of vram, much like how your system will utilize your hard drive through a page file when it runs out of vram. it is similarly bad for performance as the connection between your dram and gpu is far slower and higher latency than vram, and you'll see lots of stuttering. so basically, don't run out of vram
  6. i cant find a "power down mode" in the intel bios like i could on amd but it seems that setting the PPD timing to 0 will do the trick?
  7. whoa thats a bit chilly lol. i'd like to hit 4000mhz, i'll try without watercooling em but if i'm running in the 50s and 60s for the hottest dimms then it's watercooled ram time as for the disabling power down mode, is that setting PPT to 0? i tested with MCE on, stock clocks, vcore at 1.25 and with varying levels of LLC, using non avx p95 for the load. LLC 8 brings vcore up to 1.28, chip reports 1.18v. LLC 4 drops down to 1.10v, chip still reports 1.18v but it runs a lot lower wattage and thus temps (220w to 170w iirc). LLC 1 bluescreens immediately upon starting non avx p95 lol. so with the amount of vdroop at LLC 4 (1.25 to 1.1), what scenarios would the lower LLC levels be useful for? and what's up with what the chip reports in hwinfo versus what the motherboard reports in hwinfo?
  8. huh fair enough, so i should leave it on for my manual oc? how about all the different LLC levels? i'm sure they wouldn't have so many levels for no reason. i know it's for controlling vdroop but would you ever want a situation where you have lots of vdroop? welll i was thinking i should run 1.55 or 1.6v like you i've got some sticks running 1.35v XMP right now, and in p95 large ffts they hit 40-41C already; didnt you say that <40C is best? iirc on my old ryzen system i was running over 50C with 1.45v. and with 4-8 dimms i can imagine they'll run even hotter than just 2 sticks with an empty slot between em...
  9. yeah i've read that temps are pretty important. speaking of, someone near me has some of those ekwb monarch ram watercooling things for really cheap, presumably cause theyre a pretty silly bit of hardware. but if i plan to be pumping 1.5v or more into 8 sticks of ram, you figure itd be worth the trouble? also, i've been overclocking my 10900k + xii hero (soon to be replaced hehe), havent had it for very long and tbh i'm not great with overclocking in the first place. previous owner of both the cpu and board claimed that they ran it at 5.1ghz, 1.25v (in bios), and LLC 8. i dont really know much about the LLC levels but i punched in his settings, and it's at least stable in non avx p95 (my d15s is already really struggling here lol). but if i turn off asus's multicore enhancement and change nothing, it bluescreens on non avx p95 almost immediately. in the short time i'm able to see, load voltage is about the same as with MCE off and the power draw is the same (1.27-1.28v, 240-250w, 95C). the load voltage when i ran the chip with just MCE and no overclock seemed like it was pretty close; in hwinfo, the chip reported 1.23v at "stock" and 190w (around 80-85C iirc), the board reported 1.12v and 165w. the disparity is lesser with the overclock but it's certainly confusing and i'm not sure why MCE makes my manual OC more stable; i also have zero clue what LLC setting to use... sorry for the rambling wall of text, any chance you've got some pointers? the asus bios rates the chip at SP94 which i hear is both good and a bit meaningless, but it'd be nice to get a solid bin on this chip before i part with it.
  10. 4000c14 for daily would be pretty insane lol. but i'd hazard a guess that 1.51v on current cpu generations isnt the safest load voltage so how about really bad b die bins, like corsair who mixes v4.31 b die into their 3200c16 kits? would you figure that high voltages on those are safe too?
  11. wow i guess youve been doing this for a while lol. and yeah i pay for high end memory to work, not to burn up... but i'll have to mess with higher voltages and take your word for it, gskill dimms with their temp sensors are real nice. but yeah i'm all about more "functional" overclocks, big numbers are fun i guess but if you can't use em all the time then why bother
  12. that seems like impressively low io/sa voltage for 4000mhz, on an 8700k no less! but why no oc on the cpu? and 1.58v seems like a lot of voltage for 4000c16, dont they make 4400c16 kits that run 1.5v?
  13. ha yeah i went into this knowing that it's not the absolute best performance possible but it should be a ton of fun and it's been my favorite mobo since it released. i just didnt wanna pay 750 plus whatever exorbitant price the cpu was lol. i dont think i'd be interested in competitive overclocking though, i really prefer ambient cooling and i really only try for 24/7 stable ocs. plus i'd burn up a bunch of hardware while learning and i dont have the budget for that yet lol, i only have the budget for this through lurking sale threads and trading/flipping my way up. it's taken me a few months but i've gone from a 9700k + 1080ti to a 10900k + 2080ti while actually making money lmao hmmm if i cant hit 4000mhz on at least 4 dimms then maybe i'll spring for a 10900x. did those get the same die-sanding and other thermal optimizations that comet lake did?
  14. https://www.gamersnexus.net/hwreviews/3392-intel-i9-9980xe-review-disappointing-overclocker-but-good-when-stock seems like it shouldnt be too bad with a decent oc! what are the best options for keeping b die cool? i'm guessing that watercooling them isnt the way to go? ekwb 420mm CE and 280mm SE with noctua ippc a14 2000rpm fans, not going full blast ofc. been planning this first loop of mine for a while so i'm hoping it's decent lol this is a big part of what should be pretty exciting! but the board is technically validated at 4000mhz+ for both 4 and 8 dimm configs, so i was figuring i could get high clockspeeds plus memory interleaving, but idk what to expect from this platform in any way. and idk how low i'd be able to go on timings with that many dimms. would a cascade lake x cpu like a 10900x potentially be binned better, both for memory overclocking and voltage needed for a given clockspeed?
×