MyPC8MyBrain

after update it is still possible to undrvolt but temps with existing undervolt went up 15-20c, i would stay away from updating to bios 1.9.1, went back to 1.8.1 with no apparent issues atm,

did you try disconnecting the systems main battery running on DC power only? (when you remove cmos battery you should also have your main battery disconnected otherwise it defeats the purpose!) FYI: when you remove RAM and power the system it should beep at you constantly, if it didn't something broke before the ram gets powered,

1) Yes 2) don't remember 3) Yes 4) Yes 5) 12950HX 6) 3080Ti 7) 1.8.0

indeed, i inferred from @1610ftw post, i assumed there was more data on the original thread where he saw indicating temps, id say way, that is at the least plausible 😅 i did exactly that with my 12950HX, 26k on my none premium bin is not what most get with the same spec system its more like 22k, a good bin with more efficiency cores things should add up, plus CB23 can tolerate extreme undervolting and overclocking other application will not tolerate, i saw 26k best CB23 score in person for the 12950HX, even if we slap 1% per core increase performance its already close to 10% at least just by merits of number of cores added in the 13980HX,

with LN id expect temps to be in the negative side of temps not ranging in the plus 92c, not too far fetched i think, if he buffered his hardware with fans blasting full throttle for 10-15 min prior benching at around 0c ambient temps (aka outdoor) these temps do make some sense, i think that's a very low median range with additional 8 cores, 12950HX does 26k, you inferring 5% increase from previous gen with 8 less cores?

impressive indeed, i have seen the difference a good silicon makes first hand, id say this is a combination of good bin plus a solid and stable undervolt due to it being a good bin, combined with the ability to buffer the system with fan control and maybe the cold outdoors whether? 😉 i haven't really stressed tested my current rock solid cpu bin, it is still using factory thermal paste, with previous unsettled bins i was able to push up to 26k in CB23, i think my current bin can make even +27k properly prepped, seem to be a solid 30% (rounding) increase from just the 12950HX, at 40k - 45k that's current gen desktop cpu score territory,

thats not too much if anything your application there is bit on the "dry" side for my personal taste, I prefer the "pool" effect on both sides or at least the silicon side, when I say pool I don't mean overdo it just to make sure that there's enough LM so the entire surface acts like a single surface mirror with no craters so it looks almost like a single mirror surface, when you touch the center there should be enough LM so the covered surface remains unified, its a fine line between getting a unified LM surface and overdoing, if you are protecting the area around the cpu you should have no concerns even when overdoing, the thing about LM is not how much you apply its that it runs when hot and can get around and short other components,

i have, other than the full 330w actually drawn from the wall generating more heat, same result as with the 240w power supply no difference whatsoever,

why does anyone needs an online store installed on his computer when all it does is access webpages on its back end, what's wrong with a user accessing the web store page located anywhere on the web where these pages live anyway? it rationalizes MS forcing firewalls, defender, encryptions and random updates all to protect their store installed on your computer by design from factory, they don't care about anyone's identity they only care that you don't steal from their store they installed on your computer without paying them, its as simple as that, engagement for the soul propose of tracking users habits and predict what they can sell you through their pipeline, pure corporate greed.

we kinda don't have a choice in the matter its dictated by the big players, what choice is left, even after tuning the crap out of latest windows everything is now "app" based which you have to get from the... you guessed it the app store which is conveniently installed and runs on your PC, remove it and a slew of issue rises up, and its not a coincident its deliberate, a corporate baboon decided this many years ago today its jus his deals and collaborations coming to fruition, since windows 7 all these nonsense with security alerts 50 times a day scaring the crap out of normal users is another plot that absolutely makes zero sense but hey MS can exploit people's fear, defender and firewall that do nothing but increase customer engagement that without these alerts wouldn't know or care, and it makes a great reason to encourage people to upgrade, its a win win squeeze monopoly for everyone involved in the commercial end of selling PC's accept the end user.

a mainframe with terminal access is not a new architecture or concept it has been the basis of the industry since the beginning, what was revolutionary is the concept of personal computing as most by now lost the meaning and use a general acronym PC without fully understanding that PC's have long been gone shortly after windows XP. what we have now are Public Computers which also qualifies under the same acronym PC 😄 but technically its a managed computer by microsoft, you buy the hardware and they manage every aspect of the operating system, to see the industry going back to a mainframe architecture is somewhat funny, maybe in few years from now we will finally get back to a real PC as in Personal Computer the way it used to and should be.

i still have somewhere a dual Xeon 700Mhz in a SuperMicro board i custom built over 20 years ago 😉

everyone got spoiled now days, how about any basic 32-64MB integrated will do the trick for email, browsing the web, and watching some videos, as it did for the past decade with zero issues (mostly unless one insists on watching 4k streams), not to mention newer generation of far more capable integrated video chips exist now days, i was doing just fine before with 32mb UHD chip for casual browsing, there's a slew of application beside gaming that benefit from a powerful GPU rendering like physics engines, modeling, AI, and various simulations, all run heavy mathematics calculations that requires sometime ECC and every bit of resources to compute in a reasonable time, these gamer focused marketing campaigns will fade out in the next few years and the lines will become more and more blurred between different category needs nowadays, it used to be only gamers cared or needed these, as we learned from the past few years shortages there is far more business needs for these well beyond gamer community scope,

i like Merrell brand a bit more distinguished then having the same next door neighbors crocs, true story back in the days (04) when Crocs were just entering the US market i was almost hired by them to implement and manage their POS across the states, in a last minute turn of event i turned them down after an offer was made to go work for (no one know at the time myself included) the first social media network that shaped the face of social networking as we know it today,

that's only valid until 3080 stocks run out and shelves are only stocked with 40xx lines, that's why the rush,

hmm... i beleive you can shave around 30% more which in turns should effect your bench results few clicks north, while benching i was running latest Win11 with 47-48 process at idle (and that's on a laptop), this is from right now my current idle which I still consider extremally liberal for none benching, i left few windows gadgets running in the background like weather, spotlight, search, even windows updates is running within that scope, and few standalone executable running continues task (ThrottleStop.exe, TopRight.exe moves incoming alerts back to the top right corner etc., this is not a custom or stripped windows11 image it is off the shelf latest available win11 22H2 build downloaded from MS)

i guesstimated 70-75, thats badass, i assume around 50 at idle?

@Mr. Fox curious to learn how many background process are running for you while you bench?

any of you tried unhiding "away mode" in power policy then setting it to "no" in active policy? https://www.tenforums.com/tutorials/108441-add-remove-allow-away-mode-policy-power-options-windows.html

if that's the case its a bogus ECC! they cant be doing software based memory error correction that's laughable calling it ECC, its like having a 4 year old babysit a 3 year old, maybe its an acronym for something else NVidia made up they just call it ECC to be boujie and confusing? I'm sorry I'm still having a hard time wrapping my head around the rational behind this nonsense from NVidia to be honest,

of curse it will ECC is additional processing layer which defeats the whole premises of running a bench, for other application it is critical when its reliable stability they are after for implementing ECC on a server not for benching bleeding edge performance, lets call a spade a spade the developers made a booboo and the only way to ensure result is to enable ECC so they can get a readout they can reliably analyze, in simple word their bench software cant keep up, is that a fair assumption?

i was under the impression that they want to keep it off or something? so was last gen 3090, how is the 4090 different in its workflow for creators that it needs ECC? also what kind of ECC is that, traditionally there's no ON/OF switch for ECC unless its some software implementation or there's actually two separate memory controllers onboard, if anything i can maybe see this useful for miners but that's pretty much it. this has to be a software based ECC mutant which in my view still makes zero sense and sort of defeats the purpose of a performance bench,

i'm trying to follow this ECC debacle, some basics i am having hard time wrapping my head around, the first is why ECC is on a gpu in the first place? if for whatever justified reason ECC is needed for some gpu scenario why are you guys insisting on keeping it on if the feature is disqualified by bench tools moving forward? in other words why slow down calculation in a bench is that important? what am i missing?

it's similar case with the 4080, smaller die, less VRAM, less everything compared to the 3080,

12950HX @157w (it went even higher i just didn't capture)