23:02 Asus is using a led indicator on top of the power connector. It will alarm if the voltage of the rail drops to low (any transients not just lower power/wattage). Or if the PSU wasn't able to deliver power fast enough as a result of fast load changes. Which can cause a drop in needed voltage. This isn't just about not having enough wattage alone. Which can contribute to blackscreens, crashes, etc. And it's not limited to just the 3080 GPU either although it's demonstrated with it. It would be nice to see this on all GPUs though. As he indicates if you've been running your PSU hard (IE: running the PSU at it's peak performance rating for example) the voltage regulation may no longer be adequate for the fast load changes that are required (for most GPUs...again this isn't limited to just the 3080). Again, I would never recommend someone buy a psu with just enough peak power to run your PC.
Luckily for those that just want to game with a single GPU, it is unlikely you will constantly hit peak wattage of even a 550w power supply Knowing that most of the headroom for overclocking has been taken advantage already from the factory, the total power of the 3080 non-founders edition can be seen at around 360w (3x8-pin MSI Gaming X trio). You factor in a 12-core CPU in extreme cases and you are around ~500w. Really interested to see what AMD can do with RDNA2 efficiency.
Yeah lowered overall consumption from efficiency (Unless AMD goes all out again on voltage on the GPU core I guess?) plus the new improved node here with this 7nm tech and maybe overall improvements to the high peak transients and also the frequent drops to very low voltages that could otherwise trip up especially the older power supply models. Newer motherboards seem to have a setting where for a bit more power consumption you can change it from automatic / low power to typical I forgot the exact figures here but it's not going to be close to zero draw and it won't trip the PSU as a result. Might be more for the CPU though but long as the power supply is not totally inactive I don't think that will happen although it might also depend on the rails and more multi rail design for what's plugged where perhaps also more so for true multi rail models not just a split up single rail with some switch or "key" for changing it to a big 12v rail. There's a bit more hardware on RDNA2 but it should still be more efficient, guess user reviews and tests from sites like Igors Lab and the like will show how it works plus if there's a similar thing here with a possible skewed undervolt amount for what's a minimal performance loss again. ~Mostly, binning and such being quite varied means that not all cards work the same but that's true of hardware in general here though poor advice can be a problem. Having to test on your own for the highest reported safe voltage for the Zen2 CPU's and it's not 1.375v static as still pops up here and there it's actually closer to 1.2v - 1.25v I believe but with a bit of variance. Or for Vega and Navi the recommendation for 1.0v when some cards will need to have the GPU clock speeds reduced. And a bit of a rant here on undervolt and results and all that... Spoiler I can hit 1.150 roughly at 2050Mhz and then I have to start lowering or OCCT starts hitting errors, lots of errors. ~2 Ghz effective with some ups and downs, power consumption near 250w which can be lowered to under 200 with some tuning and ~3% - 5% performance loss at most ha ha. 1950 so 1900 Mhz effective at 1.020v and a 5% power draw increase so it stops bouncing around 1850 - 1870 and a resulting 200w draw reported instead though that's a estimated without measuring it with hardware. (Plenty of advice for setting full 50% power draw on the slider too which is rarely required though it might need 20% or more for achieving a near or above 2 Ghz clock speed consistently if going for that.) It's not perfect but for a free and fairly simple to use software it's quite good at this for a initial overclock or undervolt or just stability test and then a bit extra power and some game or program testing over time and hopefully that's then stable. (Re-test on driver update for when AMD feels like changing things or make the GPU overall work harder or get better which changes the conditions again and needs testing to confirm continued stability on the prior settings also.) EDIT: Two more days for the 3090 reviews and a bit under half a month for the Zen3 info then another few weeks and it's Navi20 info time and whenever reviews and actual availability is a thing. (And here's this handy Python script thing...yay yeah there's that as a thing too now.)
I have a couple of Corsair AX1200's. Both have never had any issues, but are both over 10 years old. Not sure weather to change one when I upgrade to Big Navi or a 3080.
The focus is to use it with an Ampere/Navi GPU, it remains to be seen what will work ok. I wouldnt jump the gun though, wait and see before changing PSU.
https://videocardz.com/newz/geforce-rtx-3080-sees-increasing-reports-of-crashes-in-games I post this because I've been reading some of those posts and from some of the descriptions their PSU is lacking. It's not to suggest all problems are PSU related. However, this is a case study because I want everyone to be ready on what you may potentially need for RDNA 2. To avoid that massive threads of ctd, etc. @OnnA All he did was basically test to see if the PC would start and play a game. And sticking his finger in the fan to sense if it was "hotter" or not. There was no analytical testing to find out loads before/after, transients and exact temps before/after and what's recommended by the PSU manufacture. Did he test this for a week to see if results would fail? Nope just a one/two times for 30 minutes each and report results. And that's just naming a few. No offense but I would never recommend that video to anyone for a 24/7 daily use. He's only doing 30 minute testing... Homeboy is trying to convince me that the PSU is fine for the 3080. But when you look at the graph it clearly tells you it's not fine!! He thinks it's fine because the PSU didn't shut down or CTD when there is more involved then that. Which is the last resort of the psu failing. The last resort should never be used to determine compatibility.
^^ maby it's bouncing back from 2GHz mark cuz of it? I bet UV at 950mV and Set max to 1950MHz will do fine there (~275tW i think)
From what I know whenever you see wild fluctuations like that power delivery is an issue. https://www.igorslab.de/en/sapphire...-less-weight-and-the-best-navi-card-so-far/6/ That's for a 5700xt Nitro. He's saying don't use 260watts but 350watts do to 1ms power draws. What kind of PSU do you think you need for that? IMO, a lot more then 750watts if you want to upgrade in the near future.
