Ubeogesh Posted January 1, 2014 Share Posted January 1, 2014 Subj. http://i.imgur.com/4ph5jhL.jpg :eek: This bug is a total disaster, it completely ruins my trust to this feature. Can anyone help me with this? I am using the 2.5.5145 version. Link to comment Share on other sites More sharing options...
MojoVersion8 Posted January 3, 2014 Share Posted January 3, 2014 80+ Unobtainium :laughing: Link to comment Share on other sites More sharing options...
zyo Posted February 23, 2014 Share Posted February 23, 2014 You are not alonehttp://i60.tinypic.com/11b3gax.jpg Link to comment Share on other sites More sharing options...
ignight Posted March 1, 2014 Share Posted March 1, 2014 Same here. 1.) "PowerOut" is most of the time higher than "PowerIn"...yay we got power-plants. 2.) On Idle both PowerIn and PowerOut are way to high. Solution: Visit your local DIY-warehouse and buy a cheap Watt-o-meter (approximately +/-5% accuracy) and plug it in the PSUs power cord. Haha, funny that cheap mainsteam DIY hardware made-in-china for power-plugs is way more accurate than a PSU internal measurement of Corsair. OH dear. Link to comment Share on other sites More sharing options...
ignight Posted March 9, 2014 Share Posted March 9, 2014 http://imageshack.com/a/img593/1871/zgbc.jpg Link to comment Share on other sites More sharing options...
ignight Posted March 9, 2014 Share Posted March 9, 2014 So basically... haha, they already tried to make up "power out" just according to the spec-curve without even measuring it... well, if there's no alternative because of broken ICs i must say im fine with it!!! The problem is that they calculate Power In wrong, what is really no rocket sience... it's "Main V * Main A" result is very accurate Power In Watt according to what my watt-o-meter tells me. Plus they make it up wrong way around for the efficiency-statistic. Whenever my CPU goes full-load... ATX 24PIN 12V still shows 0,0 Ampere.... but Power In of course raises. So funny way around, the Power Out now shows the correct curve-algorythm according to the spec but wrong way around.... the curve-algorythm must not raise the "variable" Power-Out...but needs to LOWER IT! So when consuming very low power on Idle it shows 113% efficiency, while it means 87%...lol... soon as power-draw increases to high load only for CPU (no PCI-express power-draw) the efficiency showing 101-105%, what should be 95-99%. ...really? NEIN? DOCH! OH! LOL - Now thats what i call epic awkward failing with a non-functional hardware feature beeing replaced by even worse functioning software, since it is not actual reading out power out, but calculating it from Power-In applied specification-efficiency-curve! :p DOH! Now corsair please do the math correct, and at least FIX this finally to show correctly! All you need to do is to calculate the variable shown "Power IN" with what is shown in "Main V" * "Main A", and apply efficiency-curve for calculating power-out based on correct power-in correct way-around! I request for a new BETA to apply these ultra-simple fixes quickly. THANKS! Link to comment Share on other sites More sharing options...
ignight Posted March 10, 2014 Share Posted March 10, 2014 Bump. Here is Idle / Load (only CPU) comparison and analyze. This time little down-scaled to not screw your browser-window. :D http://imageshack.com/a/img534/6067/0gpq.jpg http://imageshack.com/a/img31/5426/3f4h.jpg Having the brains to troubleshoot the problem and work out a solution makes the difference between "Professionals" and "Amateurs". Corsair is pretty "Amateur" to me in this demand...unable to deliver a fix for this crap since the beginning now ... any more monthes passing by, i start distancing from "Amateur"-Corsair-Products ! Link to comment Share on other sites More sharing options...
Corsair Employee jonnyguru Posted March 12, 2014 Corsair Employee Share Posted March 12, 2014 Hi guys! I just wanted to let you know that fixing this IS a priority. Unfortunately, the problem isn't "shady math". It's the accuracy of the sensors used to measure the AC input at low loads. If the AC current is < 1A, accuracy can be off by as much as 0.25A. Obviously THAT'S HUGE and it totally screws with the efficiency equation. We tried implementing a look up table for displaying efficiency at loads < 100W, but what we ran into is that the transition from the look up table to actual measurements, the efficiency numbers spike even to even higher numbers than what we're seeing now. This has been the case for the last two beta releases. I've put a trouble ticket in for the current issue and hopefully we can get it tweaked to the point where we're all happy. Link to comment Share on other sites More sharing options...
ignight Posted March 14, 2014 Share Posted March 14, 2014 Bump: You are right that the AC current on <1A is off 0,25A, but it's not fluctating... but steady reporting up to 0.25 A higher. But still just Main V*A without correction as it is, would be more accurate, than what LINK is actually reporting for Power In. Several AXi show the same behaviour, so we can rule it by thumb, that the smaller the Main-Current drops below 1A, the higher is the sensor reporting it. So what you need to fix it is to apply correction on the Main-A to adjust small values lower than 1A with a decreasing factor <1 i.e. 0.9 or 0.8. This would result in Main-A beeing more accurate and in Power-In beeing nearly perfectly accurate when calculated through Main-V*A. After that is done, you could continue applying theoretical efficicy-calculation on the Power-In by applying power-efficy factor according to the specification-documents to calculate popper PowerOut. ;) The problem is that the "look up table" you integrate to correct the values follows wrong math.... it raises them instead of lowering them, so it makes it exactly double-times worse except of better. You apply factors >1 for the corrections in your formulas... what is wrong... the reportigns are to high... so you have to LOWER the values by applying correction-factor <1. Think about it. Won't you agree? That's exactly what you refuse it to be: shady math! Check that calculation and probe them... i bet the results are >1 what is wrong.. Main-A needs downwards correction for <100W!!!! Simple as that! Sloppy math! Maybe you calulate that correction factor itself by formula, then: Somewhere within your formulas that make up the correction-factor you have to switch "devisor" and "divident"... and the magic would work! PS: Also i do not know why you apply these correction for "PowerIn" only, when the root problem is incorrect "Main-A" readings ..... wouldn't it be nice to have the display of "Main-A" also make properly show at first - all other Ampere dont work at all except PCIe - and aftert that take THAT corrected and displayed "Main-A" value to multiply with displayed main voltage, so the numbers obviously make mathematical sense to anyone!? I mean if all that would work, why need i devide PowerIn through Main-VOltage to find out about correct Main-A, which still would show up wrong, because correction of that is hidden under PowerIn value-calculation. ...Doh! Can you do some things right from time to time... some few values with an accuracy +/-5% would already be sufficent enough to not loose your face. Im slowly entering rage-mode, trying to explain your mistakes in detail by reverse-engineering.. you still not seeing it. Hope you finally got it. Read and understand the bold-lined! Link to comment Share on other sites More sharing options...
wytnyt Posted March 14, 2014 Share Posted March 14, 2014 jeez dude lighten up on the sarcasm and demeaning attitude a bit,im sure if it were as simple as you suggest then they would have implemented this. lots of times things look good on paper but real world testing and usage is a whole different ballgame as is most of the time just sayin from a neutral perspective is all. Link to comment Share on other sites More sharing options...
CryoGenic Posted March 15, 2014 Share Posted March 15, 2014 Jonny Guru I am using LINK version 2.4.5110 with the registry tweaks H100i firmware 1.0.5 and efficiency seems to be working correctly on the power page. AX860i at idle 88w in 87w out 98%. Main amperes in .781 and 0's in both 24 pin amps and peripheral amps, they are not working. Since everything works for the most part I did not want to risk the Beta. Thanks to wytnyt (I was skeptical) I changed from average CPU temp to water temp and the custom fan curve has been working really well now. I built this system in Oct 2013, two weeks ago I made the change to water temp which has fixed the only real problem I had the fans not responding to the custom curve, now they do. I do not load LINK at start up because when I tried stress tests monitoring the core temps LINK worked as set without loading at start up. I always had contempt for the guy who posted "I am not having any problems" instead of helping the OP out but I am not having any real problems, sorry for all the people that are, hope Foxconn gets the software written soon. Link to comment Share on other sites More sharing options...
Corsair Employee jonnyguru Posted March 15, 2014 Corsair Employee Share Posted March 15, 2014 The problem is that the "look up table" you integrate to correct the values follows wrong math.... it raises them instead of lowering them, so it makes it exactly double-times worse except of better. You apply factors >1 for the corrections in your formulas... what is wrong... the reportigns are to high... so you have to LOWER the values by applying correction-factor <1. Think about it. Won't you agree? That's exactly what you refuse it to be: shady math! It's easy to be an armchair programmer and call others "amateurs". I'm not a programmer @ Corsair, and I doubt you're a programmer at all. But we are working on the problem and that is the reason for my post. FYI: The look up table for < 100W loads is only in the second to last beta release. When it was found to not work, it was pulled out. The problem is simple, but the solution not so much. But loose cannon ranting doesn't contribute to solving the problem. Like I said, they're working on a fix. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.