ASUS Explains Transient Response, VRM Layout Basics, & LLC for Motherboards
video description
Date: 2020-05-06
Related videos
Comments and reviews: 10
bobhumplick
i know what they are going to say before they say it. doublers hurt transient response and less phases with no doulbers is better. and for transient response thats true. for most people this is the most important when overclocking. llc isnt as effective with doublers because of latency. by the time they sense a voltage drop and respond the latency from the doulbers means its too late. giagbytes 12 phase z390 boards are great vrm designs for high load and ripple which are both important. but the doublers hurt response time. and on boards like this it seems to me that llc just under the flat llc or even 2-3 steps under flat are the only ones that actually make a difference. anybody using the aorus z390 boards try llc at medium before you jump to high or turbo and beyond. you might be surprised at how little is actually needed to get the effect. not because higher llc doesnt help but because it just doesnt help on these particular boards. asus boards or boards without doublers might be different though. basically by the time the droop happens the controller cant compensate quickly enough because of the delay from the doublers. anything over about medium or high doesnt really help like it should for example i have a 9900k that will run at 1. 2v under load going by the vcore reading (the highest one, not the second sensor or vrout, but the highest vcore reading. i set medium llc and then 1. 28 and then by the time it droops it is about 1. 2v. but if i set 1. 2 or even 1. 22 or something and use a flat llc then it crashes. if the board had better transient response i could probalby run 1. 25 or so at the setting just below flat. or maybe even 1. 23 flat if the board had crazy fast response. if youre just gaming or whatever and not running a high load all the time doulbers arent always necessary now everybody's use case is different. someone like buildzoid needs vrm's that can provide a lot of power and dissipate a lot of heat but for the average user on a water cooler or especially an air cooler, no doulbers would be better. but the best solution is lower priced controllers with more than a hand ful of phases. the industry needs to move to a 16 phase controller or something so you can do soc\sa and everything else and still have 12-14 phases for the cpu.
reply
i know what they are going to say before they say it. doublers hurt transient response and less phases with no doulbers is better. and for transient response thats true. for most people this is the most important when overclocking. llc isnt as effective with doublers because of latency. by the time they sense a voltage drop and respond the latency from the doulbers means its too late. giagbytes 12 phase z390 boards are great vrm designs for high load and ripple which are both important. but the doublers hurt response time. and on boards like this it seems to me that llc just under the flat llc or even 2-3 steps under flat are the only ones that actually make a difference. anybody using the aorus z390 boards try llc at medium before you jump to high or turbo and beyond. you might be surprised at how little is actually needed to get the effect. not because higher llc doesnt help but because it just doesnt help on these particular boards. asus boards or boards without doublers might be different though. basically by the time the droop happens the controller cant compensate quickly enough because of the delay from the doublers. anything over about medium or high doesnt really help like it should for example i have a 9900k that will run at 1. 2v under load going by the vcore reading (the highest one, not the second sensor or vrout, but the highest vcore reading. i set medium llc and then 1. 28 and then by the time it droops it is about 1. 2v. but if i set 1. 2 or even 1. 22 or something and use a flat llc then it crashes. if the board had better transient response i could probalby run 1. 25 or so at the setting just below flat. or maybe even 1. 23 flat if the board had crazy fast response. if youre just gaming or whatever and not running a high load all the time doulbers arent always necessary now everybody's use case is different. someone like buildzoid needs vrm's that can provide a lot of power and dissipate a lot of heat but for the average user on a water cooler or especially an air cooler, no doulbers would be better. but the best solution is lower priced controllers with more than a hand ful of phases. the industry needs to move to a 16 phase controller or something so you can do soc\sa and everything else and still have 12-14 phases for the cpu.
reply
RJ
Quick translation and summary: Asus hires some smart dudes to explain things in terms maybe 1% or less of the population can understand, in hopes that they can justify their cheapness and/or laziness when it comes to making VRMs: P Maybe transient response is better with a twin 8-phase (so a 4 phase, but like Buildzoid said, what actual difference is there in real world use? Not one he has noticed so far. Only difference people can see and measure is higher VRM temps than the other boards at its price point, though at least not worryingly high. Maximus XI Hero at the price they're charging is a no go (it's adequate enough, but you don't pay 250-300 USD for adequate enough. At least it seems they've done better with X570, let's hope that trend continues. This is coming from a long time Asus fan and customer, though my Maximus X Hero had to be RMA'd, and now my 2nd board is crapped out as well after a few months. Even my darn Z170 Pro gaming, a cheap ass Asus board, is still going strong (okay maybe not super cheap, but a step down from the actual ROG boards, aka Hero and up. Though now even the Hero has a cheaper VRM. Bought a Z390 Aorus Master and couldn't be happier, though the BIOS takes some getting used to. Weird thing is, my PC runs so much smoother than it did on both of the Hero boards. Don't know if Asus just sent me back a shit board for RMA, but I may just have found a new go-to company for my mobos.
reply
Quick translation and summary: Asus hires some smart dudes to explain things in terms maybe 1% or less of the population can understand, in hopes that they can justify their cheapness and/or laziness when it comes to making VRMs: P Maybe transient response is better with a twin 8-phase (so a 4 phase, but like Buildzoid said, what actual difference is there in real world use? Not one he has noticed so far. Only difference people can see and measure is higher VRM temps than the other boards at its price point, though at least not worryingly high. Maximus XI Hero at the price they're charging is a no go (it's adequate enough, but you don't pay 250-300 USD for adequate enough. At least it seems they've done better with X570, let's hope that trend continues. This is coming from a long time Asus fan and customer, though my Maximus X Hero had to be RMA'd, and now my 2nd board is crapped out as well after a few months. Even my darn Z170 Pro gaming, a cheap ass Asus board, is still going strong (okay maybe not super cheap, but a step down from the actual ROG boards, aka Hero and up. Though now even the Hero has a cheaper VRM. Bought a Z390 Aorus Master and couldn't be happier, though the BIOS takes some getting used to. Weird thing is, my PC runs so much smoother than it did on both of the Hero boards. Don't know if Asus just sent me back a shit board for RMA, but I may just have found a new go-to company for my mobos.
reply
Tiki832
How about ASUS explain why 2 weeks into release and they still haven't released any proper X470 BIOS for their boards that actually fully works with 3000 series CPUs? I know the public reason they've given is that the most recent AMD AGESA release has some bugs they're trying to find and fix and that's why they're still using much older pre-release versions even in their latest BIOS release this week. but honestly when the current public BIOS releases can't handle memory above 3200Mhz when using a 3000 series CPU on a ASUS X470 board, and can't support any decent timings plus additional stability/performance issues because they're using a pre-release AGESA revision then frankly the fact that the PCI-E Lanes on fringe set ups will revert to 2. 0 bandwidth speeds instead of 4. 0 speeds which is one of the key reasons why they're not using it given by a ASUS technician when asked, really doesn't hold much weight. Especially when most other board vendors have since pushed out new BIOS revisions to use that AGESA release. Even those that were lagging behind like ASUS at the start of the week.
reply
How about ASUS explain why 2 weeks into release and they still haven't released any proper X470 BIOS for their boards that actually fully works with 3000 series CPUs? I know the public reason they've given is that the most recent AMD AGESA release has some bugs they're trying to find and fix and that's why they're still using much older pre-release versions even in their latest BIOS release this week. but honestly when the current public BIOS releases can't handle memory above 3200Mhz when using a 3000 series CPU on a ASUS X470 board, and can't support any decent timings plus additional stability/performance issues because they're using a pre-release AGESA revision then frankly the fact that the PCI-E Lanes on fringe set ups will revert to 2. 0 bandwidth speeds instead of 4. 0 speeds which is one of the key reasons why they're not using it given by a ASUS technician when asked, really doesn't hold much weight. Especially when most other board vendors have since pushed out new BIOS revisions to use that AGESA release. Even those that were lagging behind like ASUS at the start of the week.
reply
Tim
If they believe this type of VRM design is best that's fine and all, but their marketing department needs to understand that 8 power stages across 4 phases is not 8 phases. When Z390 was released their marketing page for it said that the new Intel 8 core CPUs need more power and that's why all ASUS Z390 motherboards have at least 8 power phases. But in reality the Maximus XI Hero/Code/Formula were all parallel 4 phases. Calling them a twin 8 phase without explanation was misleading too. Back then ASUS never apologised for the bad marketing, even when their staff were asked directly for clarification on the ROG forums they stayed silent, and instead quietly changed the wording in their marketing and pretended it never happened. I still have screenshot evidence of the original marketing saved on my PC. This really didn't impress me.
reply
If they believe this type of VRM design is best that's fine and all, but their marketing department needs to understand that 8 power stages across 4 phases is not 8 phases. When Z390 was released their marketing page for it said that the new Intel 8 core CPUs need more power and that's why all ASUS Z390 motherboards have at least 8 power phases. But in reality the Maximus XI Hero/Code/Formula were all parallel 4 phases. Calling them a twin 8 phase without explanation was misleading too. Back then ASUS never apologised for the bad marketing, even when their staff were asked directly for clarification on the ROG forums they stayed silent, and instead quietly changed the wording in their marketing and pretended it never happened. I still have screenshot evidence of the original marketing saved on my PC. This really didn't impress me.
reply
Jared
This has nothing to do with the video, but I just wanted to share this. I just woke up from a dream where Steve was going through a personal character development program initiated by his friends to cure his rampant alcoholism. Part of this program was to livestream him loudly humming different notes in order for him to realize how annoying his voice was. In my dream he had to keep humming the E note over and over because he was flat. It took so long that Linus and Jay had to step in to help manage his videos for some odd reason. Turns out though that his horribly flat humming was actually the sound of my phone vibrating on my desk as it's alarm went off. That being said, Steve if you could post a video of you humming an E note just to satisfy my now burning curiosity, I'd greatly appreciate it.
reply
This has nothing to do with the video, but I just wanted to share this. I just woke up from a dream where Steve was going through a personal character development program initiated by his friends to cure his rampant alcoholism. Part of this program was to livestream him loudly humming different notes in order for him to realize how annoying his voice was. In my dream he had to keep humming the E note over and over because he was flat. It took so long that Linus and Jay had to step in to help manage his videos for some odd reason. Turns out though that his horribly flat humming was actually the sound of my phone vibrating on my desk as it's alarm went off. That being said, Steve if you could post a video of you humming an E note just to satisfy my now burning curiosity, I'd greatly appreciate it.
reply
Johnny
Two things ASUS does not want to talk about regarding their anemic phase design, idle power loss due to many components having to run since some boards now use triple design you have 3 powerstages having to be on at minimum compared to others with a phase per stage, same with load balancing and thermal balancing, you can only do it in 3s compared to one phase per stage, and the final thing ASUS does not want to discuss, the fact they are testing with a dummy load and not a CPU, try testing transients with a true CPU under various loads, there's a pretty high chance you won't be able to se any difference in transient respons between a doubled design and a twin design at all. ASUS is simply trying to cheap out on controllers and skip doublers at the same time.
reply
Two things ASUS does not want to talk about regarding their anemic phase design, idle power loss due to many components having to run since some boards now use triple design you have 3 powerstages having to be on at minimum compared to others with a phase per stage, same with load balancing and thermal balancing, you can only do it in 3s compared to one phase per stage, and the final thing ASUS does not want to discuss, the fact they are testing with a dummy load and not a CPU, try testing transients with a true CPU under various loads, there's a pretty high chance you won't be able to se any difference in transient respons between a doubled design and a twin design at all. ASUS is simply trying to cheap out on controllers and skip doublers at the same time.
reply
fozz1138
Interesting video- would be cool to see the whole thing. Undershoot is where the signal line is over-filtered, the filtering makes the response to change too sluggish so the square wave change isn't vertical, it takes longer to change so you get a sloping edge rather than a vertical one. Undershoot is the opposite, where the signal is underdamped and can change too quickly, which is where that 'ringing' occurs. (or something like that; trying to remember back when I worked on pulse electronics. You have to balance the filtering to allow a fast enough response to the input signal but not so fast that you produce damaging levels of overshoot.
reply
Interesting video- would be cool to see the whole thing. Undershoot is where the signal line is over-filtered, the filtering makes the response to change too sluggish so the square wave change isn't vertical, it takes longer to change so you get a sloping edge rather than a vertical one. Undershoot is the opposite, where the signal is underdamped and can change too quickly, which is where that 'ringing' occurs. (or something like that; trying to remember back when I worked on pulse electronics. You have to balance the filtering to allow a fast enough response to the input signal but not so fast that you produce damaging levels of overshoot.
reply
Locks
This is very cool! Thank you gentlemen! ASUS, you just impressed me, which is not easy. Many of us, myself included, love to really know about the products we are using, and the fact that you allowed your project manager, Jonathan, to share knowledge with us, without going through a marketing wank that really has no idea what he/she are talking about, speaks volumes. You got our six, so I got yours. I'm currently, happily, running an ASUS board with an Intel chip and WAS going to grab an MSI board for my current Ryzen build, but I will revisit that idea, and see if I can find an ASUS one that will fit my needs instead. Again, thank you.
reply
This is very cool! Thank you gentlemen! ASUS, you just impressed me, which is not easy. Many of us, myself included, love to really know about the products we are using, and the fact that you allowed your project manager, Jonathan, to share knowledge with us, without going through a marketing wank that really has no idea what he/she are talking about, speaks volumes. You got our six, so I got yours. I'm currently, happily, running an ASUS board with an Intel chip and WAS going to grab an MSI board for my current Ryzen build, but I will revisit that idea, and see if I can find an ASUS one that will fit my needs instead. Again, thank you.
reply
rusTORK
O. K, i understand black and green marker colors. I guess it's a good start! I think that Buildzoid did huge impact on understanding how motherboards are work. Without that we probably were focused on PCI Express, DIMM and SATA slots amount. But now we got ASUS ROG PM (Power Manager) which trying to explain us transient response of their motherboard VRM. Do you get it! Now try to go in near PC store and ask their stuff in the middle of conversation about transient respond of that two motherboard? Which one is better? He will be like What the F you are talking about! .
reply
O. K, i understand black and green marker colors. I guess it's a good start! I think that Buildzoid did huge impact on understanding how motherboards are work. Without that we probably were focused on PCI Express, DIMM and SATA slots amount. But now we got ASUS ROG PM (Power Manager) which trying to explain us transient response of their motherboard VRM. Do you get it! Now try to go in near PC store and ask their stuff in the middle of conversation about transient respond of that two motherboard? Which one is better? He will be like What the F you are talking about! .
reply
rad
I have a b350 f gaming with ryzen 3600 i was playing with the pbo the other day and i noticed with pbo enabled my pc cannot resume from sleep mode. fully tested it and it is the only variable that causes this. PBO enabled my pc cannot resume from sleep (hangs on a red led after fans and cpu cooler are all on. really annoying because i had some really nice profiles set up that passed occt with no errors. i tried everything as far as turning off fast boot changing power profiles. the only thing that allowed me to resume from sleep is with pbo on auto aka off.
reply
I have a b350 f gaming with ryzen 3600 i was playing with the pbo the other day and i noticed with pbo enabled my pc cannot resume from sleep mode. fully tested it and it is the only variable that causes this. PBO enabled my pc cannot resume from sleep (hangs on a red led after fans and cpu cooler are all on. really annoying because i had some really nice profiles set up that passed occt with no errors. i tried everything as far as turning off fast boot changing power profiles. the only thing that allowed me to resume from sleep is with pbo on auto aka off.
reply
Add a review, comment
Other channel videos
