Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

RTX Power Delivery Redesign & FE Cooler | Interview

2018-09-15
hey everyone I am now joined by Qi and Justin we're gonna be talking about the cooling and Power Solutions for the new r-tx 2000 series cards already talked with Tom Peterson so if you want some of the content on overclocking and frequency response be sure to check that video so for today let's let's start with the cooling solution I guess the cooling is the first time that I remember anyway that Nvidia has done an on blower card for reference so let's let's go through some of the basics here I guess before that this video is brought to you by the thermal take level 20 VT micro ATX case the level 20 VT takes the high quality at level 20 design and makes it more affordable and shrinks it down to a micro ATX form factor at that with fully modular paneling it's possible to rearrange this case into whatever configuration you prefer for a micro ATX case that can be a discussion piece in a home theater system click the link in the description below you've got a dual axial solution you want to walk through some of the top level changes other than that yeah for sure you know we're really excited about our new coolant solution for the r-tx 20 series you know it's the first time we've done a Founders edition with a dual axial cooling solution and in fact it's the first time ever that a vapor chambers been paired with a dual axial coolant solution and it provides really an incredible amount of cooling performance you know when you compare to our previous generation the the GTX 1080 founders Edition you know you see temperatures running it like 20 degrees or so lower you'll see acoustics at you know up to one one-fifth quieter one fifth of the noise compared to the previous generation so really some pretty significant improvements and and really the key is that we've combined the dual axial fan with the vapor chamber for the first time ever in a graphics card right so that was my first question about it is the vapor chamber completely or mostly obstructs the PCB so then how do you go about dealing with the fact that air can't really get through the vapor chamber to the prm's what's what's connecting the via rammus to the rest of the cooler yeah so under so we have a base plate that sits on top of the VRMs that's thermally connected to them and in that base plate we have a heat pipe so the heat from the VRMs is thermally connected into that baseplate it spreads out through that heat pipe and can be exhausted out of the system so then for the base plate and the fin stack are those soldered together is that they're all pads what's what's the connection so the vapor chamber is soldered on to the fence deck so it's got great thermal connection there and that vapor chamber does a great job of just spreading out the heat throughout really evenly throughout the whole fence stack because really the key to the performance of the thermal solution is to be able to get the heat which is concentrating the GPU really effectively and efficiently spread throughout all the fins so that when the air blows down on it you're really doing a great job of getting more air over more of the heated area and be able to remove it from the fans and I guess I'll pitch this one out there to for the fans so I noticed and I don't we didn't talk about this before but I notice that the fans are slightly more spaced than some of the partner solutions so in the past partner solutions the duel fans had to be fairly close together sometimes because it's larger sometimes they just do it that way is there any specific reasoning for the space you know I mean did you just optimize for where the vrm and the GPU are basically just optimize for the solution for the GPU and for the entire layout you know one of the other relatively new things that we did on the fan is that these are not these are 13 blade fans we added some more blades to that and again that gives you more airflow at a lower rpm for for better airflow it and with less noise right so Cheetham let's go on to the power stuff so for first immediate question what are we looking at for phase design I mean we've got core in memory what can you walk me through for the TI models and the non TI okay so we have the most number of phases ever on a founders edition design and we did that to be able to really be able to lean into overclocking and give everyone a wonderful overclocking experience and the best efficiency ever on Founders Edition card and so with the TI we have 13 phases of core 3 phases of memory for the non TI 80 have 8 phases of core 2 phases of memory and for the 70 class card we have six phases of core and two phases of memory and all of this put together with the dynamic phase shed technology allows us to be efficient throughout the power draw range so even when you're just having it plugged in and you're doing something like gaming on the web or something that's less taxing it's gonna be super efficient in your system so let's let's break down that word dynamic phase technology that you're working on now I start on the presentation the viewers haven't seen it yet what can you walk us through that yeah so when the power supply's of these switching power supplies how they work is they turn on and off really quick to shunt the current to the GPU and when you have a lot of switching at very low current draw then all of that energy is being burned during the switching process so the more phases you have the less efficient you are at low power at high power you're actually going to spend most of the time in the conduction mode and so you're you want to be have more phases to can have less conduction loss and so what we're doing is as the power goes down when you use when you're running a lower power workload you turn off phases that you don't need and you reduce the amount of switching loss right and one of the other things we saw in the presentation is there's is it 60 watts of overclocking overhead or something like there's additional power overhead from Pascal yeah there is definitely more power overhead from Pascal I'm sorry power overclocking overhead from Pascal not only is there it is that coming from the phase additional phases which provides more current it also comes from the fact that we've provided bigger input connectors than we really needed to and we have this advanced current monitoring technology and you'll be able to see the current monitors on the board you've done a great board analysis before I love those things they're random so so yeah the current monitoring allows us to really dial in the current to make sure that everything is cool and and just wonderful for overclocking in the card so for other stuff here you went through the phase design phase count on that are there any we were talking about the power balancing power monitoring is it true I heard that there is some more power balancing between the two power headers on the boards is that accurate yeah I don't know how you found that out but thank you for calling it out go ahead free advertising um yeah so there is we have new power balancing technology as well because each phase is around 30 watts it can be about 30 watts and and if you try to try to draw it from one connector or another connector depending on how the GPU is pulling power from the phases you might not be able to fully utilize all the current available on a particular input connector so we have this really high precision power balancing capability now that really improves the amount of it well we've had power balancing in the past but this is much faster and much more accurate so we're able to fully utilize the inputs available input power my last question for you back to the phase count so whenever we hear is only like twelve phase the instant assumption is it's doubled or is two PW PWM signal split between two fats or something like that are you guys doubling at all what's what's the what's the approach to 13 phase or 8 phase designs on these yeah so we have an all new power controller for for these cards and they're able to drive with just a lot more phases power doublers take up real estate on the PCB our phased uploads really take up real estate that we didn't want to spend we wanted to spend as much of that real estate and putting down fats and making it super efficient and in fact we also spent a bunch of space if you'll notice in the founders Edition on twice it's nearly twice as many capacitors and so that makes is the supplies ultra quiet and that actually also helps power efficiency so we wanted to spend the beat the board area that way instead of putting down like phase doublers and other things that we could have done and it's it's just not gonna as you pass current through the PCB it's less efficient as well so makes a lot of sense so for more information as always there's a link in the description below for our article and thank you Qi and Justin for joining me we'll see you all next time
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.