Gadgetory

hey Ron we're back for 23 episode of ask GN and this is the series where if you have questions about hardware technology news whatever post them below will try to get to it for the next video this one we're doing in the middle of a thunderstorm so i apologize if there's some background noise while i'm talking here but let's get into the first question is from Lucretia's 22 much for that name disgusts me or not is there any point to changing the core clock of a modern GPU for overclocking as opposed to just increasing the power target and or temperature limit since with GPU boost the card will overclock as far as it can anyway the short answer to this is yes depends on what you're doing so for the RX 480 as an example we tested this where because the if you just increase the power target it still got a range so it's saying I'm going to stay between X megahertz and why megahertz where x and y might be something like of top of my head out I think 12 1100 1200 to 1300 almost depend on what kind of offset you configure but it's got a range and so we just increase the power target and don't touch anything else on something like the RX 480 what will happen is it'll push as high as it can when it can and then when it feels as if some sort of thermal threshold is being hit it'll throttle back and so you lose some frames there you potentially introduce some low one percent or point one percent low frame rates which impacts your frame times which creates stuttering so the way to avoid that would be to go in there and also configure a higher fan speed or fan speed curve or manually configure the clock rate and set it to be something more stable and consistent or restrict the range of that clock rate so that's one reason on nvidia cards like the gtx 1080 it's a similar answer they had that voltage frequency curve these days that was new with boost through point O boost through point O and and these newest version of boosts are both pretty complex in the way they function versus previously but increase in the power target certainly will do a bit for you it'll sustain the high end of the configured clock rate we're longer depending on because it also increases the temperature limited step so it increases the two together so by increasing the temperature limit in the power target limit you're moving or reducing some mitigating factors two variables that could throttle back the clock rate so you'll sustain a higher overall clock but you'll get a bit just from doing the percentage offset I would recommend manually overclocking in general just as good practice because you'll get more out of it and you can kind of fine tune where the limits are next question is from at CBU Sybil you four days ago who says not sure if you will know much about this but here's a question with doing a more budget system for cad type software ptc creo specifically would you recommend buying the cheaper geforce card or with the cheaper geforce card be fine compared to getting a quadro card the projects right now our small mechanical parts so something like CAD computer-aided design or drafting I'm not an expert in but I can give you some advice based on personal experience with it to some extent with CAD software and with render software like premier or if you're working with solidworks blender anything like that one thing you'll want to check is first of all is the software khuda aur opencl accelerated to open CL and and it's supported well then AMD cards can work pretty well with it if it's CUDA accelerated then obviously you go nvidia and you fir for professional or production use cases you really want to buy your card based on what your software will support in terms of acceleration on the GPU on the hardware side so you don't want to push all that on the cpu so by your card based on that if it's good at let's say it's cuda accelerated i don't know if that's the case for ptc creo specifically but if it's cuda accelerated you're buying an nvidia card if it's opencl maybe consider am diva nvidia card quadro may quadro will work but it's always going to be more expensive than geforce car which i guess is where the passion comes from so as an example for adobe premiere we use geforce cards and even though quadra would kind of work a bit better the price offset isn't worth it to us so we just use g4s you can even use multiple of them if you want if the software supports multi-gpu it doesn't need to be at over an SLI bridge just multi-gpu MDA so GeForce can be as good or comparable to quadro but you'll want to check will the software specifically support g-force and then look around for benchmarks online unfortunately mountain expert in that area one thing i will say for adobe software in the past we've had to go into txt files and configuration files and add the card that we're using so for example in cs5 you have to go into a txt file add in geforce gtx 980 or 780 or whatever and at that point it will use that card for acceleration up until that point it would only detect and use certain quadro cards so you may have to change your software's configuration files to specifically detect and call a GTX card and obviously the psalter needs to support that at some level as well so basically the answer is the unfortunate you're gonna have to do some more research but GTX can work as well or close enough to a quadro card and performance where it may be worth buying just because it's cheaper especially if it supports multi-gpu next question is from Miguel Felix who says so I'm curious how much of an improvement can you expect from each release of drivers from nvidia or Andy would it be noticeable in terms of FPS yes it would be noticeable in terms of FPS we have tested some of this a good example that we tested recently as published was when doom came out before Vulcan doom came out with OpenGL only and the AMD r9 390 X we tested had pretty bad performance I think it was comparable to a gtx 960 which should never be the case and so Andy push drivers that improved performance so much i think it was twenty-five or thirty percent performance increase in FPS so it improved it so much that it outpaced the 960 and i think going from memory was tied close to whatever with the nine 70 so you can definitely expect bigger improvements than a zero FPS noticeable gain depends on the driver update depends on the game so a driver update doesn't just do a blanket to increase in performance for all games it will specifically target certain games target elements of that game whether it's post FX processing or texture or batch processing or whatever whatever the game is doing in both AMD and NVIDIA will tune for that that game specifically and then you get a performance improvement there's no real general you can expect X percent but I've seen as much as thirty percent and I've seen as low as one percent so definitely arrange next question is Alice star says about these aio GPUs that's all in one for those who don't know with liquid cooler about these aio GPS I've experienced issues with an aio cpu cooler which need topped up after only six months due to air bubbles getting caught in the pump what is the end causing noise what is the situation if this needs to be done with one of these aio GPUs that don't have a fill port so that's correct these generally an aio by a sort of nature of what it is an all-in-one it will not have a fill port it's not meant really to be an open loop system that you're filling and maintaining its supposed to be easily deployable you use it for like five years and then you throw it out and get something else with your new computer so they are not meant to be refilled that does mean that they have some advantages over open loops one is time to is price there's a big ones but they also have a biocide in them which I figure what it is on top of my head but they would biocide in them and that that helps make sure there's no bacteria buildup which you will sometimes get in open loops if you don't also put some of that chemical n would you can buy from Lowe's or something but so that's those are the advantages now in terms of topping them off you can't do that with these a iOS but you shouldn't need to the only reason you get air bubbles with noise that I can think of as on a consistent basis would be if you install the radiator upside down so the recommended installation for almost all radiators I've ever worked with let's say you're you've got a case here and I've actually got one here so we just built this for that video that was countering the motherboard article and if you mount your radiator in the back of the case and the rear fan spot what's going on then basically you get what's in this case here so you see I have the tubes down here that's where you want them if you install this with the tubes at the top of a rear-mounted or vertical radiator front-mounted tubes at the top can it makes the pump work harder the pump is not pulling the the liquid down and that can cause air to enter the sort of the system and we talked to rob teller from a stack about this but it's real problem causes air bubbles and that can create the noise you're talking about so just install it correctly and it shouldn't really be a problem next question is from Ramos arena skin of Rome who says serious question i'm hearing the I've heard that before i'm hearing the 490 might be 24 80s and one card is that possible if so it doesn't really fix the crossfire issue of old games not using multiple GPUs so what's the point so the first question is it possible yes you can definitely put two gps on one card and it's been done several times some of the old nvidia 90 series so i think the 690 maybe was one of them recently 590 those were two GPU cards as two GPUs on one PCB and then they share the ram or they have two separate i should say separate stacks of ram so it might be labeled as for example and they did this and they did this with the fire pro might be label that's like 12 or 16 gigabytes of erm but it's still independent effectively banks for each GPU but yeah it's possible and then the next question was doesn't fix the crossfire issue the issues you have with crossfire or sli you will pretty much also have with a dual GPU single card because it's still going to behave the same that needs to be tasked independently or correctly by the developers of the game anything that's got interdependent frames like post effects and things like that will be more difficult for sli and crossfire configurations to resolve and that includes multi-gpu single cards as for the 490 specifically I honestly haven't looked at any rumors I don't know anything about what it's supposed to be but if it is to RX for 80s then that would basically just be two of the Al's matter chips on a single card which andy has done in the past and it's normally been in the 90 series so that seems like a reasonable potential assumption and it would be more powerful you would get the same performance if you look at our our X 480 crossfire benchmark it would be almost the same performance with one potential change to that a potential change would be you're forcing a lot more data potentially through the PCIe slot so when you're pushing all this extra base because your dual processing you're doing AFR so it's one frame per GPU so that's more in theory double the the frame output from a single card not necessarily how it works in real life but that's the idea so when you increase the bandwidth and push it all through one port there is a chance that you lose some performance as opposed to two RX for 80s that have the exact same clock right and everything in crossfire so keep that in mind but with a dual GPU single card normally there's either a price benefits or you can just get two of them and now you have four cards that only take two slots so that's okay i guess the pen what you're trying to do hopefully that answers that but yes the last video at the end we edited in the 50,000 thing thank you all for helping us reach 50,000 subscribers we are already on the way well on the way to 60,000 so as always subscribe tell your friends to watch hey channeling the post real video on it helps out directly i'll see you all next time