You might also like our stupid Vacuum video. We're sad that it got so many views. Maybe give it some more? ua-cam.com/video/pm3icDVqyQ0/v-deo.html Patreon version is up! www.patreon.com/posts/17502842
Rich W: That doesn't fit Ask GN, and we filmed it before that news went out. As stated elsewhere, it's in our next HW News video. Those go up on Mondays.
For Ask GN, would thermal pads work fine if I'm working on multiple mobos and cpus, it seems like a waste of time and thermal paste setting up the cooler and cleaning. It would only be used for a short amount of time, like bios updates and to check if component are working properly.
Sure Lock well i guess all that would really be is a rental service. But if its your only vehicle i dont think youd net much profit vs the risk it poses of theft or accidents/wear and tear
Hey Steve, do you have any opinions on the articles released by HardOCP. Pcper, and Forbes (etc) on the NVidia Geforce Partner Program? It seems like an interesting topic that could greatly affect the consumer market as a whole.
and all GN will do is yet more damage control for Nvidia. mark my words. instead of pointing out how incidious this program is they will throw shades at HardOCP and probably also AMD.
I say we hold our tongue till we get more of a grasp of the situation. If the allegations are true, this could definitely mean something big for us consumers. Let's not try and swing statements like that around because we cannot throw judgement calls with material that has not been officially validated (for now). The things we want to be true may be different from the things that are actually true, so just calm down and sort the facts out of the wall of news.
jonrty007 and they never will be validated. Nvidia isn’t suddenly gonna come out “yeah, you got us guys. we tried to screw over the industry.” and neither will any OEM’s or AIB’s go on record publicly for this. they are too afraid. but remember Nvidia has throw out AIB’s before for going into nagotiation with AMD about offering their GPU’s as well. sadly the most “proof” we will get is how much this fits with their MO as well as taking our lesson from when Intel pulled this. the only good thing is this got busted before it could take effect. Its both sad and hilarious how people ask for proof when this kind of stuff comes out and its Nvidia behind it but when unverified accusations are made and its about AMD everybody jumps right onto it and no proof is needed. the accusation is proof enough.
Nvidia should stop GPP right now. For all of you who might not know what GPP is; GPP (Geforce Partner Program) is a contract in which OEMs (Asus, MSI, etc) are asked to only sell Nvidia products in their gaming line up. This means Asus would no longer be releasing and AMD ROG products and the same goes for of the other OEMs. If an OEM decide not to sign the contract they will basically be excluded from Nvidias network and since Nvidia is the largest GPU supplier it would hurt that specific OEM hard. This is some really shady business from Nvidia.
Nintendo did something really similar back in the NES days. Basically, if you released a game for the NES, you couldn't release it on another console, and they could punish you by taking your license away or shorting you on cartridges, since Nintendo controlled the manufacture of carts. They used the license to control all sorts of other aspects, as well, like censorship. They ended up getting sued pretty hard for anti-competitive practices.
How do you clean an anti-static mat? Say for instance the GN anti-static mat, or another popular brand (the one I currently have) such as ModRight's mod mat. As I believe they compose of the same materials (or maybe they don't, you could explain this as well) what is the best way to clean them? Thanks!
love this channel + love the ask GN segment. As soon as I finish school and get a job I'll be buying your products to help support you! keep up the good work
I have a question. Let's assume I have watercooling on CPU+VRM (monoblock) and GPU, and the radiator with fans is placed externally outside the case (it's way too big to fit in any case that exists). So there's no active airflow at all inside the case (no fans). There is some passive airflow - the case has perforated bottom and top covers, and the case is standing a few inches above the ground, so the air can naturally enter and leave the case. In that scenario, are there any components in the PC that can overheat due to zero active cooling inside the case?
Hey Steve, When i was stress testing my new cpu overclock a few months ago i noticed an odd phenomenon: With an unstable overclock my computer would crash in both Aida64 and Prime95(just the two tools i used), but with both of them running at the same time it was perfectly stable. I would think that having both tools stress test at the same time would be even more taxing if anything, so what caused this? My cpu is a Ryzen 5 1600 by the way.
Devin Dykstra where you hitting the thermal throttle limit? it could have been a case of two avx loads were causing so much extra heat that it forced a cpu downclock before it could crash due to high temps
I recommend running OCCT to check stability, just don't do 24 hs of that, is the most intensive program for overclocking, do 4 hours and if you get any erros you need to lower your frequency or up your voltage.
Nico O. This. Neither Prime or Aida are actually as stressful as people think. OCCT will pretty much guarantee you have a stable OC if it passes a 24 hour run
When running both at the same time the CPU has to interrupt one instruction to run the other. It's not running them at the same time but going back and forth. This is why you shouldn't have anything running while you're stress testing. I just use Handbrake because synthetics are so hot and 24 hours stable in LinPack or IBT doesn't guarantee stability. You could still crash in a game and likely will. How relevant each stress test is changes as architecture changes so it's usually best to look at the guides on overclock dot net pertaining to your (assuming recent Intel) CPU Darkwizzie does a great job on them and there's probably a lot of things you could learn that are very helpful.
Question for Ask GN: How important is the Hot Spot temperature on GPUs, as reported by monitoring software such as HWinfo? For example, on my Vega 64 Liquid Edition (factory water cooler), the GPU Hotspot temperature will hit a MAX of 105 C, average of 57 C during a 3DMark Firestrike benchmark. But during that time, the GPU Thermal Diode only hit a max temp of 49 C and the HBM only reached 45 C. Should I be concerned? I am coming from an R9 390X so I am used to 90 C temps but 105 seems quite bad, especially considering that it is a water cooled card!
Steve really blew answering the question about CPU's versus GPU's. The reason AMD and Intel do not have AIB partners is because they do. Nvidia and AMD have AIB partners for graphics cards exactly like AMD and Intel have partners who create MoBos. Nvidia and AMD make the GPU which is actually the silicon chip on the graphics card not the whole card exactly like the CPU which gets put in the socket on the MoBo. The difference is the GPU is soldered into the card at the factory and the CPU is installed by the enduser or by the system integrator.
Also don't forget about chipsets and other components which are mandated by Intel for the motherboards. All mobo manufactures are basically AIB partners.
It was "mentioned", but in a very tortured and round about way. It could have been so much more clear cut. I mean the first thing you should have done is beat his face in for confusing GPU versus graphics card. It's really not something people who talk about "enthusiast features" should be mixing up.
Question for Ask GN How big was the leap between the previous best apu's (which I think was Intel 630?), to today's current best apu's (ryzen 5 2400g)? If we had that same proportional leap in performance with a new apu from teh 2400g, what kind of performance would this theoretical apu have? Are we talking 1050ti territory?
If anyone's curious, what Steve didn't mention about the Skyrim physics problem is that theirs a way to fix it by changing a havok physics setting in a .ini file. You can look up the exact details of the fix by going to the "PCgamingwiki" site
Considering what you said about being focused on games and knowing about game settings, would you guys consider doing a game settings guide talking about the different settings and what they do with respect to the hardware?
@Steve Burke continuing the "high end APU" Question. Is it physically possible to fit a 8 core 16 thread ryzen die, and Vega 56 with HBM2 on the same substrate? Forget the theoretical limitations it might have will it fit under threadripper IHS?
For the CPU manufacturers question, the real answer comes down to you cannot sell the silicon without packaging it, because the part of the die that is facing down must be protected from all contaminants and it would be very cost prohibitive to setup some way to take individual dies from Intel or AMD, remove any protective material from the die and package it. Basically the failure rates and costs of doing that at a separate factory than where the die is manufactured would be far too high to make it worth doing. That being said I do not know of a reason you can't have a company that removes the stock IHS and puts on a different one or changes the TIM used, other than this would be very expensive parts.
Does anyone know if coil wine is a sign of a dying or faulty gpu? I also just installed a new power supply seasonic prime 1000w and now my fans rev up when starting up the pc ,is this normal ?
Conductonaut comes with a disclamer not to use with aluminuim heatsinks. It would be interesting to see what it would do to one over a longer period of time. There were a couple of videos out but they kept the system running around 24h which is not enough to have a good insight on what would happen for a longer time.
Steve what do use as your personal rig? Would be interesting to know considering all of the different hardware that you cover on the channel and to see what components you prefer personally.
I know you answered a question similar to this. As a preface, I get a few folks asking me what they should upgrade their system to for streaming. They say that their PC place games fine when they aren't streaming but once they are they drop frames and-so-and-so-forth. Usually I just suggest that they get another cheaper system that captures, and encodes, and streams the picture instead of buying a more expensive processor. To me, having a budget of say 1.250 a user could spend 750 on a "gaming" system and 500 on a a used or just cheaper capturing, encoding, and streaming system and get a more fluid and pleasurable experience, given a lot of streamers I know of prefer 720p60fps over 1080p30fps. I am curious if you would do a content piece on this type of price point and compare the performance of such a concept? I know there are lot of things to go along with this, such as does the person edit and upload highlights and do a lot of after affects et cetera.
Would you show us guides on how to repaste certain laptop coolers (with normal or liquid metal paste), as iFixit doesn't have the tutorials on how to dismantle some gaming laptops like the Asus ROG GL503VS Scar Edition or the MSI GE63VR Raider.
@Multiple fan header question: As there probably is a fuse somewhere you won't burn the motherboard, but you will kill the fan header. I mixed up the plug for a fan up with a plug for my D5 pump on a MSI Xpower motherboard once (I blame bad cable management). The motherboard survived just fine, but the fan header was useless after booting the computer for the first time. The fan headers for the Xpower mobos are rated for 2A and the pump at 23W, but I suspect that the startup current of the pump is higher than its rated 1.9A (equivalent of 23W at 12V). Also: I'm an idiot.
Hey Steve, what settings you recommend tweaking when overclocking a CPU and which ones you recommend leaving untouched? I watched a lot of CPU guides for overclocking CPU's and people just talk about increasing core voltage and frequency, but in my experience tweaking the LLC Level was crucial to find stability and lower voltages when idle, also there isn't a clear answer anywhere on how to set Adaptive Voltage when OC is done. Thanks Man.
I'm having a hard time figuring out how to use all the fan headers on my Maximus X Hero. Any help? My H150i AIO suggests connecting it to the CPU_FAN header (I assume just to report pump speed?). How would the CPU_OPT header operate then? It has shared control with CPU_FAN, but CPU_FAN isn't actually controlling anything. If I put case fans on the AIO_PUMP or W_PUMP+ headers, they default to full speed and won't slow down until I enter Windows and AI Suite kicks in. The BIOS doesn't seem to allow fan curves for the AIO_PUMP and W_PUMP+ headers. If I instead put the AIO on the AIO_PUMP header, any case fans I put on CPU_FAN seem to vary the speed too fast and ignore hysteresis. People act like you can use any of these headers for case fans, but any headers beyond CHA_FAN's have given me trouble. I eventually just gave up and used a splitter, but I shouldn't have to with all these empty headers. I don't know why the have to complicate them with settings for water cooling and shared control. How do I best use them as plain old fan headers?
Hey Steve, Your discussion of GNs origins made me wonder, where/how would you recommend someone start learning about more detailed info on different game engines/graphics/programming? Was your education on games organic or did you have a more structured plan? I have a full-time job and am considering playing around with game design as a hobby. The problem I run into is that while I'm not completely clueless, my knowledge is very haphazard and inconsistent. I am struggling to find good free learning resources online (eg. subreddits/youtube channels/dedicated websites/etc.). The other thing I struggle with is not knowing an engine to use, right now I'm thinking Unity would be good to just start somewhere. Spending to join something like Lynda isn't out of the question, I just don't know what the best resources are short of going to an actual school. I would love to experiment with different ideas for games similar to something like Stardew valley, Factorio, Oxygen not Included and other fun non-AAA steam games. I expect to fail miserably but want to have fun and try anyway. P.S. I know some of those games are pretty different, I just listed some examples off the top of my head.
I remember when I met the fps related physics issue the first time. I've tried the good old Vice City on my current build. Without frame limiter the fps got up to 500-1000 range and cars were moving like a huge inert mass.
Regarding the message they sent you about the boot kit: That makes it sound like not sending it back means you won't be allowed to use ANY of their warranty services in the future. Are they legally allowed to do that? It seems fair (at least on the surface), but I'm just curious if the law is on their side (or if I'm misunderstanding).
8:30 No, it's not "400fps" where you get physics bugs, it's pretty much ANYTHING over 60fps. I was trying to play it at just 90fps and it still bugs out often enough that it's almost impossible to get past the opening wagon ride.
I reckon that once AMD has integrated Infinity Fabric into their GPU technology, we'll see a similar approach to GPU construction. At that point you could devise some sort of twin-die socket with the CPU-module on the one (composed of however many Zen- or further iterative cores happen to be available) and the GPU-module on the other, right next to each other. You can even design the GPU as such that you only use 1-10% of the actual hardware for "iGPU'" workloads to keep power consumption down. But high-end? Let's make a high-end product first.
Vega doesn't have any Infinity Fabric in it, but that isn't why Vega flopped. Mostly overvolting out of the box, and a poor reference cooler combined with the mining boom and RAM shortage limiting supply to the normal consumer. It is however something that will help AMD make bigger and more powerful GPUs in the future without having to worry about yields. Die shrinking has the downside of making monolithic products (like Intel's Skylake-X CPUs) very difficult to manufacture. With smaller die sizes however, the yields are far greater, which means less money is lost per wafer and "more" can be made for "cheaper", and thus sold "cheaper" if AMD chooses so. Last I heard Navi is getting the first GPU implementation of Infinity Fabric, and the design after that will be brand new instead of the umpteenth iteration on GCN.
I have had two Corsair AIO fail, the first I replaced, the second I sent back, back when the 4930k was used to heat a small apartment building, AIO were a must, is that still true today. The same 4930K is overclocked and stable using the Noctua NH-U14S, did I buy in to the HYPE?
Could you make a video comparing some of the more popular cases with the maximum size & number of fans rather than just "stock"? I feel like some cases only come with 1 fan stock and no one I know runs their case with only stock fans (most go in the trash).
As for APUs becoming "High End" I could see a MCM that has CPU+GPU+Memory modules under the IHS. This would be a way to create smaller, cheaper silicon modules and "glue" them together. I suspect they have already been attempting to use Infinity Fabric for a GPU module and this would be a drastic improvement at lowering the cost of consoles.
The question at about 21:00 makes think of this: instead of APUs, do you think socketing GPU chips on motherboards could become an option? Instead of using a PCIe slot for graphics cards, there could be another socket for GPU chips (makes me think of old math coprocessors.) The motherboards would need to change quite a bit to offer proper power delivery for this new component, and there would be extra memory slots besides RAM ones: the (HBM?) DIMMs :) The benefit of all that: upgrade paths would have quite a different granularity compared to the replacing entire graphics cards of today. You can choose your own cooler just for the GPU chip. You could theoretically increase how much memory your GPU has, without changing the GPU chip... The fact that both GPU and CPU chips would be parallel on the same board plane instead of perpendicular to each other could also possibly offer easier cooling arrangements
13:12 Because a CPU is a silicon core on a relatively simple circuit board with not much room for variability. The variable is the motherboard which provides the memory, voltage controllers, power delivery, etc. A GPU in its turn is a different type of relatively simple processor which requires its own "motherboard" with memory, power delivery, voltage controllers, etc. So the silicon in GPUs (and CPUs) is practically the same, what differs is the motherboard - Intel/AMD make the silicon product, partners make the motherboards, and it is valid for both GPUs and CPUs.
Any chance you guys can take a somewhat older case that has side panel fan mounts and test with a system built with a GPU with an open air cooler design, where side panel fans are set up as intake, and then again as exhaust? maybe even mix up the testing wehre cpu is air cooled and also using AIO on cpu in the above mentioned tests. It's a lot of testing and I get it if you don't have the time to do it, but it seems like it would be interesting content to see. If you've already done a video on it, point it out to me I'd like to see it. I've got the Antec 1100, but I don't have the disposable income or industry connections to test any of this out. It woudl be interesting to see just how much difference(s) it would make by having that extra airflow around the GPU.
Is it normal that the HDD is the loudest part of my system ? I have 7 fans in my system running at around 700 to 800 rpm and you can still hear the HDD over them
Actually ARM CPUs are not manufactured by ARM, but rather licensed out to other Manufacturers. So the CPU "AIB" stuff kinda exists. Just not for x86 AFAIK
Blessman Fred Yea, They have misleading specs however, as a proper comparison of ARM against x86 Would mean a computation per clock ratio of 3 of Arm : against 1 of x86 Wow that got unexpectedly complex... o:
Hey Steve. I have a hopefully simple question that I would like you to answer in the next "Ask GN" episode. It is: Where can I find what the IPC of a given CPU is? As far as I know, neither AMD or Intel list the IPC numbers for any of their processors anywhere, so I was wondering if you would know of any place to find those numbers. Thanks!
Why not integrate hbm on a cpu die since its so much faster than ddr; use the hbm as a cache etc then use ddr as expandable memory; when needed/applicable. The one question about apus went into some detail; but such a question too would be relevant to the question of "custom cpus" As you said it is hard to cram gpu and cpu (apu) in the size of an apu but in a cpu without a gpu there should be plenty of room to integrate hbm?
Hey Steve, how do you think LM and Copper Heatsink will go in long run? I ask specifically for gaming laptops that have no IHS on CPU but do come with copper heatsinks. Consider there is no leakage of LM onto board as necessary precautions have been taken care of.
Hey Steve, I am in the market for a new phone and as funny as it might sound I was checking out new phones and noticed that they are shipped with different processors, different gpu's and so on. This potentially creates a new segment in the review community that hasn't emerged yet as much - testing the cpu, gpu of a mobile phone, if something I would love to see GN do a segment on those. My question is - is GN planning to look into mobile hardware since it is getting better and more powerful each iteration and there are a few different configurations already(Mali/Adreno/Bionic custom GPU), since you already have a wide and broad experience in testing? Thank you.
A GPU is like a computer on its own. It has the processor chip (GPU). It has its own RAM. It is on a PCB. It has output ports. It has extra features. A CPU is just the processor chip. So, I suppose, it makes sense to have external companies assemblies the video card.
I was wondering about the other part of APUs. Obviously the graphics will never be able to compete with discrete graphics of the same generation but will the CPUs ever catch up? Will we ever get a 6 or 8-core Ryzen APU? They've all been in the lowish end 4-core segment for a really long time.
@Gamers Nexus I'm sleeving my own full-length custom cables for my pc but I'm stuck. Two questions for you: why is it that some (but not all) 24-pin ATX cables have 24 pins on the mobo but on the PSU side it's 10 + 18 pins? Also, why is it that certain ATX cables, like my 18awg EVGA cable set has 27 wires on the 24-pin connector (which means that 3 of the wires are doubled)? I haven't seen a single video that clearly explains PSU cables - there's a lot of intricacies that often go overlooked, I'd love to hear your explanations - they always rock. :)
so i was wondering what thermals would be like on solid front panel cases especially ones with out power supply shroud with bottom fan mounts if you reversed the airflow and had front exhaust with bottom and rear intake could you test this out
Thankyou very much for answering my question! I have now watched this video on the cost of HBM2 vs GDDR5 and I see a potential benefit of HBM for end users. But in the implementation from AMD vs Nvidias GDDR5 implementation it appears that there is no end user benefit to HBM over GDDR5. Is that fair?
Hey Steve, I have a question about the CPU/GPU dies... why the chip is that smaller from the actual PCB that is underneath? A bigger die will not spread the heat way better than a small die with stuck of things one of top of the other? 🤔
idea, Threadripper APU, take the 2 "spacer" dies and put Graphics dies in those locations, basically take a GPU half of the APU die that is twice the size of the one in the current APUs so it's a full die size then crossfire them together through the infinity fabric or some crazy shit like that.
Has anyone had heat spreader fall off of Corsair Vengeance tall memory. Mine did when thermal tape let go. Machine wouldn't post until I removed them. May have shorted component near memory or on memory causing no post. Memory survived. Might make for a good topic for Ask GN.
Hey Steve, What do you think of BCLK overclocking on Ryzen and what benefits do you think it offers? Also do you BCLK overclock on any of your computers.
I think the only way an APU would be able to compete today is on a platform like the TR-4 socket, where the substrate is big enough to handle more GPU/HBM onboard RAM and such. How the power delivery would work and how to cool something that monstrous, I'm not sure about. It could be an interesting APU, but would it be cost effective and worth the trouble? I have my doubts.
What do you think of the battle between AMD and Intel with regards to the amount of cores? While both parties are adding more and more cores, little is being done on the (consumer) software side to actually make proper use of all of them. Especially beyond 8 threads there are very little programs that utilize all cores. I would personally rather see them focus on making the cores themselves faster and adding more features to them, than increasing the costs without actually having a big real-world performance boost. I know that the frequency is nearing the limits of the semiconductors on the current transistor sizes, but I suspect there has to be quite some headroom left in the architecture and chip layout.
Question for AskGN So a few months ago when i was installing a new motherboard, i accidentally bent my usb 3 header and now can't connect my usb 3 ports on my case. Is it a good idea to just leave it or bend it back using tweezers or something similar?
Hey Steve, could a pc be cooled with an air sealed case with strong airflow through a water cooling radiator, with the heat being dumped outside of the case? (I.e. larger radiator outside, or well water?
#AskGN in regards to the High end APU question. What if AMD used a TR4 sized substrate and socket for an APU? What do you think they could realistically fit on there?
Hey @gamersnexus I will be performing my first custom water cooled pc soon, and was wondering about bacteria growth. I've heard about putting in chemicals to prevent growth, but would passing the clear tube through a strong uv light tunnel kill any growth effectively? Thanks for the input
Difference between gddr5 vs ddr4? yes we know ones for graphics and ones for system, but what makes them good at those things? I know their manufacturing is similar, but how do they differ? And with the release of ddr6 (ddr5? Cant remember) is it possible that we are working towards a "jack of all" memory that could handle both?
I'm sure this has been asked before, but do you think sometime in the future graphics cards will become customizable, Similar to how cpu's are? You would buy your own gpu chip, board, memory, cooler, etc. Is there a reason why this is not already a thing? Thanks.
The physics solution for the sake of fluidity is calculated every frame cycle. Physics solvers are designed to work with 1m, 1s scales. Increasing or decreasing that resolution results in pronounced floating point overflow and loss of precision. So, technically that frame rate limitation is computer architecture limitation. Bullet, for example, has a scaling parameter which keeps dimensions at optimal scales, however, changing that at runtime is a NO-NO.
18:14 Will APU's Go High End? With the near release of Intel Kaby Lake G with Radeon Graphics (not really an APU), and the rumors of an AMD Fenghaung APU with 28CU's + HBM2, I would not consider it high end for a desktop, but for laptops, definitely in the upper half. Do you think we will see this kind of processor soon from AMD? As HBM would really help these APU's with there biggest weakess of memory bandwidth. And also, do you think we will see these Fenghaung APU on AM4 or TR4? AMD has teased 32 core "Exascale Heterogeneous Processor" before.
hey steve with the r3 2200g or even the 2400g how big of an air cooler would you need to effectively cool it with out the fan while overclocked on both cpu and gpu while under full load im just looking for basically total silence and just having the one rear fan as an intake and nothing else probably use the dark base 600
Hey Steve, do you think that the Ryzen Threadripper "2" being released in the future will still use DDR4? If I was investing into a TR-4 system now but didn't really want to spend a lot on a fancy motherboard (although, from what I have read TR-4 will be used for both Threadripper 1 and Threadripper 2) and just get a decent Threadripper now. Switching out the MB in the future to take advantage of whatever new features Threadripper 2 has. At least then I can just move the DDR4 RAM, Graphics Card, Power Supply and Case from the old one to the new one, buying new MB and CPU only.
I don't agree. The point of going bigger package size would be to fit a more robust GPU in there. I'm not saying make one for Tr4 platform, but what if the package were that size. Beyond that, nobody said it would have to be budget at that point. Even then, if they could fit significant performance in the package, say an r7 and vega 56 class for ~$429, I would consider that budget anyway.
I caught a fan on fire once. It was a 3.5-3.6v (or somewhere in that range) and I hooked up 5v power to it and the cable literally went up just like a fuse going to a bottle rocket. Scary as hell and sucked since that fan couldn't be replaced.
What about an APU with HBM? Is sort of "on chip memory" more likely to happen with cpus first or do they really not need say 2GB of "cache/ultrafast sysem memory?"
The CPU question at 13:12 technically does happen with CPU, or rather the chipsets on the motherboard. The 'AIB' partners in this case would be the Mobo manufacturers.
When do you see developers utilizing more than four cores for games? Assuming that CPUs keep going the way they're going. Any predictions on what they'll be able to do with the new resources available?
#askgn Hi! Why AMD or Intel cant make processors without IHS like in GPU's? I mean not all of them just like "special edition" or part of HEDT just to have choice. I know that IHS shields the die so it wont break but its not a big problem looking at how its going with graphic cards that have big non referent cooling. Will there be any Kaby Lake G mini itx boards? Also can we count mobo's with soldered processors (ie. in laptops or NUC's) as CPU AIB?
Steve do you think it's possible to desolder a cpu from a laptop with a higher end i5, with a better i7 model? I've seen some pretty awesome videos of how to desolder, and reball the cpu points on the pcb and reinstall the same cpu that had stopped working and it fixed it. I think it would be difficult but maybe possible? Would you ever attempt to try it?
Can you do some t3sting and comparisons for replacing the VGA cooler with something like the Rajintek Morpheus 2? On Cards like Galax 1080ti HOF or Evga SC2 1080ti cooler replaments. How does card perform n vrms cooling etc. Are they worth it
You might also like our stupid Vacuum video. We're sad that it got so many views. Maybe give it some more? ua-cam.com/video/pm3icDVqyQ0/v-deo.html
Patreon version is up! www.patreon.com/posts/17502842
Steve, are you planning to address the GeForce Partner Program leak? I had really hoped to see it mentioned in this Ask GN.
Rich W: That doesn't fit Ask GN, and we filmed it before that news went out. As stated elsewhere, it's in our next HW News video. Those go up on Mondays.
Rumor has it the Thermaltake lvl 20 case is being delayed by a sudden shortage of Dyson fans.
For Ask GN, would thermal pads work fine if I'm working on multiple mobos and cpus, it seems like a waste of time and thermal paste setting up the cooler and cleaning. It would only be used for a short amount of time, like bios updates and to check if component are working properly.
Your vacuum video sucks hard 😂
If you want to make some good money in your store you should sell Miner Voodoo Dolls
or sell old voodoo video cards
John Lacquey I truly believe you might be on to something here🤔
Better get on it, quick, before your idea gets stolen. 🕵️♂️
Just give me a little credit, or a 1080 ti, I'll be happy.
😣🤔
I walked into frys electronics and went to the graphics card section. The shelves looked like a ghost town.
Mankest Demes 50000 cards were stored here, now it's a ghost town.
Lukas Tran This was Fry’s in Duluth GA. Not sure about other places 🤔
Those miners *POISONED OUR WATER SUPPLY, BURNED OUR CROPS, AND BROUGHT THE PLAGUE UNTO OUR WALLETS*
they did?
John Galt
>not letting a stranger drive your car when youre not using it
And people pretend their logic is flawless
joshcogaming I would let a stranger drive my car if they paid and were guaranteed to return it.
Sure Lock let us know how that goes
Sure Lock well i guess all that would really be is a rental service. But if its your only vehicle i dont think youd net much profit vs the risk it poses of theft or accidents/wear and tear
Hey Steve, do you have any opinions on the articles released by HardOCP. Pcper, and Forbes (etc) on the NVidia Geforce Partner Program? It seems like an interesting topic that could greatly affect the consumer market as a whole.
It's included in tomorrow's news video!
Great! Looking forward to hearing your insights on the issue!
and all GN will do is yet more damage control for Nvidia. mark my words. instead of pointing out how incidious this program is they will throw shades at HardOCP and probably also AMD.
I say we hold our tongue till we get more of a grasp of the situation. If the allegations are true, this could definitely mean something big for us consumers.
Let's not try and swing statements like that around because we cannot throw judgement calls with material that has not been officially validated (for now).
The things we want to be true may be different from the things that are actually true, so just calm down and sort the facts out of the wall of news.
jonrty007 and they never will be validated. Nvidia isn’t suddenly gonna come out “yeah, you got us guys. we tried to screw over the industry.” and neither will any OEM’s or AIB’s go on record publicly for this. they are too afraid. but remember Nvidia has throw out AIB’s before for going into nagotiation with AMD about offering their GPU’s as well. sadly the most “proof” we will get is how much this fits with their MO as well as taking our lesson from when Intel pulled this. the only good thing is this got busted before it could take effect. Its both sad and hilarious how people ask for proof when this kind of stuff comes out and its Nvidia behind it but when unverified accusations are made and its about AMD everybody jumps right onto it and no proof is needed. the accusation is proof enough.
Nvidia should stop GPP right now.
For all of you who might not know what GPP is;
GPP (Geforce Partner Program) is a contract in which OEMs (Asus, MSI, etc) are asked to only sell Nvidia products in their gaming line up. This means Asus would no longer be releasing and AMD ROG products and the same goes for of the other OEMs. If an OEM decide not to sign the contract they will basically be excluded from Nvidias network and since Nvidia is the largest GPU supplier it would hurt that specific OEM hard.
This is some really shady business from Nvidia.
Nintendo did something really similar back in the NES days. Basically, if you released a game for the NES, you couldn't release it on another console, and they could punish you by taking your license away or shorting you on cartridges, since Nintendo controlled the manufacture of carts. They used the license to control all sorts of other aspects, as well, like censorship. They ended up getting sued pretty hard for anti-competitive practices.
Asus, MSI, etc. aren't OEMs though. If they were you woulndt find Asus on the package of an Nvidia card.
Or did i get something wrong?
How do you clean an anti-static mat? Say for instance the GN anti-static mat, or another popular brand (the one I currently have) such as ModRight's mod mat. As I believe they compose of the same materials (or maybe they don't, you could explain this as well) what is the best way to clean them? Thanks!
love this channel + love the ask GN segment. As soon as I finish school and get a job I'll be buying your products to help support you! keep up the good work
I have a question. Let's assume I have watercooling on CPU+VRM (monoblock) and GPU, and the radiator with fans is placed externally outside the case (it's way too big to fit in any case that exists). So there's no active airflow at all inside the case (no fans). There is some passive airflow - the case has perforated bottom and top covers, and the case is standing a few inches above the ground, so the air can naturally enter and leave the case. In that scenario, are there any components in the PC that can overheat due to zero active cooling inside the case?
Hey Steve, When i was stress testing my new cpu overclock a few months ago i noticed an odd phenomenon:
With an unstable overclock my computer would crash in both Aida64 and Prime95(just the two tools i used), but with both of them running at the same time it was perfectly stable.
I would think that having both tools stress test at the same time would be even more taxing if anything, so what caused this?
My cpu is a Ryzen 5 1600 by the way.
How many tests did you run?
It seems like that may have been luck of the draw, but I may be wrong.
Devin Dykstra where you hitting the thermal throttle limit? it could have been a case of two avx loads were causing so much extra heat that it forced a cpu downclock before it could crash due to high temps
I recommend running OCCT to check stability, just don't do 24 hs of that, is the most intensive program for overclocking, do 4 hours and if you get any erros you need to lower your frequency or up your voltage.
Nico O. This.
Neither Prime or Aida are actually as stressful as people think. OCCT will pretty much guarantee you have a stable OC if it passes a 24 hour run
When running both at the same time the CPU has to interrupt one instruction to run the other. It's not running them at the same time but going back and forth. This is why you shouldn't have anything running while you're stress testing. I just use Handbrake because synthetics are so hot and 24 hours stable in LinPack or IBT doesn't guarantee stability. You could still crash in a game and likely will.
How relevant each stress test is changes as architecture changes so it's usually best to look at the guides on overclock dot net pertaining to your (assuming recent Intel) CPU
Darkwizzie does a great job on them and there's probably a lot of things you could learn that are very helpful.
Question for Ask GN:
How important is the Hot Spot temperature on GPUs, as reported by monitoring software such as HWinfo? For example, on my Vega 64 Liquid Edition (factory water cooler), the GPU Hotspot temperature will hit a MAX of 105 C, average of 57 C during a 3DMark Firestrike benchmark. But during that time, the GPU Thermal Diode only hit a max temp of 49 C and the HBM only reached 45 C. Should I be concerned? I am coming from an R9 390X so I am used to 90 C temps but 105 seems quite bad, especially considering that it is a water cooled card!
The first question was spot on and completely got it right.
Steve really blew answering the question about CPU's versus GPU's. The reason AMD and Intel do not have AIB partners is because they do. Nvidia and AMD have AIB partners for graphics cards exactly like AMD and Intel have partners who create MoBos. Nvidia and AMD make the GPU which is actually the silicon chip on the graphics card not the whole card exactly like the CPU which gets put in the socket on the MoBo. The difference is the GPU is soldered into the card at the factory and the CPU is installed by the enduser or by the system integrator.
Ken Shaw this was the answer I was expecting. He eluded to it when he was talking about Intel motherboards.
We are all thinking the same thing
That was mentioned.
Also don't forget about chipsets and other components which are mandated by Intel for the motherboards. All mobo manufactures are basically AIB partners.
It was "mentioned", but in a very tortured and round about way. It could have been so much more clear cut.
I mean the first thing you should have done is beat his face in for confusing GPU versus graphics card. It's really not something people who talk about "enthusiast features" should be mixing up.
That timestamp overlay is 🔥🔥🔥🔥🔥
Oh, the first question here is so relevant to today!
How things change, eh Steve?
Question for Ask GN
How big was the leap between the previous best apu's (which I think was Intel 630?), to today's current best apu's (ryzen 5 2400g)? If we had that same proportional leap in performance with a new apu from teh 2400g, what kind of performance would this theoretical apu have? Are we talking 1050ti territory?
If anyone's curious, what Steve didn't mention about the Skyrim physics problem is that theirs a way to fix it by changing a havok physics setting in a .ini file.
You can look up the exact details of the fix by going to the "PCgamingwiki" site
Considering what you said about being focused on games and knowing about game settings, would you guys consider doing a game settings guide talking about the different settings and what they do with respect to the hardware?
Thoughts on the Nvidia GeForce Partner Program?
Discussed in tomorrow's news video!
Thanks!
@Steve Burke continuing the "high end APU" Question. Is it physically possible to fit a 8 core 16 thread ryzen die, and Vega 56 with HBM2 on the same substrate? Forget the theoretical limitations it might have will it fit under threadripper IHS?
For the CPU manufacturers question, the real answer comes down to you cannot sell the silicon without packaging it, because the part of the die that is facing down must be protected from all contaminants and it would be very cost prohibitive to setup some way to take individual dies from Intel or AMD, remove any protective material from the die and package it. Basically the failure rates and costs of doing that at a separate factory than where the die is manufactured would be far too high to make it worth doing. That being said I do not know of a reason you can't have a company that removes the stock IHS and puts on a different one or changes the TIM used, other than this would be very expensive parts.
Does anyone know if coil wine is a sign of a dying or faulty gpu? I also just installed a new power supply seasonic prime 1000w and now my fans rev up when starting up the pc ,is this normal ?
Viking
Fans reving up to max-rpm during boot is completely normal
Teemu Keinänen thank you !
Conductonaut comes with a disclamer not to use with aluminuim heatsinks. It would be interesting to see what it would do to one over a longer period of time. There were a couple of videos out but they kept the system running around 24h which is not enough to have a good insight on what would happen for a longer time.
You and Buildzoid keep tempting me with the X299 DARK!
Solder is a few degrees behind due to the lower distance between the die and the IHS compared to solder.
Steve what do use as your personal rig? Would be interesting to know considering all of the different hardware that you cover on the channel and to see what components you prefer personally.
Actually, I'd love to see a video on what these graphic settings mean and what performance/hit you can expect from them
I know you answered a question similar to this. As a preface, I get a few folks asking me what they should upgrade their system to for streaming. They say that their PC place games fine when they aren't streaming but once they are they drop frames and-so-and-so-forth. Usually I just suggest that they get another cheaper system that captures, and encodes, and streams the picture instead of buying a more expensive processor.
To me, having a budget of say 1.250 a user could spend 750 on a "gaming" system and 500 on a a used or just cheaper capturing, encoding, and streaming system and get a more fluid and pleasurable experience, given a lot of streamers I know of prefer 720p60fps over 1080p30fps. I am curious if you would do a content piece on this type of price point and compare the performance of such a concept?
I know there are lot of things to go along with this, such as does the person edit and upload highlights and do a lot of after affects et cetera.
Would you show us guides on how to repaste certain laptop coolers (with normal or liquid metal paste), as iFixit doesn't have the tutorials on how to dismantle some gaming laptops like the Asus ROG GL503VS Scar Edition or the MSI GE63VR Raider.
@Multiple fan header question: As there probably is a fuse somewhere you won't burn the motherboard, but you will kill the fan header. I mixed up the plug for a fan up with a plug for my D5 pump on a MSI Xpower motherboard once (I blame bad cable management). The motherboard survived just fine, but the fan header was useless after booting the computer for the first time. The fan headers for the Xpower mobos are rated for 2A and the pump at 23W, but I suspect that the startup current of the pump is higher than its rated 1.9A (equivalent of 23W at 12V). Also: I'm an idiot.
Hey Steve, what settings you recommend tweaking when overclocking a CPU and which ones you recommend leaving untouched? I watched a lot of CPU guides for overclocking CPU's and people just talk about increasing core voltage and frequency, but in my experience tweaking the LLC Level was crucial to find stability and lower voltages when idle, also there isn't a clear answer anywhere on how to set Adaptive Voltage when OC is done. Thanks Man.
Why aren't GPUs socketable? Why can't I choose what card board, cooler, and vram I want?
I'm having a hard time figuring out how to use all the fan headers on my Maximus X Hero. Any help? My H150i AIO suggests connecting it to the CPU_FAN header (I assume just to report pump speed?). How would the CPU_OPT header operate then? It has shared control with CPU_FAN, but CPU_FAN isn't actually controlling anything. If I put case fans on the AIO_PUMP or W_PUMP+ headers, they default to full speed and won't slow down until I enter Windows and AI Suite kicks in. The BIOS doesn't seem to allow fan curves for the AIO_PUMP and W_PUMP+ headers. If I instead put the AIO on the AIO_PUMP header, any case fans I put on CPU_FAN seem to vary the speed too fast and ignore hysteresis.
People act like you can use any of these headers for case fans, but any headers beyond CHA_FAN's have given me trouble. I eventually just gave up and used a splitter, but I shouldn't have to with all these empty headers. I don't know why the have to complicate them with settings for water cooling and shared control. How do I best use them as plain old fan headers?
This channel is so high end!
Question: could you please revisit the Dark Base 700 with more fan configurations?
I would love to see the potential of this case.
Hey Steve,
Your discussion of GNs origins made me wonder, where/how would you recommend someone start learning about more detailed info on different game engines/graphics/programming? Was your education on games organic or did you have a more structured plan?
I have a full-time job and am considering playing around with game design as a hobby. The problem I run into is that while I'm not completely clueless, my knowledge is very haphazard and inconsistent. I am struggling to find good free learning resources online (eg. subreddits/youtube channels/dedicated websites/etc.). The other thing I struggle with is not knowing an engine to use, right now I'm thinking Unity would be good to just start somewhere.
Spending to join something like Lynda isn't out of the question, I just don't know what the best resources are short of going to an actual school. I would love to experiment with different ideas for games similar to something like Stardew valley, Factorio, Oxygen not Included and other fun non-AAA steam games. I expect to fail miserably but want to have fun and try anyway.
P.S. I know some of those games are pretty different, I just listed some examples off the top of my head.
I remember when I met the fps related physics issue the first time. I've tried the good old Vice City on my current build. Without frame limiter the fps got up to 500-1000 range and cars were moving like a huge inert mass.
Regarding the message they sent you about the boot kit: That makes it sound like not sending it back means you won't be allowed to use ANY of their warranty services in the future. Are they legally allowed to do that? It seems fair (at least on the surface), but I'm just curious if the law is on their side (or if I'm misunderstanding).
8:30
No, it's not "400fps" where you get physics bugs, it's pretty much ANYTHING over 60fps.
I was trying to play it at just 90fps and it still bugs out often enough that it's almost impossible to get past the opening wagon ride.
It's only an issue if the devs program the game to base physics calculations on a hardcoded frame rate. Q3, UT99 run just fine at 600fps.
Those Anti-Static Modmats look great. I use a Venom Anti-Static Conqueror mat, and they're superb, wouldn't go back to a regular mouse mat.
What chair do you and your crew work/recommend for daily use?
Thank you for your continued great work!
I reckon that once AMD has integrated Infinity Fabric into their GPU technology, we'll see a similar approach to GPU construction. At that point you could devise some sort of twin-die socket with the CPU-module on the one (composed of however many Zen- or further iterative cores happen to be available) and the GPU-module on the other, right next to each other. You can even design the GPU as such that you only use 1-10% of the actual hardware for "iGPU'" workloads to keep power consumption down.
But high-end? Let's make a high-end product first.
Isn’t that how Navi is supposed to work? And why Vega flopped so hard?
Vega doesn't have any Infinity Fabric in it, but that isn't why Vega flopped. Mostly overvolting out of the box, and a poor reference cooler combined with the mining boom and RAM shortage limiting supply to the normal consumer.
It is however something that will help AMD make bigger and more powerful GPUs in the future without having to worry about yields. Die shrinking has the downside of making monolithic products (like Intel's Skylake-X CPUs) very difficult to manufacture. With smaller die sizes however, the yields are far greater, which means less money is lost per wafer and "more" can be made for "cheaper", and thus sold "cheaper" if AMD chooses so.
Last I heard Navi is getting the first GPU implementation of Infinity Fabric, and the design after that will be brand new instead of the umpteenth iteration on GCN.
When will we see the 24-pin disappear for a slimer more powerful cable ?
do you have to size up a FF helmet cause of your locks
I have had two Corsair AIO fail, the first I replaced, the second I sent back, back when the 4930k was used to heat a small apartment building, AIO were a must, is that still true today. The same 4930K is overclocked and stable using the Noctua NH-U14S, did I buy in to the HYPE?
Could you make a video comparing some of the more popular cases with the maximum size & number of fans rather than just "stock"? I feel like some cases only come with 1 fan stock and no one I know runs their case with only stock fans (most go in the trash).
As for APUs becoming "High End" I could see a MCM that has CPU+GPU+Memory modules under the IHS. This would be a way to create smaller, cheaper silicon modules and "glue" them together. I suspect they have already been attempting to use Infinity Fabric for a GPU module and this would be a drastic improvement at lowering the cost of consoles.
The question at about 21:00 makes think of this: instead of APUs, do you think socketing GPU chips on motherboards could become an option? Instead of using a PCIe slot for graphics cards, there could be another socket for GPU chips (makes me think of old math coprocessors.) The motherboards would need to change quite a bit to offer proper power delivery for this new component, and there would be extra memory slots besides RAM ones: the (HBM?) DIMMs :) The benefit of all that: upgrade paths would have quite a different granularity compared to the replacing entire graphics cards of today. You can choose your own cooler just for the GPU chip. You could theoretically increase how much memory your GPU has, without changing the GPU chip... The fact that both GPU and CPU chips would be parallel on the same board plane instead of perpendicular to each other could also possibly offer easier cooling arrangements
13:12 Because a CPU is a silicon core on a relatively simple circuit board with not much room for variability. The variable is the motherboard which provides the memory, voltage controllers, power delivery, etc. A GPU in its turn is a different type of relatively simple processor which requires its own "motherboard" with memory, power delivery, voltage controllers, etc. So the silicon in GPUs (and CPUs) is practically the same, what differs is the motherboard - Intel/AMD make the silicon product, partners make the motherboards, and it is valid for both GPUs and CPUs.
Any chance you guys can take a somewhat older case that has side panel fan mounts and test with a system built with a GPU with an open air cooler design, where side panel fans are set up as intake, and then again as exhaust?
maybe even mix up the testing wehre cpu is air cooled and also using AIO on cpu in the above mentioned tests.
It's a lot of testing and I get it if you don't have the time to do it, but it seems like it would be interesting content to see.
If you've already done a video on it, point it out to me I'd like to see it.
I've got the Antec 1100, but I don't have the disposable income or industry connections to test any of this out. It woudl be interesting to see just how much difference(s) it would make by having that extra airflow around the GPU.
What about taking the CPU and GPU and the area size of threadripper and have like a full gpu, apu cpu,?
I had the same idea, it would be interesting to see
I noticed the Santa Cruz shirt from awhile back and had a feeling you shredded some gnar
Is it normal that the HDD is the loudest part of my system ? I have 7 fans in my system running at around 700 to 800 rpm and you can still hear the HDD over them
John Smith
How old is your hdd?
Actually ARM CPUs are not manufactured by ARM, but rather licensed out to other Manufacturers. So the CPU "AIB" stuff kinda exists. Just not for x86 AFAIK
Bit of a noob question... what are some ARM CPUs? Are they mostly mobile chips or something?
Blessman Fred Yea, They have misleading specs however, as a proper comparison of ARM against x86 Would mean a computation per clock ratio of 3 of Arm : against 1 of x86
Wow that got unexpectedly complex... o:
Is GN going to look into the GPP controversy with Nvidia?
Hey Steve. I have a hopefully simple question that I would like you to answer in the next "Ask GN" episode. It is:
Where can I find what the IPC of a given CPU is?
As far as I know, neither AMD or Intel list the IPC numbers for any of their processors anywhere, so I was wondering if you would know of any place to find those numbers. Thanks!
Why not integrate hbm on a cpu die since its so much faster than ddr; use the hbm as a cache etc then use ddr as expandable memory; when needed/applicable.
The one question about apus went into some detail; but such a question too would be relevant to the question of "custom cpus"
As you said it is hard to cram gpu and cpu (apu) in the size of an apu but in a cpu without a gpu there should be plenty of room to integrate hbm?
Hey Steve, how do you think LM and Copper Heatsink will go in long run? I ask specifically for gaming laptops that have no IHS on CPU but do come with copper heatsinks. Consider there is no leakage of LM onto board as necessary precautions have been taken care of.
This is probably stupid but could you take a cpu that is not solder to the heat spreader and soldier it?
Hey Steve, I am in the market for a new phone and as funny as it might sound I was checking out new phones and noticed that they are shipped with different processors, different gpu's and so on. This potentially creates a new segment in the review community that hasn't emerged yet as much - testing the cpu, gpu of a mobile phone, if something I would love to see GN do a segment on those. My question is - is GN planning to look into mobile hardware since it is getting better and more powerful each iteration and there are a few different configurations already(Mali/Adreno/Bionic custom GPU), since you already have a wide and broad experience in testing? Thank you.
Great work, Steve!
A GPU is like a computer on its own. It has the processor chip (GPU). It has its own RAM. It is on a PCB. It has output ports. It has extra features.
A CPU is just the processor chip.
So, I suppose, it makes sense to have external companies assemblies the video card.
I was wondering about the other part of APUs. Obviously the graphics will never be able to compete with discrete graphics of the same generation but will the CPUs ever catch up? Will we ever get a 6 or 8-core Ryzen APU? They've all been in the lowish end 4-core segment for a really long time.
@Gamers Nexus I'm sleeving my own full-length custom cables for my pc but I'm stuck. Two questions for you: why is it that some (but not all) 24-pin ATX cables have 24 pins on the mobo but on the PSU side it's 10 + 18 pins? Also, why is it that certain ATX cables, like my 18awg EVGA cable set has 27 wires on the 24-pin connector (which means that 3 of the wires are doubled)?
I haven't seen a single video that clearly explains PSU cables - there's a lot of intricacies that often go overlooked, I'd love to hear your explanations - they always rock. :)
so i was wondering what thermals would be like on solid front panel cases especially ones with out power supply shroud with bottom fan mounts if you reversed the airflow and had front exhaust with bottom and rear intake could you test this out
If there was a ryzen threadripper with vega igpu, the quad channel memory may help the apu alot.
Funnily enough I had a similar problem with the physics engine in my game. I haven't yet figured out how to get around that...
Thankyou very much for answering my question! I have now watched this video on the cost of HBM2 vs GDDR5 and I see a potential benefit of HBM for end users. But in the implementation from AMD vs Nvidias GDDR5 implementation it appears that there is no end user benefit to HBM over GDDR5. Is that fair?
1050 ti cards are slowly but surely going down in price, which model do you think will follow suit? I’m thinking 1060
Hey Steve, I have a question about the CPU/GPU dies... why the chip is that smaller from the actual PCB that is underneath? A bigger die will not spread the heat way better than a small die with stuck of things one of top of the other? 🤔
idea, Threadripper APU, take the 2 "spacer" dies and put Graphics dies in those locations, basically take a GPU half of the APU die that is twice the size of the one in the current APUs so it's a full die size then crossfire them together through the infinity fabric or some crazy shit like that.
for real why dont 2 way sli cards come packaged together with a 240mm radiator AIO.
Why do laptop manufacturers use LPDDR3 on 8th gen cpu's instead of DDR4 ?
Has anyone had heat spreader fall off of Corsair Vengeance tall memory. Mine did when thermal tape let go. Machine wouldn't post until I removed them. May have shorted component near memory or on memory causing no post. Memory survived. Might make for a good topic for Ask GN.
Do you think we will ever see a glue architecture GPU to get around silicon size limitations?
Hey Steve, What do you think of BCLK overclocking on Ryzen and what benefits do you think it offers? Also do you BCLK overclock on any of your computers.
I think the only way an APU would be able to compete today is on a platform like the TR-4 socket, where the substrate is big enough to handle more GPU/HBM onboard RAM and such. How the power delivery would work and how to cool something that monstrous, I'm not sure about. It could be an interesting APU, but would it be cost effective and worth the trouble? I have my doubts.
What do you think of the battle between AMD and Intel with regards to the amount of cores? While both parties are adding more and more cores, little is being done on the (consumer) software side to actually make proper use of all of them. Especially beyond 8 threads there are very little programs that utilize all cores. I would personally rather see them focus on making the cores themselves faster and adding more features to them, than increasing the costs without actually having a big real-world performance boost.
I know that the frequency is nearing the limits of the semiconductors on the current transistor sizes, but I suspect there has to be quite some headroom left in the architecture and chip layout.
Question for AskGN
So a few months ago when i was installing a new motherboard, i accidentally bent my usb 3 header and now can't connect my usb 3 ports on my case. Is it a good idea to just leave it or bend it back using tweezers or something similar?
btw i have an asus prime z270 ar
I expected "Steve Shredding" to be a section where you just play some sweet guitar.
Hey Steve, could a pc be cooled with an air sealed case with strong airflow through a water cooling radiator, with the heat being dumped outside of the case? (I.e. larger radiator outside, or well water?
#AskGN in regards to the High end APU question. What if AMD used a TR4 sized substrate and socket for an APU? What do you think they could realistically fit on there?
Hey @gamersnexus I will be performing my first custom water cooled pc soon, and was wondering about bacteria growth. I've heard about putting in chemicals to prevent growth, but would passing the clear tube through a strong uv light tunnel kill any growth effectively? Thanks for the input
Difference between gddr5 vs ddr4? yes we know ones for graphics and ones for system, but what makes them good at those things? I know their manufacturing is similar, but how do they differ? And with the release of ddr6 (ddr5? Cant remember) is it possible that we are working towards a "jack of all" memory that could handle both?
gddr5 much higher bandwith and much higher latency than ddr4, also a much, much, much higher power consumption than ddr4
I'm sure this has been asked before, but do you think sometime in the future graphics cards will become customizable, Similar to how cpu's are? You would buy your own gpu chip, board, memory, cooler, etc. Is there a reason why this is not already a thing?
Thanks.
The physics solution for the sake of fluidity is calculated every frame cycle. Physics solvers are designed to work with 1m, 1s scales. Increasing or decreasing that resolution results in pronounced floating point overflow and loss of precision. So, technically that frame rate limitation is computer architecture limitation. Bullet, for example, has a scaling parameter which keeps dimensions at optimal scales, however, changing that at runtime is a NO-NO.
18:14 Will APU's Go High End?
With the near release of Intel Kaby Lake G with Radeon Graphics (not really an APU), and the rumors of an AMD Fenghaung APU with 28CU's + HBM2, I would not consider it high end for a desktop, but for laptops, definitely in the upper half.
Do you think we will see this kind of processor soon from AMD? As HBM would really help these APU's with there biggest weakess of memory bandwidth.
And also, do you think we will see these Fenghaung APU on AM4 or TR4? AMD has teased 32 core "Exascale Heterogeneous Processor" before.
hey steve with the r3 2200g or even the 2400g how big of an air cooler would you need to effectively cool it with out the fan while overclocked on both cpu and gpu while under full load im just looking for basically total silence and just having the one rear fan as an intake and nothing else probably use the dark base 600
Hey Steve, do you think that the Ryzen Threadripper "2" being released in the future will still use DDR4? If I was investing into a TR-4 system now but didn't really want to spend a lot on a fancy motherboard (although, from what I have read TR-4 will be used for both Threadripper 1 and Threadripper 2) and just get a decent Threadripper now. Switching out the MB in the future to take advantage of whatever new features Threadripper 2 has. At least then I can just move the DDR4 RAM, Graphics Card, Power Supply and Case from the old one to the new one, buying new MB and CPU only.
What if they made Tr4 sized APU's?
Reiji Oenothera it would be the price of a thread ripper lols kinda kills to point of Apus being budget
I don't agree. The point of going bigger package size would be to fit a more robust GPU in there. I'm not saying make one for Tr4 platform, but what if the package were that size. Beyond that, nobody said it would have to be budget at that point. Even then, if they could fit significant performance in the package, say an r7 and vega 56 class for ~$429, I would consider that budget anyway.
Holy shit did you put progress bars on the currently discussed topic on the list? I wish everyone did that now.
I caught a fan on fire once. It was a 3.5-3.6v (or somewhere in that range) and I hooked up 5v power to it and the cable literally went up just like a fuse going to a bottle rocket. Scary as hell and sucked since that fan couldn't be replaced.
What about an APU with HBM? Is sort of "on chip memory" more likely to happen with cpus first or do they really not need say 2GB of "cache/ultrafast sysem memory?"
Expensive and then you're also dealing with memory supply issues.
The CPU question at 13:12 technically does happen with CPU, or rather the chipsets on the motherboard. The 'AIB' partners in this case would be the Mobo manufacturers.
I still want to see you benchmark that upgrade CPU for S&G
When do you see developers utilizing more than four cores for games? Assuming that CPUs keep going the way they're going. Any predictions on what they'll be able to do with the new resources available?
#askgn
Hi! Why AMD or Intel cant make processors without IHS like in GPU's? I mean not all of them just like "special edition" or part of HEDT just to have choice. I know that IHS shields the die so it wont break but its not a big problem looking at how its going with graphic cards that have big non referent cooling.
Will there be any Kaby Lake G mini itx boards?
Also can we count mobo's with soldered processors (ie. in laptops or NUC's) as CPU AIB?
Steve do you think it's possible to desolder a cpu from a laptop with a higher end i5, with a better i7 model? I've seen some pretty awesome videos of how to desolder, and reball the cpu points on the pcb and reinstall the same cpu that had stopped working and it fixed it. I think it would be difficult but maybe possible? Would you ever attempt to try it?
Can you do some t3sting and comparisons for replacing the VGA cooler with something like the Rajintek Morpheus 2? On Cards like Galax 1080ti HOF or Evga SC2 1080ti cooler replaments. How does card perform n vrms cooling etc. Are they worth it