Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

Ask GN 74: NVIDIA GPP, HBM Saving Memory Prices?

2018-03-26
everyone welcome back to another episode of ask GN as always leave your questions in the comment section below for next week's episode after this one we'll have a separate extended ask GN for patreon backers you can get access to it by going to patreon.com/scishow and Rose Nexus and speaking of this may be the last video that we shoot without a shelf behind me but we'll see how that goes if you're not familiar with it that's also on the patreon behind-the-scenes stuff so a couple interesting questions for this one we are looping back on the GPB topic at least briefly I don't have a ton of updates I have some stuff I can talk about though before that this video is brought to you by Thermaltake and the view 71 enclosure the view 71 is a full tower case that's capable of fitting three video cards and most configurations it's also one of the better cooling cases in our recent case testing bench lineup the view 71 has hinged a tempered glass doors on either side that make it easy to open and show off and it comes with at least one rain fan though you can get the RGB version if you prefer learn more at the link in the description below this first question is something we brought up in a news video briefly and we'll talk about again now this comes from John rty double-oh-seven who says hey Steve do you have any opinions on the articles released by hard OCP pc / and forbes etc on the nvidia geforce partner program it seems like an interesting topic and could greatly affect the consumer market as a whole I have a whole separate sheet on this one we don't often prepare entire extra sheets for this because this required some research so let's try and get into this this is a very difficult topic to talk about and it's primarily because there's really two sources of information there's Nvidia's blog and there is hard OCPs original article those are our primary source of information we went out and tried to speak with board partners s eyes analysts anyone we could find who would well who would say anything or not say anything because both are really equally useful right now and we talked about this in the news video where at the time we had no additional information from our contacts we were not able to get any information that validated or invalidated the story people generally I didn't know or wouldn't talk about it and sometimes it's hard to distinguish between the two I've spent the last week talking to a few more people so here's what we're gonna do first of all in complete effort of fairness and for benefit of the doubt and because there's really only one source of information I did reach out to NVIDIA I did offer them the opportunity to talk with me either directly about the hard OCP story which reasonably they did not want to comment on or talk about the GPP in general thus far we have not received a response to our inquiries so I would like NVIDIA to consider speaking with us in more detail about I want to say their stance on this but I don't even need their stance on the discussion of GPP as it's been presented currently I want their stance on GPP as a program and how does it benefit their partners and the consumers that's what I would like to talk with them about right now but I don't have that today we did want to present that opportunity to them no-count presently so what I do have is a list of people as percentages with whom I've spoken about the program or well we know it exists but we don't know to what extent the hard OCP story is accurate right now the only person who really knows that really is Nvidia and to some extent Kyle as Kyle from heart OCP notes that he has some sort of documentation which is not public so let's go through the numbers I have this can be very brief and very direct 35% of the people we spoke with genuinely had no idea the legitimacy of the heart ocv claims or the illegitimate II the the kind of answer I got was that this is all discussed way above their pay grade and this is part of the problem this is why it's hard for me to dig into this like a lot of people want because our contacts are often marketing PR maybe some technical marketing and very limited engineering resources none of those people would be involved in these discussions for the most part now some of them may have overheard things or may speak with their higher-ups so that would be kind of the only angle you have and that's still second-hand information which is why it's very difficult to talk about this story because I can give Nvidia a chance to talk I can give board partners a chance to talk if no one talks to me I have nothing to add so that's the challenge right now let's go through the rest of them so 35% genuinely no idea and I believe them what was going on how accurate the story is or how inaccurate it is is discussed above their pay grade 15 percent unsure and 15 percent more think that this is a non-story and an AMD PR opportunity which is referencing the the fact that it was an AMD sort of vended story 20 percent board partners would not comment at all specifically told me they would not comment and typically these are people who do talk to me even if it's off record this time however they would not do so 15 percent more were upset at the terms of the GPP and indicated arm-twisting by nvidia and and then nvidia would not comment so what we've got here is basically a whole huge percentage of no information no comment I don't know and then we have is it equal percentages here equal percentages equal counts of people who have told me that they their team was upset at the terms of GPP or who have told me that they think this is an AMD PR opportunity so basically once you rule out all the people who said I don't know I won't talk to you or some variant of those things you know that's with 50% of people who answered me saying we don't like gvv and 50% who said this seems like a non-issue or it seems like an AMD PR op I don't know where to go with that information so I've been trying guys I've been to talk to more people it's very difficult to get anything here and that doesn't necessarily indicate that Nvidia is doing something wrong it does definitely indicate that my contacts aren't high enough up the chain to give me a proper answer like we all want I'm not saying that Nvidia is definitely in the clear I'm not saying that they are in the wrong I'm saying that we have no meaningful way to make a determination at this point and it's not my story Kyle's apparently got the documents and we don't so that's that's really pretty much as far as I can go with it now saying all of that let's briefly talk about two things here one of them is why would and Vidia create this program and think that it's good so we're going to take their position for that the other one is why would people or partners see this program as bad so that's the opposite side I'm gonna present both of those to my present understanding and then we'll move on until I get more information let's start with Nvidia's I gave Nvidia the opportunity to talk to me about it we were pointed to the blog post I read the blog post and it more or less outlines why they think this is a good thing if you haven't read it in effort of trying to read every one side it is worth reading so my interpretation of the blog post and of Kyle's information that he's posted assuming accuracy of both of those we're assuming Nvidia's not lying and their blog post we're assuming Kyle's not lying in his post so taking both of those two only sources of information I have my interpretation is that Nvidia in an honest effort is trying to preserve their marketing dollars that they give to vendors I want to list a bunch of companies but I also don't let's just say EVGA cuz they only make Nvidia so we're not going to hurt any feelings so so let's say EVGA Nvidia wants to take their money that they give EVGA and their marketing Development Fund and they want that money to be used to advance an Nvidia product under a known presumably EVGA brand and that's so that if EVGA also sells AMD products which I guess EVGA rearranges conveniently eega so EVGA has got a vga product and nvidia wants to make sure that the EVGA v EGA is not sold under the same branding as the EVGA nvidia product they don't want a let's call it an FTW that is for both if they are providing money to advance that FTW product line I think that's that's the gist of the primary argument that Kyle has posted and and that would be how I would interpret it if giving the benefit of the doubt to Nvidia so from that interpretation I would say it's reasonable to not want your marketing money spent on a brand which carries a competitor's product I get that however there are also concerns so some of those might be as presented by Kyle some of them would be AMD being sort of backed out because this comes down to whether we're talking about established brands or not this is a big question I had friend Vidya but let's talk about that let's go down this path right now so with Nvidia's agreement if they are saying to vendors you have to create an nvidia brand or or establish an existing brand as in video only in order to receive these benefits because we don't want to give you money to develop something that you are also selling with your competitors products so then the question is let's say there's a company that makes a a brand called dog the democracy of gamers the company that makes dog do they have to take that established well-respected democracy of gamers brand and for lack of a real word exclusive eyes it under Nvidia products or can they say you know what Nvidia we like this GBP idea however we spend a lot of time developing the dog brand and we would like to be able to sell both AMD and NVIDIA under that brand how about this what if we create a dog aires brand that is only nvidia products so that's my question is that okay or or does the entire dog brand have to be Nvidia or can the company say dog is going to be vendor agnostic and we're gonna create a new one to get me benefits of GPP and it's not gonna sell I am the product so is it then ok for the vendor to produce a new brand that is exclusive 10 video or does it have to be an established one these are all the questions I have absolutely no answers to them so the reason I'm saying all this is because there's a lot we're trying to consider here to give everyone a fair shake but no one's talking to me so and that's not necessarily because they don't want to it's not necessarily because they're trying to hide something but as I've stated a lot of them genuinely know nothing PR people would never be clued in to this kind of discussion in general or the half that spoke to me don't like it the half the other half think it's a non-issue you can see why I'm kind of exasperated at this whole topic because I have nowhere to go with it well look into it further I'll let you know if there are updates my final word on it is I would absolutely be concerned if this program exists in the fashion which it has been presented thus far by heart OCP and I would very much like Nvidia to provide more of the transparency that they have promised by just talking to us or other media about just the upsides what are what are you trying to do with this program not any gotcha questions not trying to corner anyone I just I just want something it's something to go out so yeah that's that's where it is that's where we'll leave it a second question might as well just make that its own video second question is from strand Easter who said hi Steve did Coolermaster respond to you in any way after you release your CMH 500p mesh review and no no they did not : I stir didn't say a word to us that was easy next question as rock by the way that case was if I didn't make it clear was actually good I liked what they did with it but uh we didn't praise the first one so screw us I guess next question asrock unpredictable reference card Steve with the current memory crisis will using HBM type memory become an alternative solution not really and the very easy answer to it is because the same people who make G DDR and DRAM which has problems in terms of availability and price they are the same people who make HB m so HB m is primarily samsung right now and microns been getting into it so you have the same vendors they run on the same out of the same fab so you can have the same problems HP M is also very expensive anyway a couple times more expensive than gddr5 for example so because of that let's even if you pretend hbm's not going up in price other types of memory are still cheaper and HP I'm also requires significant redesigns of architecture to integrate so no no it wouldn't help it wouldn't help unfortunately next question is from Rama Kiara a week ago he said HBM is stupid 1080 beats Vega without HBM so AMD you had no visible using HBM and making the card more expensive we actually have a great video on this it's I think you can just search for HP m verse gddr5 on the channel or how much does HBM cost where we talk about why AMD chose HP m so it definitely makes the card more expensive you are correct I disagree with the simple answer that HB m is stupid HP m and video uses it too there are good reasons to use HP m they use it in their volta products right now so an AMD use it in some of their high-end products HBM has significantly lower power consumption than g ddr memory which is very important because you have a power budget for those cards and it also has advantages in bandwidth and in architectures that can leverage the bandwidth like the Volta products it's definitely not stupid because in those instances using gddr5 would limit you for AMD their products there although they were somewhat backed into a corner with their choices when developing things like Fiji and it did not pan out well betting on HBM didn't necessarily work for them in spite of all that they also sort of had to use it because these products became so power hungry that AMD had really no option we talked about it in the GDR 5 versus HP m2 video basically the extra couple of watts here and there for every piece of gddr5 memory on the card will push the power budget to ranges that are unreasonable also AMD gains a lot from HP am i overclocking they are bandwidth bound in some cases for their memory and that neccessity it's a higher bandwidth memory choice so no HP M is not stupid it does make the card more expensive I'll give you that and it does need to be used in obviously scenarios that make sense Warren said we all know it's bad idea to run a CPU with no thermal paste between it and the heatsink but how dangerous is it really doesn't the CPU forcefully turn itself off to prevent damage anyway if it gets too hot yes so if you kind of it'll thermal shutdown in theory if you're your t.j.maxx sent when when your motherboard senses t.j.maxx depend on what you've configured or what's configured by default you'll either shut down the system or throttle in either case it's protecting your CPU from further harm as long as you can identify that the issue is thermals and a thermal shutdown and you fix it before before routinely running into that scenario you'll be fine and even if you run into it a few times it's probably still fine because things like passengers take some damage they aged faster at 105 degrees Celsius for example but you're still talking against thousands of hours as long as t.j.maxx is tripping a shutdown or a hard throttle that lowers temperature reasonably then you'll be ok and if it doesn't then there's a problem definitely if you're if you're forcing it to run against t.j.maxx without any kind of protection yes there's a problem Sean Lyons says how important is the hot spot temperature on GPUs as reported by monitoring software such as hardware info for example on my Vegas 64 liquid edition the GPU hot spot temperature will max at 105 average 57 during fire strike and the GPU thermal diode only hits 49 I don't know if it's been answered since Sun last time we looked at Vega there was no concrete answer on where the hot spot temperature is however I did sort of pick up a a an idea for where the hot spot temperature is and that seems to be between the HBM and the GPU so I came up with this basically by testing one of those I think was the power color card with very poor thermal contact on the HBM and the temperature was significantly higher for the GPU hot spot which indicated to me that it was not in the middle of the GPU diet dye because that had full contact to the cooler and the HBM did not have any contact at all in this test to the cooler so my assumption was that the hot spot is somewhere between like the dye and the two hpm dyes somewhere triangulated in there that's that was my guess maybe someone's answered it concretely by now sike says why don't laptop manufacturers like Apple and Microsoft use liquid metal their products shouldn't be a problem if applied properly and would help massively with cooling as a customer you can't open their laptops anyway so there's no danger of someone doing something that's stupid with lucre metal later there's always danger of someone doing something stupid and you can still open the products liquid metal should not be used in that type of thing for a number of reasons one of them being that of its direct eye contact as opposed to an IHS which is off in the case you run into potential corrosion or pitting issues you run into issues of not tight enough app installation of the cooler where you could have liquid metal spillage over the sides shorting things and killing them parts tend to loosen a bit with age laptops move around a lot they go through a lot of shock and vibration it's possible that things lose now for some liquid metal gets out short something you'd be better off soldering if you wanted to improve your thermals that much but yeah there's there's a lot of good reasons to not do it Zeta our CLC is becoming common enough for you to include them and your thermal case testing alongside the air cooler the problem here is dismounting that we've done some of this as one-off testing it's best to do it as one-off testing because then we can test the concept demonstrate how the CLC performs against the air cooler under different conditions generally with different mounting orientations and locations but the problem is if we're standardizing our case test bench introducing a lot of mounts and dismounts of the cooler starts to cause concern for long-term use and for consistency of testing especially for all that different thermal paste application so that would be our main concern I'd do it as one-off couger 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 mod rights mod mat I believe they composed of the same materials or maybe they don't you could explain this as well they actually don't they're very different what is the best way to clean them thanks okay so here is one of the mod right mats and we did competitive analysis as we ramped into producing our own I'm biased uh we make the other mod mat but I do obviously think that our material is superior we did have the advantage of getting to look at all the others and figure out how to make a superior material for ours so this one is more of a you can see it's like a rubber ours is is rubber it's rubberized but this is like more like a tire rubber so it does have a different like feel to it leave some at least this one leaves some residue on my hands when I interact with it but cleaning them is about the same so for cleaning you could if you're gonna be like hyper-paranoid and you want to do it like a cleanroom wood which is not necessary at all we just this is just ESD service and mat cleaner it's called that comes with our industrial stuff we buy for business great testing and it's it's what is it even composed of it's just a cleaner it's really trivial you could honestly you could use rubbing alcohol to clean hours but you spray it on and then you take like a rag but paper towels we find don't really work great for this for theirs or for hours because the paper towel will kind of disintegrate as you're cleaning it because the surface is a bit gritty and for good reason it's so that things don't move around and because the Matt weave is I mean it's just great here by Nature but it's just he's a normal drag so that answers your question how to clean it the material though I did want to point out just cuz you said you think the material is the same it's it's way different ours is it it's it's a rubberized surface and underside and it's made in a factory that makes products for clean rooms and this one is more of a pure rubber so in terms of how it feels it's a lot of different in terms of how the ink applies to them both it's a lot different I'm not saying either is better or worse in terms of the ink application I'm just saying materials react and respond a lot differently to the processes we put them through so yeah that's that answers your question I think anyway clearly biased on that but I'm also not afraid to say that I'm confident ours is a lot better next question over the last one I think you should really diversify from random tech tips really diversify the Gion portfolio into hair products I think you'd be successful well let me announce our new gamers Nexus anti-static hairspray you can pick it up on the store and we will sell used versions if you want them for actually more because I touched it and so the value went up so that's it for this one thank you for watching go to patreon.com/scishow and access to grab the extra version of this and otherwise leave questions in the comment section below for next time I'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.