Gadgetory

this is a preview of what we're looking at today it's a composite overview of Vega GPU mounting pressure and here's a look at why we're testing it when Vega launched at least two primary modes of GPU packaging were found across the GPUs with either raised up molding encasing the silicon or the absence thereof Tom's Hardware initially reported that HBM potentially ran 40 micrometers lower than the GPU and the initial proposal by Tom's was that AIB partners may have more difficulty designing for the multiple packaging types today we're testing to see the mounting pressure and thermal impact from and these various Vega GPU packages before that this video is brought to you by thermal Grizzly makers of the conductor hot liquid metal that we recently used to drop 20 degrees off of our coffee leak temperatures thermal grizzly also makes traditional thermal compounds we use on top of the IHS like cryo not and hydronaut pastes learn more at the link below just for an explanation of what we're doing and these Vega GPUs chip and at least two types of packaging if not three one of them uses an epoxy resin that raises to fully encompass the GPU and HBM to flattening the height and protecting the inner poser in the process we found this mounting method on three of our four Vega 64 referenced cards two branded by gigabyte and on loan from a generous reader and one from Andy as a press sample the second method is resin --less assembly which leaves the inner poser less protected and sort of recessed from the GPU and HBM this also means a more topographical contact area we found this resin 'less method on our Vega 64 Strix card from Asus and on our Vega 56 reference card from the initial press batch our Vega frontier Edition card used the epoxy resin method of the first one as did the three Vega 64 reference cards in speaking with board partners it seems the resin list versus epoxy packages are not chosen by the partners and that is basically dependent on when the batch of chips went out thus far there is no hard data to suggest that either package type is superior to the other they originate from different factories and so they are inconsistent in assembly but that's not necessarily bad we'll be testing that here our first order of business is to determine the mounting pressure from the various coolers with a chemically reactive contact paper so this paper turns sort of an off red when it is when pressure is applied to it from either side that would be from the cooler and from obviously the GPU side and this reacts to show us where there is contact with the vapor chamber now where this doesn't really apply is the fact that in real life you have thermal paste that sits between the vapor chamber or the cooler in the case of the asus strix card and that is what would bridge the contacts for an imperfect service between the GPU and the cold plate which be the imperfect part of the service so that's something that's not accounted for in these pressure tests but we'll look at that a little bit later right now we're just focused on the variance in pressure contacting the GPU and potential differences between the different packaging types for the silicon that's what sparked all of this and then of course the sort of extra bonus round of that is the variants in the vapor chambers themselves and whether these various cylinders within the chamber will impact the contact in a meaningful way we're testing a total of six devices some of these tests will include swapping coolers between GPUs which will help determine if limited mounting contact is more results of the vapor chamber cooler or of the package assembly again most of these are done without thermal paste so keep that in mind but we do have one that was conducted with their own paste getting right into it we'll start with our vega 56 GPU as that provides the most varied results to quickly understand how this all works with its stock reference cooler that came on our sample our vega 56 GPU shows minimal contact with the vapor chamber for the HB m2 and it's also got spotty contact for the GPU proper we can highlight where the HBM would be Illustrated if it had fuller contact but on this particular implementation we're not seeing the HBM really show through the GPU itself has some contact to the chamber but cylindrical shapes appear in each of the first two tests this isn't a mistake there are actually cylindrical columns within vapor chambers preventing the chambers from collapsing on themselves this is normal for all vapor chambers the fact that we see them here means that there's a greater contact pressure in the area from those columns and if there's enough interest in the comments we may dissect one of these coolers in a few days this contact patch overall is theoretically suboptimal but remember that thermal compound exists exactly for this purpose hypothetically the vapor chamber cooler should outweigh the negatives of more limited contact area than a flat cold plate with copper heat pipes and thermal paste again is meant to fill those gaps it's just a matter of whether it does its job and we'll see soon hpm also has a low heat flux so it can better get away with less contact than something like a GPU could here's a look at our hybrid mod melting from way back when Vega 56 first got here we first thought our hybrid model would have worse contact on the vapor chamber but repeat tests and improved better contact with the ASA tech TLC plate that's on the flow 360 this is because it's flat to slightly convex with a z' attack designs and so we see greater contact with the hpm remember this GPU is one of the recessed package GPUs lacking any epoxy resin around the borders the GPU coverage is also significantly better with a Vega 56 hybrid mod than with the reference cooler if we were to superimpose them for instance moving on the next GPUs are our three Vega 64 reference cards two of which were purchased by a reader and are on loan all three units use the epoxy resin packaging and so the footprint up here is much wider than that of the previous silicon packages this is because our contact paper is reacting to the Pennsy and pressure applied by both the raised epoxy and the silicon and then the heat sink on the other side and the two gigabyte branded reference cards have significantly better HBM coverage than our previous bake of 56 tests and have overall better GPU coverage you'll notice however that the earlier sample Vegas 64 reference card that we were sent early in the launch cycle has poorer contacts when compared to the two gigabyte units which come from the same batch as one another to test why this is we planted our Vega fifty-six vapor chamber atop the best Vega 64 in the test labeled as v60 4-1 and found the following it's got much worse coverage despite being tested on the exact same package this indicates a level Natisha with the Vega 56 vapor chamber in our model not all of them necessarily but that there's a quality-control difference between the vapor chamber plates this also carries over to our Vega 64 press sample of the card the earlier one and for further testing we took the best Vega 64 cooler that be our V 64 - one again and stuck it to the Vega 56 card we now see significantly better coverage rivaling that of our hybrid mod the coverage is much better on the GPU but still lacks for the HBM we think this may have something to do with the lower HBM tube package height as initially reported by Tom's hardware but we are not yet fully positive the next test is our Vega 64 Strix model from Asus in this one we see the best GPU coverage we've yet spotted with still lacking HBM coverage the Strix coldplay is nearly perfectly flat as it's smooth and nickel-plated without the supporting rods of a vapor chamber to get in the way we think that this correlates with the recessed package design as our Strix 64 card lacks the epoxy resin of course this is a sample size of just a few cards so we have to implore that correlation of course as an equal causation but it is a potential finding what is notable though is that the cylindrical divots do not appear with the flat cold blade of the Strix card which is attributable to its cooler design that does not inherently mean a superior cooler has vapor chambers make up for it in other ways and of course the Strix has different fans but it is another potential point of data for when the Strix undergoes review testing the final test was conducted on our Vega front erudition card which we purchased in the very first round of vega products this GPU uses an epoxy resin package just like the three Vega 64 reference cards unfortunately it also has similarly poor contact to our early sample vega 64 and early sample vega 56 again a correlation doesn't equal causation but a point worth exploring would be whether the models like the gigabyte Vegas 64 units underwent improved quality control for cooler contact pressure we can't be certain without more units to test if you'd like to send them for loan though tweet at gamers and access to get in touch for now we can only note that our Vega frontier Edition card had it more limited contact we tried applying thermal compound underneath the contact paper just to see if the coverage area appeared improved but coverage although improves a bit didn't really look like it worked with this test it's just not conducive to working with the pressure paper so we'll discard that for now let's pull the most important results and then look at their thermal behavior this is the best coverage Vega 64 vapor chamber versus the worst coverage chamber from our Vega 56 card we've kept the fan the base plates and the face plate and I've swapped only the vapor chambers the GPU is also the same we have controlled for ambient and apply to Delta modifier where necessary and also ran the test multiple times with power virus burn ins and lower speeds at about 4700 rpm the Vega 64 vapor chamber which had fuller coverage keeps us at around 76 point four degrees on the GPU ninety-five point two on the GPU hotspot and eighty-two point three on the HBM temperature reading swapping the vega 56 chamber onto the card we end up with a GPU temperature that is about one degree lower more or less within variants along with hotspot temperatures that are 0.7 degrees different with invariant a hotspot temperature again is not appreciably different but the HBM temperature is it's a full three degrees higher with the Vega 56 chamber which again had worse contact and is outside of variance and is repeatable bringing the comparison shop back up on the screen you can see why there's such a difference on the HBM temperature fortunately an extra three degrees won't hurt anyone there's definitely quality control variants here but we're not running significantly hotter than the cooler already operates just for parity here's a frequency chart to demonstrate that the tests operated at the same frequencies we can also show GPU only power which plots the cards as consuming equal amounts of power for each test so the takeaway here there's a few of them one is that yes there are different types of packages and they do impact how the cooler contacts the silicon but that's primarily because there's fuller contact with the raised epoxy resin packages because they contact more of the cooler so that's potentially somewhat beneficial to the the ability to dissipate that heat as we saw in some testing but it's not a huge difference ultimately you're talking maybe a couple degrees at most and that's just not appreciable because the Vega cooler is already bad enough that it's just it's not gonna make a difference you're already in the 80s for HBM so going from 81 to 83 at 4700 rpm and an ambient of about 20 degrees is just not important what is noteworthy is that of course buying the AV partner cards will not guarantee anything we spoke to them about this and the abbey partners as far as everyone we've spoken to is aware have no ability to say to Andi we want the resin list package or we want the resin package so it's kind of luck of the draw based on when the batch was run and to whom it was sold as for other items of note thermal paste basically makes up for the poor contact of something like this this is the Vega 56 unit which had among the worst contact out of all of our tested units and it seems to be more indicative of a quality control issue with the vapor chamber more so than the GP packaging that said the resin list packages did consistently show worse HBM coverage towards the top part of the hbm's in the middle of the entire package than with the resin packages so that's another item of note and basically corroborates what Tom's Hardware was saying closer to launch now does it matter as far as thrown performance it doesn't appear like it really does but it also that could potentially depend a bit on the cooler type but for these not really the AV partners that have the flat cold plates like this one should get better contact than the vapor chamber because it doesn't have the cylindrical bars in it which is a normal thing to have in a vapor chamber however the vapor chamber is able to partially make up for the contact area by just being a vapor chamber then again the blower reference cards suck on NVIDIA and Andy devices and that's for reasons other than just the cold blade and the heatsink so yeah that's the data kind of cool stuff not hugely impactful to how you use the device or how it performs necessarily what is noteworthy one more time though is that if you open these up and reseal them make sure you really tighten it down to spec and make sure there's thermal paste in there because it's gonna matter but that's all for now as always you can support this type of research by going to patreon.com/scishow solves that directly or store that gamers nexus dotnet to pick up a shirt like this one or one of our new stickers subscribe for more I'll see you all next time