Gadgetory

it feels like forever ago that we got the asus strix vega card but we've been unable to review it until a time at which asus and AMD come to a conclusion of vio is functioning properly with drivers we then got the power color card in the red devil and had some similar issues with it prior to the most recent driver update 11.4 and what we're doing today is looking at these three cards to see if it's finally time to review them because it's getting pretty close to consumer availability and that means that there needs to be reviews online so all we're doing today is checking the frequency response using different driver revisions on these three different cards and different V BIOS revisions in some cases to determine if AMD and their board partners have resolved the issues of drivers and the clocks not quite working properly before we get into that this content is brought to you by the Thermaltake Flow RGB closed-loop liquid cooler which is a three hundred sixty millimeter radiator plus three one twenty fans that are RGB illuminated if then we'll take it ring fans at that this is a 4.5 done a detect pump which is one of the faster pumps you can learn more at the link in the description below so let's establish the basics here it's been pretty exhausting going back and forth with all the board partners and AMD trying to figure out when can we review these cards we've been waiting because here's the thing when we get the product in we've got a few options and one is if we're bar good obviously there's that's it end of story if we're not embargoed it becomes a question of when does it make sense to review it for the consumer if I reviewed this thing like almost two months ago and we got it it would be completely irrelevant information by the time it ships because it hasn't shipped yet but it will soon should by end of year and obviously that information would be so old that's completely irrelevant so we wait and obviously that gives the vendors time to fix problems or to improve performance if there are no problems to fix so that's what we've done now because it's getting time for these to launch it's five times review them so people actually know what they're getting so we're looking back at this one again and we're looking at the red devil verse is retesting the Vegas 64 reference card the big things here to look at are the frequencies so the frequency reporting in driver version 17 dot 11.3 was broken to a point of actually not just reporting but Andy even lists this as a problem as a known issue on their website so it's real issue and it's known by Andy so they've fixed some of that in 17th out 11.4 but not all it the rationally fakes an adrenaline or whatever the next major driver update is we've heard rumors of December from people who aren't at AMD so secondhand information but that'll fix most of it can't wait forever though with cars are shipping soon so 17.10.2011 teen dot 11 X so that was the conundrum now we're testing for frequency so let's let's just establish the frequency first and see if any of them actually clock properly the red devil and the asu's card should theoretically clock a bit higher than the reference card the pen which be bios is used so we'll look at that first with just firestrike for now this is a frequency chart of the reference rx Vega 64 card from AMD using driver version 17 dot 11.4 and the first bios which position with stock settings the spikes will be uniform across all devices because the test is recorded from an automated suite of all 3d marked fire strike an x by benchmarks that we scripted here's the asus strix card with the latest speed bios revision from a few weeks ago the card is reporting frequencies adjust below the reference card about 20 megahertz deficit on average but the question remains whether that's a reporting bug or even accurate reporting or if it's actually performance so we'll have to look at performance for that which we'll do momentarily and calls back to our initial vega review where people thought they were overclocking for example to 1900 megahertz which wasn't happening but the frequency would report that way so there is a clear history of incorrect frequency reporting on Andy drivers right now which means that we don't actually know if it's really slower or if it's just reported incorrectly you will have to test it next this next line is for the power color red devil card when using a pre-production v bios switch set number three to the far right switch position and this is Justus he behaves differently from production the frequency jumps up to about 80 to 100 megahertz higher than reference at times and it's consistent in its additional performance that's more from removing the power limitation than actual pre overclocks this next line is the middle switch for the power color card which they define as a balanced mode the clocks are reporting a bit higher frequency than reference at times but are mostly tied we tend to be equal or about ten megahertz higher finally here's the silent mode for the power color line this one is nearly the same as balanced mode though maybe one to five megahertz lower the differences in clock here are with invariance and error at this point as the power color spec actually defines the clock as the same for all three BIOS revisions if you're asking then why the higher power target one is clocking higher it's simply because it has a higher power budget not because they've actually pre overclocked at higher but that's just how the boost mode works the same frequency target is applied to balanced silence and OC and that sixteen thirty megahertz core with 945 megahertz HBM - the only real differences are the fan targets and the power budget this table shows all of the power color switch differences highlighting a 40 watt increase to tgp with OC mode and varying temperature targets that dictate the fan curb and here's how that relates to scoring in game this is sniper leap forward direct x12 nicing compute 4k resolution in high settings this game is one of the most sensitive to any changes on anthe hardware so should show the differences if they exist the reference at Vega 64 card with 17.11.2013 and 51 with older drivers 1781 the Vega 64 card got about 58 FPS average so we've seen some uplift on 17 at 11.40 engine all for the ACS card the somebody briefly recommended testing on 17 10.1 this is prior to the launch of 11.4 so we did that on 17 out 10.1 we were seeing a measurably better performance than on 17 not 11.3 and this was consistent and pretty much all games that we tested the performance is roughly equal to 17 not 11.4 so whatever was broken in dot 3 seems to have been resolved and that 3 did have more known clock reporting bugs so that's part of that though 79 11.4 has some of its own problem still they're just not as severe this would explain part of the performance disparity regardless of the minor differences here the asus card is not able to exit margin of error for differences between reference and we typically do see differences for AIB partner cards that are greater than the tolerance for error the power colored card for instance plots 61 to 64 FPS here depend on which meet BIOS were using and for the red devil we are seeing an actual difference with the higher TDP on switch number 3 we're measuring a gain of about six point nine percent on the red devil over the reference 64 on 17 to 11 dot for you didn't switch number to balance mode we're at 3.2 percent so it's outside of error but not by much well enough but it's definitely working that's the point and that's a start the Strix card seems somewhat stuck but we can't confidently state why that is if it's expected or if it's an AMD a sign-off issue of some kind that remains to be seen so to recap in some of the tests depending on what they were the asus card was originally doing better on 1710 that one then dot 11.3 the version in dot 11.4 fixed some of these problems and the major difference here is that power color is actually plotting more change so we're seeing up to 7 percent performance uplift with the higher TGV switch that's actually more wattage 40 more watts available to the GPU which is why you see that change and then other than this it seems like power colors is pretty much working on 17 11.4 there are more issues on dot 3 then now so I think we're ready to test this card and review properly you can look for that soon the asu's card still has an issue so at this point we're waiting for a soos to ship the card make it available for pre-order or something like that at which point consumers can get it so we're gonna review it or for ASIS - hopefully before launch say we figured out the problem it was this handshake with our card and and these drivers or whatever we fixed it here's a better version so yeah that's what we're kind of waiting on right now let's look at fire strike scores briefly - just kind of reinforce what we've seen thus far fire strikes next for fire strike ultra the power color Vega 64 with the OC switch is hitting 56 21 points averaged over multiple passes this drakes falls down to below reference levels of performance when using drive versions 17 11.4 and dot three especially and this is repeatable over multiple passes the maximum performance Delta between the lowest score and the highest-scoring single card is 6.5 percent that's not huge and we're targeting single-digit FPS changes but the drivers are clearly interacting with the Strix card in very specific and potentially incorrect ways with fire strike extreme the power color card lands at top again with the OC switch at 11,000 22 points clearly demonstrating that the over power Headroom is outputting higher o'clock so things are working properly and the clocks are reporting at least relative to each other properly we don't know if there are a hundred percent accurate and can't really be sure but based on performance we can see that they are at least somewhat relative to each other accurate so that's our 80 to 100 Hertz increase from earlier the maximum performance change is still about 5.2 percent increase for the best power color number versus the worst number on the charge and the performance win and FPS would range from 55 to 58 FPS with the fire strike test just to give you a frame of reference that we won't make a chart for that the Strix card is once again performing it worse than the reference card which we think comes down to some kind of driver handshake with fire strike 1080p the scaling is identical to what we saw before showing a performance range of 4.5% from top to bottom we'll also include times by numbers linked in the article in the description below if you want to see those so that's really it this is pretty easy video the power color is ready to review we don't normally do a preview test to validate if it's time to review them but the idea here is to set the stage so everyone's clear on how the drivers interact with a different cards so it there's a good understanding of why we're reviewing what we are when we are to recap this is ready to go we'll review it very soon already tested this and retested it with the newest drivers so these will be on a level ground we are focusing on out of box testing we will do overclock testing power consumption and the usual suite noise testing all that stuff with the power colored card this one we've done thermals on but the challenge is that at this point we're just waiting to do the game testing and we will only wait up until a point that you can buy the card or up until the point that asus tells us is ready and in a final state for review until then there's no point in reviewing it because it won't represent the final product and if we're just publishing it two weeks ahead of the launch of that product that changes a day before launch then we'd have three do it all anyway and it's just going to be easier for everyone if we wait but only so long so that's everything that's Vega it's taking a long time to get to this point Andy has had supply limitations and issues so partner cards are scarce to begin with and then they haven't worked for like two months but we don't know whose fault that is we don't know if that's AMD we don't know if that's the board partners it could have something to do with the V bio signature that was introduced it could have entirely nothing to do with that and be all about the drivers it could be the board partners building the BIOS with lower clocks than we expect I don't know but hopefully it's fixed and check back subscribe so that you can check that content when it is up and as always patreon.com slash to helps out directly you can go to extort gamers nexus dotnet to make a shirt like this one probably the most direct way to help us out short of just watching and subscribing to the content so thank you for watching I'll see you all next time