**Gaming Card Review: Sapphire Radeon RX 5700 XT**

We leave most of the gaming benchmarks and the baseline stuff to the initial reference design review where we look more at gaming performance than anything else. So, in these reviews, we assume you already want it and all we really care about is comparison from one of that model versus the next.

But here, because this is getting so expensive, not that 440 is particularly high, but it's encroaching on 2070 super territory. You can find them a lot of the partner models are about $500. So, should you buy it? Is a tough question to answer when considering RTX versus the other RX models. Here's what we think.

If you are looking at an Rx 57 or XT and you're definitely gonna buy one of those, you just don't know which one this is by far the best cooler that we've yet tested on those models. There's a lot more we haven't tested but it's unlikely that many will compete directly with the performance of this card.

This quarter, you can tell by looking at the design as as I mean it's pretty obvious that it's gonna be one of the best performers for cooling. It costs more but from strictly a standpoint of 5700 T's the reason you buy this aside from having some additional power to play with an overclocking although you can do that through soft power play tables, mind you but if you watched our stream with Joe that's a massive pain and they're not particularly fun to work with so you get a bit more power out of the e bios.

You have some more flexibility on cooling. You can push the cooling harder and get more clock headroom potentially sort of like competitive or an enthusiast overclocking standpoint we can recommend the card it's good for that from a cooling standpoint if you are really neurotic about noise and you just you can't stand noise, a benefit of these larger designs is that you can drive the fan rpm down quite low and still have better or equivalent thermals to a lot of the less beefed up designs. The smaller cards especially so the point where you can you basically can't hear the fans over anything else in the system so that's a benefit too.

And you do pay for that an extra 10, 20 bucks versus the other models like the pulse or something but those are the main benefits if you don't care about either of those things probably not worth it but if you're interested in an enthusiast just kind of hacking around with the card especially with BIOS mods because you've got backups here or you really enjoy the idea of having a tunable fan curve to the point where you still have good thermals on the device but you can drive down the noise to be below your system ambient noise or system case fan noise stuff like that it's good.

It's just those are the primary reasons. So, we do like the design the cards well done. It performs very well. It is superior and the charts for thermals especially noise normalized the auto settings are a bit they're a bit soft. We think that Sapphire probably should have gone more aggressive with the fan curve on that but that's nothing you can't solve by just setting your own fan curve or potentially flashing view bios if they release updates.

Versus the 20 70 super, that is more difficult it's basically you'll you have potential for more gaming performance from a twenty seventy super if you're willing to spend the extra money and you're not that distant from it. So then it just becomes a question of would you rather buy a lower and twenty seventy super that is going to be a bit louder on average and not have as good of a cooler design comparatively but will have more headroom for a high-end gaming performance or would you rather buy the high-end 5700 XT get all the quality of life features but then lose a bit of the gaming performance.

And that's going to come down to basically what monitor you have what refresh rate are you trying to hit what resolution is you trying to hit and if this is priced at 440 it's not unreasonable it's just at the high end. We like seen high-end cards because they're fun they're fun to work with and there's definitely buyers for them so good job Sapphire on the design card as well. It's we don't really have any complaints about it. It's just it becomes really difficult on the price.

**In Conclusion**

If you are interested in an enthusiast just kind of hacking around with the card especially with BIOS mods because you've got backups here or you really enjoy the idea of having a tunable fan curve to the point where you still have good thermals on the device but you can drive down the noise to be below your system ambient noise or system case fan noise stuff like that it's good. It's just those are the primary reasons.

So, we do like the design the cards well done. It performs very well. It is superior and the charts for thermals especially noise normalized the auto settings are a bit they're a bit soft. We think that Sapphire probably should have gone more aggressive with the fan curve on that but that's nothing you can't solve by just setting your own fan curve or potentially flashing view bios if they release updates.

Versus the 20 70 super, that is more difficult it's basically you'll you have potential for more gaming performance from a twenty seventy super if you're willing to spend the extra money and you're not that distant from it. So then it just becomes a question of would you rather buy a lower and twenty seventy super that is going to be a bit louder on average and not have as good of a cooler design comparatively but will have more headroom for a high-end gaming performance or would you rather buy the high-end 5700 XT get all the quality of life features but then lose a bit of the gaming performance.

"WEBVTTKind: captionsLanguage: enit's been a few GPU launch cycles since board partners really cared to invest heavily in to AMD video card designs and and these cards for as competitive as some of the GPS have been haven't been particularly exciting we've gotten a lot of refreshes in the last few years but that's changed with navi sapphires our RX 5700 XC Nitro+ is on our bench for review today as one of the most physically imposing kitted out RX 5700 xt models we've yet seen with a reported MSRP of theoretically $440 baseline we're looking to see whether the price hike is worthwhile over not only other partner models but also over the RT x-series cards that hover around the $500 mark today we're reviewing the sapphire Nitro+ in depth for noise thermals overclocking power and more before that this video is brought to you by be quiet and it's straight power 11 series power supplies the straight power 11 PSU is shipped from 450 Watts up to a thousand Watts accommodating most of the gaming PC build requirements you'd encounter and focuses on delivering a higher quality power supply that doesn't sacrifice on efficiency or stability noise is also a heavy point for the straight power 11 using a 135 millimeters silent wins 3 fan that can spin as low as 200 rpm for quieter low load operation learn more at the link in the description below this is a sapphire nitro plus it is a higher end card from the 5700 xt line it's quite large and we're gonna have separate teardown on this one and it deserves a standalone piece because it is a as stated very kitted out it's got a lot of interesting features and my cooler design on the board so we're gonna do a separate video talking about all that today we're focusing heavily on the unmodified and unopened thermals acoustics and power targets because power targets are a big deal and a big separator on these partner cards this when I was a triple bios which typically you only really want a dual V BIOS dual v BIOS is as you work more and more with video cards you start to appreciate it because it gives you this safety in that if you want to try something if you get interested in modern or store overclocking with BIOS mods things like that you're gonna flash it on there and there's no real risk of damage because as long as you don't flash all three of them or all two of them in the case of dual of you buyers you always have a backup you can always fix it and you can restore both be bios's this one's got three which is exiting the point of necessity and just getting into being excessive but we'll test all three of those just for their stock behaviors it does give you some room to play if you want to try a couple of different mods on if you get into maybe competitive benchmarking stuff like that but it does also have a practical use it's just that getting into this class of cards obviously more expensive and as you encroach upon territory of a twenty somebody super or r-tx cards you start deviating from the comparison that really would it would matter for the most part which is how does it do versus other partner model fifty-seven rx tees and drawing closer to the comparison of but should you just buy something that's a bit more expensive still and get more performance so we'll talk briefly about that today the big thing though is that thermals and noise especially with these larger coolers you'll be able to drive down the noise level at a given temperature and so that's something that's important to look at so it's a triple axial fan card it's got a triple v bios and then it's supposed to be four hundred forty dollars and it's got a thirty dollar RGB fan add-on which we don't have here but once you're doing that you're really a you've basically priced it out to a completely different tier of card because you now have four hundred seventy dollars so anyway let's get through all the testing will look pretty detailed at the thermals and do some line graphs on frequency behavior then we can talk about overclocking and if it's actually worth it there almost have been getting trickier to measure even with 40 DB a noise normalized thermal testing varying maximum power budgets on each card make the comparisons imperfect and so a card which draws 220 watts will be disproportionately hot when compared to a card draw in two hundred watts even with the same noise levels before getting into thermal testing we need to look at power consumption of each B BIOS on the Sapphire Nitro+ there are three of them we're using GPU only power draw for this so this doesn't count memory power consumption and vrm efficiency losses we're just looking at GPU here's a chart showing the Sapphire I'd show power limits for each of the three V BIOS options ignored the hard downward spikes these are reporting errors and have no bearing on the actual results for the default V bios position the leftmost one power consumption runs at about 220 watts constantly this is higher than most of the 5700 XT cards actually all of them that we've measured so far and for the most part that's a good thing that means more overclocking Headroom in theory or more clocking Headroom in general over or not the middle V bios measured the lowest of the three plotting at about 195 watts average and closer to competition the rightmost V bios also plotted at about 220 watts for reference if we look at an average DV a power graph to best understand the different cards on the market the Sapphire Nitro+ is the most power hungry of these models meaning that it has more power allowance for boosting the clocks without software or powerful a table modifications the MSI evoke only has 1 V bios and that V bios runs at 200 watts average EMA size of oak is roughly equivalent to one of the V biases for sapphire also at nearly 200 watts and the Sapphire 5700 xt pulso cv bios is also in this range the gigabyte gaming OC runs a bit lower at 190 watts average with the reference card the lowest measuring 179 watts average GPU only our next test will look at 40 DB a noise normalize thermals but there are a few explanations first as always testing methodology is defined a bit more in the article below but the basics are that this testing allows us to better look for efficiency of cooler designs by eliminating the ability for coolers to blast their way to the top of the charts with it louder or faster fans it's not exactly a fair comparison if there's a 60 DB a cooler doing better than a 40 DB a cooler obviously and so we normalize them the downside of this is that power really should also be considered because the Nitro has a V BIOS that pulls 220 watts and another that pulls about 195 200 will measure both the 195 watt version will be directly comparable to every other partner model tested thus far well the 220 watt version will pull more power than the others thus running warmer noise measurements will be taken with a DB meter at 20 inches away in a room with a noise floor of 26 DB romanian temperature is measured second a second and AC is controlled to 21 degrees Celsius with offset modifiers applied if necessary deviation and ambience is less than plus or minus one degree Celsius and humidity is controlled to approximately 45% starting with noise normalized GP thermals then we'll be looking at both the edge and junction temperature where the latter is a measurement to the hottest spot on the die with the former being a measurement of the literal edge temperature of the dice one of the cooler parts configured to roughly the same power target as the competition each power target is plotted on the left axis the Sapphire Nitro+ is easily the most efficient cooler on the chart Sapphire manages an impressive seventy two point five degree Junction temperature winning the chart by a longshot it's not even close the next best is a set of four cards all around the 81 to 83 degree range edge temperature is nearly the same as the Gigabyte rx1r xt that we tested but edge temperature is less important than junction temperature as junction is what dictates boosting Headroom and a potential thermal limiter for overclocking that's said with the TJ Max of 110 degrees this is a non-issue for all of the non-reference cards on this chart Sapphire has a massive incredible lead in this chart and if you're wondering why that would even be relevant if we're operating at levels where lower thermals don't really grant any advantages keep the following in mind Sapphire could be run instead at lower noise level still perhaps at 32 to 35 DBA in our bench while still retaining equivalent thermals to other coolers that might be louder like 40 DBA this offers a noise advantage although at the obvious cost of wealth more money that's for the 220 watt Nitro+ version it's managing the same thermals as the pulse the modded evoke and the gigabyte card but while pulling 10% more power this is why if you just measured that without accounting for the power difference it really wouldn't look like that impressive of a cooler the sheer mass of the Nitro alone is helping here not to mention the increased cooling capabilities provided by the triple axial design GTD our six and vrm Moss testing is next for GGG our six thermals the Nitro Plus at 200 watts again comes out ahead of everything else in fact the Nitro is so far ahead that it's 220 watt V by experience is the next best one on the chart despite the 10% increase in power consumption to the core at 200 watts noise normalize thermals place the nitro at 68 degrees G DDR 6 temperature with the 220 watt option at 70 degrees followed by gigabytes previous chart leading 70 4.7 degree results at the end of the day as we stated in the gigabyte review the actual difference is sort of irrelevant once you get down to the top three quarters of this chart everything below the pulse and thermals is and below meaning up on this chart is distant enough from exceeding comfortable thermals that they'd all do fine there's no real benefit from running mosfet or GD r6 a couple degrees or even 115 degrees cooler when all the results are already so far from the maximum permissible values and you're probably cooling based on GPU anyway these don't behave like GPUs there's no additional butene Headroom if a mosfet is cold or if a memory module is cold although there is potential for damage if it runs too hot but in this instance too hot means 125 degrees Celsius and above on MOSFETs or about 105 110 degrees on the GD are six modules and caps of course we're rated a bit lower but they don't really get that hot on video cards keep in mind that sticking any of these cards in a case will raise the effective T ambient from 21 degrees to maybe the mid 30s in our environment or worse if you're in a warmer ambient and that's assuming a well cooled case and maybe a light CPU load if you're the type of person who prefers to run the card stock and without controlling fan speeds we've got some testing with the cards left to auto to govern their fan speeds for out-of-the-box thermals so have fires thermals look a lot worse than the 40 DB a result one left to auto control but that all comes down to how the fan curve is written the end result is that the Nitro+ ends up mid-pack with the right v bios which about eighty eight point six degrees Junction with the last V bios predictably next door and the lower 200 watt v bios ends up at about 91 degrees Junction but with quieter fan speeds this chart certainly tells a different story from the other and if a reviewer tested only with all the settings it make the nitro look a lot worse than reality dictates we'd suggest setting a custom fan curve for the card if you're going to buy one although these temperatures are acceptable they're too warm for a cooler this good and it comes down to how the fans are programmed in v bios well plot fan behavior now with the leftmost V bios the fan ramps to about 1230 rpm or roughly 37 to 38 DBA initial ramp is much harder hitting about 1500 rpm but it calms down they're after the middle of eubiose which sticks closer to 1000 rpm and acts as a more silent option matched it's 195 watt power draw at 1000 rpm this card produced about thirty-six point five to thirty six point seven DBA of noise at twenty inches in our bench the RPM is fairly fixed otherwise with no meaningful fluctuations to speak of as for the full acoustic scale the range is 36.4 DBA to fifty nine point seven DBA with our measurement we're in the final ten percent fan of pwm doesn't really move the RPM at all or much but this is an AMD driver issue that's consistent on all cards it's not a sapphire then two of the 3v by OSA stick closer to the 1200 rpm mark with one closer to one thousand rpm running at 40 DBA puts you at around 1400 RPM for a point of reference to the previous charts there's plenty of headroom to scale up for heavier overclock and without using crazy cooling so this card may be a good candidate for power play tables frequency responses last before talking overclocking for frequency response we saw the higher power target via bios options hitting about 1960 megahertz with the lower 195 watt power budget configuration running at around 1900 megahertz again a point of reference the msi evoke not shown here but we have the data in our review holds between 1940 and 1960 megahertz so we'll need to look at some overclocking to further differentiate needs on to performance and overclocking results briefly again the focus is not on gaming performance here because they're all pretty much the same this one's a bit different it's special but most of the 57 hard XD is one of the 2070 supers all those cards within a given GPU skew most of them are going to be basically identical to each other this wouldn't be different though so this card had a particularly strong memory with no issues achieving a 9 50 megahertz memory set and will need to use power play tables later to see if we can push things further this would actually be a really good liquid-nitrogen candidate and we'll see if we can get the L&T pot mounted on it because the frequent the memories are so good we're not clear obviously on why that is most the time it's just luck of the draw it could be the IMC it could be the memory cells probably the IMC and whether or not sapphire does any real bidding here we're not sure but it's probably luck of the draw core frequency seemed to succeed at a set frequency between 21 20 and 21 40 megahertz which is really good that's the max that AMD will allow basically it's pretty damn close for the 5700 XT you can do more with power plate tables definitely and if you're into that sort of enthusiast tweaking we absolutely recommend it it might be something we'll do separately if there's enough interest in it and do a separate content piece on we've done it before but this card might be worth doing it again please note that this is not the same though as get frequency in this chart you're looking at the labeling is set that's the number we typed in the number that comes out is always a bit different you can see that in the time spot results some of the overclock scaling down on the Nitro+ is looking pretty good it's actually the 2070 super and the rx-7 hardac see reference model are added here for points of reference the 2070 super managed a significantly stronger memory score and gt2 than all of the 5700 X T's and that's a significant portion of where it's higher scoring is coming from because gt2 weights pretty heavily so this indicates that Navi is still stuck on memory clocks although the 5700 XC nitro did encroach upon the total score of the 20 so d super not shown here because blow out the chart but the weighted scores are 10 won 8-1 points 10,000 won 81 for the 2070 super and 98 48 points to the Nitro+ establishing a lead for the super of about 3.3 percent won the super is stock overclocking it would get you more obviously well the fifties or 100x the reference under the quad nitrogen we scored 10 6 8 7 or 70 3.2 FPS gt1 and 58.8 FPS GG 2 lagging behind a gt2 really hard there because the our reference model is stuck about 900 915 megahertz memory Nitro+ is the best on-air overclock we've yet used for the 5700 XT although it remains to be seen how much of that is from luck of the draw again your mileage will vary not may vary but it will vary that said we're really happy with this particular card and it's a fantastic candidate for ln2 and for PowerCLI tables hopefully that's true for all of them but we can't say shout of the Tomb Raider at 4k the 5700 I've seen nitro with the stock V bios ends up at forty nine point eight FPS average ranking roughly tied with the 27 super reference card are about 2 points percent ahead of the Evoque and further ahead of the 5200 icy poles over o'clock in the Nitro+ but just past a 1080 TI FTW 3 and equivalent to the Radeon 7 that distance from the r-tx 2080 gain overclocking the 2070 super puts it past the Nitro+ OC to be fair at 1440p the Nitro+ runs about tied with RT x 2070 super and gtx 980ti holding a lead over the RX 2200 xt pulse is 85 PS result about 5.5 percent that's a much larger game then we'll see elsewhere and is absolutely a noteworthy one just don't expect to repeat everywhere in GTA 5 at 4k there are 6700 xt Nitro+ stock V BIOS Landers performance at 50 point 5 FPS average ranking it as roughly equivalent with the other 5200 XT cards on the chart there's no meaningful gain at stock overclocking the card manually to a set frequency not the same as get again of 21 40 megahertz and 9 50 megahertz memory the Nitro plus ends up at 54 FPS average positioning at not distant from the 2070 super the 2070 super reference card 7 FPS results maintains a lead of approximately 5.9 percent over the overclocked 5700 XT although the 2070 super could be further overclocked to keep that distance jumping to 61 0.6 FPS average here granted a gain of about 14% over the hybrid 5700 XT with an overclock the Nitro+ manages the lead of 3.6% at 1440p vr x 5700 xt Nitro+ performs at 102 FPS average positioning that equivalent to the Evoque and the charts with the lows within the wider error of low performance the Nitro plus ends up about 1% over the reference rx 2 D 1 or XT or sapphire pulsar x57 r xt when either is that it's 100 FPS average overclocking boosts it notably to 107 a gain of 5 percent of the stock Nitro+ this also ties it with an overclock to 2017 super-sorry non super non a die card which was among the worst 2070 models before being discontinued twice and video discontinued the non a model which this is that was the only a most RP model at the time and then also discontinued the 27 t the RT x 2080 maintains a lead of about 7% and strange brigade with DX 12 and at 1440p the 5700 i've seen Plus runs at about 128 FPS average passing the 5700 XC pulls by 2.6 percent and the pulse is a card that we were pretty happy to recommend previously it said well overclock is Nitro puts it about tied with the stock 20 Sony super and 20 80 while overclocking the super allows it to leapfrog again typically these partner board reviews we're assuming that you already have decided you want the GPU that comes on the board that be the 5700 XT here and that's because we leave most of the gaming benchmarks and the baseline stuff to the initial reference design review where we look more at gaming performance than anything else so in these reviews we assume you already want it and all we really care about is comparison from one of that model versus the next but here because this is getting so expensive not that 440 is particularly high but it's encroaching on 2070 super territory you can find them a lot of the partner models are about $500 so should you buy it is a tough question to answer when considering RTX versus the other rx models here's what we think if you are looking at an Rx 57 or XT and you're definitely gonna buy one of those you just don't know which one this is by far the best cooler that we've yet tested on those models there's a lot more we haven't tested but it's unlikely that many will compete directly with the performance all of this quarter you can tell by looking at the design as as I mean it's pretty obvious that it's gonna be one of the best performers for cooling it costs more but from strictly a standpoint of 5700 T's the reason you buy this aside from having some additional power to play with an overclocking although you can do that through soft power play tables mind you but if you watched our stream with Joe that's a massive pain and they're not particularly fun to work with so you get a bit more power out of the e bios you have some more flexibility on cooling you can push the cooling harder and get more clock Headroom potentially sort of like competitive or a an enthusiast overclocking standpoint we can recommend the card it's good for that from a cooling standpoint if you are really neurotic about noise and you just you can't stand noise a benefit of these larger designs is that you can drive the fan rpm down quite low and still have better or equivalent thermals to a lot of the the less beefed up designs the smaller cards especially so the point where you can you basically can't hear the fans over anything else in the system so that's a benefit too and you do pay for that an extra 10 20 bucks versus the other models like the pulse or something but those are the main benefits if you don't care about either of those things probably not worth it but if you're interested in an enthusiast just kind of hacking around with the card especially with BIOS mods because you've got backups here or you really enjoy the idea of having a tunable fan curve to the point where you still have good thermals on the device but you can drive down the noise to be below your system ambient noise or system case fan noise stuff like that it's good it's just those are the primary reasons so we do like the design the cards well done it performs very well it is superior and the charts for thermals especially noise normalized the auto settings are a bit they're a bit soft we think that Sapphire probably should have gone more aggressive with the fan curve on that but that's nothing you can't solve by just setting your own fan curve or potentially flashing view bios if they release updates versus the 20 70 super that is more difficult it's basically you'll you have potential for more gaming performance from a twenty seventy super if you're willing to spend the extra money and you're not that distant from it so then it just becomes a question of would you rather buy a lower and twenty seventy super that is going to be a bit louder on average and not have as good of a cooler design comparatively but will have more Headroom for a high-end gaming performance or would you rather buy the high-end 5700 XT get all the quality of life features but then lose a bit of the gaming performance and that's going to come down to basically what monitor you have what refresh rate are you trying to hit what resolution is you trying to hit and if this is priced at 440 it's not unreasonable it's just at the high end and we like seen high-end cards because they're fun they're fun to work with and there's definitely buyers for them so good job sapphire on the design card as well it's we don't really have any complaints about it it's just it becomes really difficult on the price and I have a hard time picking aside between the options but hopefully all that helps you and if it doesn't and just look at all the data and decide on your own make up your own mind you have enough information to do it thanks for watching subscribe for more check back for the tear down of the card we'll talk more about the internals opinions on that stuff and you can also go to patreon.com/scishow and access or store documents access data to help us out directly like by buying one of our shirts or our mod mats or toolkits and I'll see you all next timeit's been a few GPU launch cycles since board partners really cared to invest heavily in to AMD video card designs and and these cards for as competitive as some of the GPS have been haven't been particularly exciting we've gotten a lot of refreshes in the last few years but that's changed with navi sapphires our RX 5700 XC Nitro+ is on our bench for review today as one of the most physically imposing kitted out RX 5700 xt models we've yet seen with a reported MSRP of theoretically $440 baseline we're looking to see whether the price hike is worthwhile over not only other partner models but also over the RT x-series cards that hover around the $500 mark today we're reviewing the sapphire Nitro+ in depth for noise thermals overclocking power and more before that this video is brought to you by be quiet and it's straight power 11 series power supplies the straight power 11 PSU is shipped from 450 Watts up to a thousand Watts accommodating most of the gaming PC build requirements you'd encounter and focuses on delivering a higher quality power supply that doesn't sacrifice on efficiency or stability noise is also a heavy point for the straight power 11 using a 135 millimeters silent wins 3 fan that can spin as low as 200 rpm for quieter low load operation learn more at the link in the description below this is a sapphire nitro plus it is a higher end card from the 5700 xt line it's quite large and we're gonna have separate teardown on this one and it deserves a standalone piece because it is a as stated very kitted out it's got a lot of interesting features and my cooler design on the board so we're gonna do a separate video talking about all that today we're focusing heavily on the unmodified and unopened thermals acoustics and power targets because power targets are a big deal and a big separator on these partner cards this when I was a triple bios which typically you only really want a dual V BIOS dual v BIOS is as you work more and more with video cards you start to appreciate it because it gives you this safety in that if you want to try something if you get interested in modern or store overclocking with BIOS mods things like that you're gonna flash it on there and there's no real risk of damage because as long as you don't flash all three of them or all two of them in the case of dual of you buyers you always have a backup you can always fix it and you can restore both be bios's this one's got three which is exiting the point of necessity and just getting into being excessive but we'll test all three of those just for their stock behaviors it does give you some room to play if you want to try a couple of different mods on if you get into maybe competitive benchmarking stuff like that but it does also have a practical use it's just that getting into this class of cards obviously more expensive and as you encroach upon territory of a twenty somebody super or r-tx cards you start deviating from the comparison that really would it would matter for the most part which is how does it do versus other partner model fifty-seven rx tees and drawing closer to the comparison of but should you just buy something that's a bit more expensive still and get more performance so we'll talk briefly about that today the big thing though is that thermals and noise especially with these larger coolers you'll be able to drive down the noise level at a given temperature and so that's something that's important to look at so it's a triple axial fan card it's got a triple v bios and then it's supposed to be four hundred forty dollars and it's got a thirty dollar RGB fan add-on which we don't have here but once you're doing that you're really a you've basically priced it out to a completely different tier of card because you now have four hundred seventy dollars so anyway let's get through all the testing will look pretty detailed at the thermals and do some line graphs on frequency behavior then we can talk about overclocking and if it's actually worth it there almost have been getting trickier to measure even with 40 DB a noise normalized thermal testing varying maximum power budgets on each card make the comparisons imperfect and so a card which draws 220 watts will be disproportionately hot when compared to a card draw in two hundred watts even with the same noise levels before getting into thermal testing we need to look at power consumption of each B BIOS on the Sapphire Nitro+ there are three of them we're using GPU only power draw for this so this doesn't count memory power consumption and vrm efficiency losses we're just looking at GPU here's a chart showing the Sapphire I'd show power limits for each of the three V BIOS options ignored the hard downward spikes these are reporting errors and have no bearing on the actual results for the default V bios position the leftmost one power consumption runs at about 220 watts constantly this is higher than most of the 5700 XT cards actually all of them that we've measured so far and for the most part that's a good thing that means more overclocking Headroom in theory or more clocking Headroom in general over or not the middle V bios measured the lowest of the three plotting at about 195 watts average and closer to competition the rightmost V bios also plotted at about 220 watts for reference if we look at an average DV a power graph to best understand the different cards on the market the Sapphire Nitro+ is the most power hungry of these models meaning that it has more power allowance for boosting the clocks without software or powerful a table modifications the MSI evoke only has 1 V bios and that V bios runs at 200 watts average EMA size of oak is roughly equivalent to one of the V biases for sapphire also at nearly 200 watts and the Sapphire 5700 xt pulso cv bios is also in this range the gigabyte gaming OC runs a bit lower at 190 watts average with the reference card the lowest measuring 179 watts average GPU only our next test will look at 40 DB a noise normalize thermals but there are a few explanations first as always testing methodology is defined a bit more in the article below but the basics are that this testing allows us to better look for efficiency of cooler designs by eliminating the ability for coolers to blast their way to the top of the charts with it louder or faster fans it's not exactly a fair comparison if there's a 60 DB a cooler doing better than a 40 DB a cooler obviously and so we normalize them the downside of this is that power really should also be considered because the Nitro has a V BIOS that pulls 220 watts and another that pulls about 195 200 will measure both the 195 watt version will be directly comparable to every other partner model tested thus far well the 220 watt version will pull more power than the others thus running warmer noise measurements will be taken with a DB meter at 20 inches away in a room with a noise floor of 26 DB romanian temperature is measured second a second and AC is controlled to 21 degrees Celsius with offset modifiers applied if necessary deviation and ambience is less than plus or minus one degree Celsius and humidity is controlled to approximately 45% starting with noise normalized GP thermals then we'll be looking at both the edge and junction temperature where the latter is a measurement to the hottest spot on the die with the former being a measurement of the literal edge temperature of the dice one of the cooler parts configured to roughly the same power target as the competition each power target is plotted on the left axis the Sapphire Nitro+ is easily the most efficient cooler on the chart Sapphire manages an impressive seventy two point five degree Junction temperature winning the chart by a longshot it's not even close the next best is a set of four cards all around the 81 to 83 degree range edge temperature is nearly the same as the Gigabyte rx1r xt that we tested but edge temperature is less important than junction temperature as junction is what dictates boosting Headroom and a potential thermal limiter for overclocking that's said with the TJ Max of 110 degrees this is a non-issue for all of the non-reference cards on this chart Sapphire has a massive incredible lead in this chart and if you're wondering why that would even be relevant if we're operating at levels where lower thermals don't really grant any advantages keep the following in mind Sapphire could be run instead at lower noise level still perhaps at 32 to 35 DBA in our bench while still retaining equivalent thermals to other coolers that might be louder like 40 DBA this offers a noise advantage although at the obvious cost of wealth more money that's for the 220 watt Nitro+ version it's managing the same thermals as the pulse the modded evoke and the gigabyte card but while pulling 10% more power this is why if you just measured that without accounting for the power difference it really wouldn't look like that impressive of a cooler the sheer mass of the Nitro alone is helping here not to mention the increased cooling capabilities provided by the triple axial design GTD our six and vrm Moss testing is next for GGG our six thermals the Nitro Plus at 200 watts again comes out ahead of everything else in fact the Nitro is so far ahead that it's 220 watt V by experience is the next best one on the chart despite the 10% increase in power consumption to the core at 200 watts noise normalize thermals place the nitro at 68 degrees G DDR 6 temperature with the 220 watt option at 70 degrees followed by gigabytes previous chart leading 70 4.7 degree results at the end of the day as we stated in the gigabyte review the actual difference is sort of irrelevant once you get down to the top three quarters of this chart everything below the pulse and thermals is and below meaning up on this chart is distant enough from exceeding comfortable thermals that they'd all do fine there's no real benefit from running mosfet or GD r6 a couple degrees or even 115 degrees cooler when all the results are already so far from the maximum permissible values and you're probably cooling based on GPU anyway these don't behave like GPUs there's no additional butene Headroom if a mosfet is cold or if a memory module is cold although there is potential for damage if it runs too hot but in this instance too hot means 125 degrees Celsius and above on MOSFETs or about 105 110 degrees on the GD are six modules and caps of course we're rated a bit lower but they don't really get that hot on video cards keep in mind that sticking any of these cards in a case will raise the effective T ambient from 21 degrees to maybe the mid 30s in our environment or worse if you're in a warmer ambient and that's assuming a well cooled case and maybe a light CPU load if you're the type of person who prefers to run the card stock and without controlling fan speeds we've got some testing with the cards left to auto to govern their fan speeds for out-of-the-box thermals so have fires thermals look a lot worse than the 40 DB a result one left to auto control but that all comes down to how the fan curve is written the end result is that the Nitro+ ends up mid-pack with the right v bios which about eighty eight point six degrees Junction with the last V bios predictably next door and the lower 200 watt v bios ends up at about 91 degrees Junction but with quieter fan speeds this chart certainly tells a different story from the other and if a reviewer tested only with all the settings it make the nitro look a lot worse than reality dictates we'd suggest setting a custom fan curve for the card if you're going to buy one although these temperatures are acceptable they're too warm for a cooler this good and it comes down to how the fans are programmed in v bios well plot fan behavior now with the leftmost V bios the fan ramps to about 1230 rpm or roughly 37 to 38 DBA initial ramp is much harder hitting about 1500 rpm but it calms down they're after the middle of eubiose which sticks closer to 1000 rpm and acts as a more silent option matched it's 195 watt power draw at 1000 rpm this card produced about thirty-six point five to thirty six point seven DBA of noise at twenty inches in our bench the RPM is fairly fixed otherwise with no meaningful fluctuations to speak of as for the full acoustic scale the range is 36.4 DBA to fifty nine point seven DBA with our measurement we're in the final ten percent fan of pwm doesn't really move the RPM at all or much but this is an AMD driver issue that's consistent on all cards it's not a sapphire then two of the 3v by OSA stick closer to the 1200 rpm mark with one closer to one thousand rpm running at 40 DBA puts you at around 1400 RPM for a point of reference to the previous charts there's plenty of headroom to scale up for heavier overclock and without using crazy cooling so this card may be a good candidate for power play tables frequency responses last before talking overclocking for frequency response we saw the higher power target via bios options hitting about 1960 megahertz with the lower 195 watt power budget configuration running at around 1900 megahertz again a point of reference the msi evoke not shown here but we have the data in our review holds between 1940 and 1960 megahertz so we'll need to look at some overclocking to further differentiate needs on to performance and overclocking results briefly again the focus is not on gaming performance here because they're all pretty much the same this one's a bit different it's special but most of the 57 hard XD is one of the 2070 supers all those cards within a given GPU skew most of them are going to be basically identical to each other this wouldn't be different though so this card had a particularly strong memory with no issues achieving a 9 50 megahertz memory set and will need to use power play tables later to see if we can push things further this would actually be a really good liquid-nitrogen candidate and we'll see if we can get the L&T pot mounted on it because the frequent the memories are so good we're not clear obviously on why that is most the time it's just luck of the draw it could be the IMC it could be the memory cells probably the IMC and whether or not sapphire does any real bidding here we're not sure but it's probably luck of the draw core frequency seemed to succeed at a set frequency between 21 20 and 21 40 megahertz which is really good that's the max that AMD will allow basically it's pretty damn close for the 5700 XT you can do more with power plate tables definitely and if you're into that sort of enthusiast tweaking we absolutely recommend it it might be something we'll do separately if there's enough interest in it and do a separate content piece on we've done it before but this card might be worth doing it again please note that this is not the same though as get frequency in this chart you're looking at the labeling is set that's the number we typed in the number that comes out is always a bit different you can see that in the time spot results some of the overclock scaling down on the Nitro+ is looking pretty good it's actually the 2070 super and the rx-7 hardac see reference model are added here for points of reference the 2070 super managed a significantly stronger memory score and gt2 than all of the 5700 X T's and that's a significant portion of where it's higher scoring is coming from because gt2 weights pretty heavily so this indicates that Navi is still stuck on memory clocks although the 5700 XC nitro did encroach upon the total score of the 20 so d super not shown here because blow out the chart but the weighted scores are 10 won 8-1 points 10,000 won 81 for the 2070 super and 98 48 points to the Nitro+ establishing a lead for the super of about 3.3 percent won the super is stock overclocking it would get you more obviously well the fifties or 100x the reference under the quad nitrogen we scored 10 6 8 7 or 70 3.2 FPS gt1 and 58.8 FPS GG 2 lagging behind a gt2 really hard there because the our reference model is stuck about 900 915 megahertz memory Nitro+ is the best on-air overclock we've yet used for the 5700 XT although it remains to be seen how much of that is from luck of the draw again your mileage will vary not may vary but it will vary that said we're really happy with this particular card and it's a fantastic candidate for ln2 and for PowerCLI tables hopefully that's true for all of them but we can't say shout of the Tomb Raider at 4k the 5700 I've seen nitro with the stock V bios ends up at forty nine point eight FPS average ranking roughly tied with the 27 super reference card are about 2 points percent ahead of the Evoque and further ahead of the 5200 icy poles over o'clock in the Nitro+ but just past a 1080 TI FTW 3 and equivalent to the Radeon 7 that distance from the r-tx 2080 gain overclocking the 2070 super puts it past the Nitro+ OC to be fair at 1440p the Nitro+ runs about tied with RT x 2070 super and gtx 980ti holding a lead over the RX 2200 xt pulse is 85 PS result about 5.5 percent that's a much larger game then we'll see elsewhere and is absolutely a noteworthy one just don't expect to repeat everywhere in GTA 5 at 4k there are 6700 xt Nitro+ stock V BIOS Landers performance at 50 point 5 FPS average ranking it as roughly equivalent with the other 5200 XT cards on the chart there's no meaningful gain at stock overclocking the card manually to a set frequency not the same as get again of 21 40 megahertz and 9 50 megahertz memory the Nitro plus ends up at 54 FPS average positioning at not distant from the 2070 super the 2070 super reference card 7 FPS results maintains a lead of approximately 5.9 percent over the overclocked 5700 XT although the 2070 super could be further overclocked to keep that distance jumping to 61 0.6 FPS average here granted a gain of about 14% over the hybrid 5700 XT with an overclock the Nitro+ manages the lead of 3.6% at 1440p vr x 5700 xt Nitro+ performs at 102 FPS average positioning that equivalent to the Evoque and the charts with the lows within the wider error of low performance the Nitro plus ends up about 1% over the reference rx 2 D 1 or XT or sapphire pulsar x57 r xt when either is that it's 100 FPS average overclocking boosts it notably to 107 a gain of 5 percent of the stock Nitro+ this also ties it with an overclock to 2017 super-sorry non super non a die card which was among the worst 2070 models before being discontinued twice and video discontinued the non a model which this is that was the only a most RP model at the time and then also discontinued the 27 t the RT x 2080 maintains a lead of about 7% and strange brigade with DX 12 and at 1440p the 5700 i've seen Plus runs at about 128 FPS average passing the 5700 XC pulls by 2.6 percent and the pulse is a card that we were pretty happy to recommend previously it said well overclock is Nitro puts it about tied with the stock 20 Sony super and 20 80 while overclocking the super allows it to leapfrog again typically these partner board reviews we're assuming that you already have decided you want the GPU that comes on the board that be the 5700 XT here and that's because we leave most of the gaming benchmarks and the baseline stuff to the initial reference design review where we look more at gaming performance than anything else so in these reviews we assume you already want it and all we really care about is comparison from one of that model versus the next but here because this is getting so expensive not that 440 is particularly high but it's encroaching on 2070 super territory you can find them a lot of the partner models are about $500 so should you buy it is a tough question to answer when considering RTX versus the other rx models here's what we think if you are looking at an Rx 57 or XT and you're definitely gonna buy one of those you just don't know which one this is by far the best cooler that we've yet tested on those models there's a lot more we haven't tested but it's unlikely that many will compete directly with the performance all of this quarter you can tell by looking at the design as as I mean it's pretty obvious that it's gonna be one of the best performers for cooling it costs more but from strictly a standpoint of 5700 T's the reason you buy this aside from having some additional power to play with an overclocking although you can do that through soft power play tables mind you but if you watched our stream with Joe that's a massive pain and they're not particularly fun to work with so you get a bit more power out of the e bios you have some more flexibility on cooling you can push the cooling harder and get more clock Headroom potentially sort of like competitive or a an enthusiast overclocking standpoint we can recommend the card it's good for that from a cooling standpoint if you are really neurotic about noise and you just you can't stand noise a benefit of these larger designs is that you can drive the fan rpm down quite low and still have better or equivalent thermals to a lot of the the less beefed up designs the smaller cards especially so the point where you can you basically can't hear the fans over anything else in the system so that's a benefit too and you do pay for that an extra 10 20 bucks versus the other models like the pulse or something but those are the main benefits if you don't care about either of those things probably not worth it but if you're interested in an enthusiast just kind of hacking around with the card especially with BIOS mods because you've got backups here or you really enjoy the idea of having a tunable fan curve to the point where you still have good thermals on the device but you can drive down the noise to be below your system ambient noise or system case fan noise stuff like that it's good it's just those are the primary reasons so we do like the design the cards well done it performs very well it is superior and the charts for thermals especially noise normalized the auto settings are a bit they're a bit soft we think that Sapphire probably should have gone more aggressive with the fan curve on that but that's nothing you can't solve by just setting your own fan curve or potentially flashing view bios if they release updates versus the 20 70 super that is more difficult it's basically you'll you have potential for more gaming performance from a twenty seventy super if you're willing to spend the extra money and you're not that distant from it so then it just becomes a question of would you rather buy a lower and twenty seventy super that is going to be a bit louder on average and not have as good of a cooler design comparatively but will have more Headroom for a high-end gaming performance or would you rather buy the high-end 5700 XT get all the quality of life features but then lose a bit of the gaming performance and that's going to come down to basically what monitor you have what refresh rate are you trying to hit what resolution is you trying to hit and if this is priced at 440 it's not unreasonable it's just at the high end and we like seen high-end cards because they're fun they're fun to work with and there's definitely buyers for them so good job sapphire on the design card as well it's we don't really have any complaints about it it's just it becomes really difficult on the price and I have a hard time picking aside between the options but hopefully all that helps you and if it doesn't and just look at all the data and decide on your own make up your own mind you have enough information to do it thanks for watching subscribe for more check back for the tear down of the card we'll talk more about the internals opinions on that stuff and you can also go to patreon.com/scishow and access or store documents access data to help us out directly like by buying one of our shirts or our mod mats or toolkits and I'll see you all next time\n"