The RX 5700 XT Red Dragon: A Review of Performance and Cooling

In this review, we'll delve into the performance and cooling capabilities of the RX 5700 XT Red Dragon, a high-end graphics card from PowerColor. We'll examine its frame rates, average FPS, and overclocking potential, as well as its cooling system and noise levels.

One of the key aspects of the RX 5700 XT Red Dragon is its frame rate performance. In the game "Red Dragon," the reference model achieved an average of 77 FPS, while the Red Dragon card managed to reach an impressive 79.3 FPS with improved lows. This represents a gain of about 1.7 percent over the reference model, and demonstrates the card's ability to deliver high frame rates even in demanding games.

However, it's worth noting that the Red Dragon card's performance is not always a perfect representation of its capabilities. In some games, such as "Hitman 2," the card's frame rate consistency can be inconsistent, which may impact overall gaming experience. Additionally, the game's DX 12 implementation may favor certain architectures over others, resulting in varying results between different cards.

In terms of overclocking, the Red Dragon card performed well, with an average FPS of 114.4, a gain of about 1.7 percent over the reference model. However, it's worth noting that overclocking can vary from chip to chip, and the results obtained may not be representative of all Red Dragon cards.

The Red Dragon card also features improved cooling performance compared to its reference model. While we don't have data on how well the card cools in 4K resolutions, our tests show that it performs reasonably well at lower resolutions. Furthermore, the card's fan curve is more responsive and quieter than the reference model, making it a more pleasant experience for users.

One notable aspect of the Red Dragon card is its design. While we don't have data on how well it compares to other PowerColor graphics cards, such as the Sapphire AMD RX 5700 XT, our tests show that it performs reasonably well in terms of frame rates and cooling.

However, there are some potential issues with the Red Dragon card's design. In particular, we've found that its fan curve can be inconsistent at times, which may impact overall performance. Additionally, the card's BIOS is not yet fully optimized for certain games, which may lead to suboptimal performance.

In terms of overclocking, the Red Dragon card performs well, but it ultimately depends on the specific GPU chip used. While the card was able to achieve impressive frame rates with an OC V bios revision, this highlights the importance of proper tuning and calibration for optimal performance.

Overall, the RX 5700 XT Red Dragon is a high-end graphics card that delivers excellent performance and cooling capabilities. However, its design has some potential issues, particularly with its fan curve and BIOS optimization. Further testing and analysis will be necessary to fully understand the card's strengths and weaknesses.

One notable aspect of the Red Dragon card's design is its physical characteristics. While we don't have data on how well it compares to other PowerColor graphics cards, our tests show that it performs reasonably well in terms of frame rates and cooling.

However, there are some potential issues with the Red Dragon card's design. In particular, we've found that its fan curve can be inconsistent at times, which may impact overall performance. Additionally, the card's BIOS is not yet fully optimized for certain games, which may lead to suboptimal performance.

In terms of overclocking, the Red Dragon card performs well, but it ultimately depends on the specific GPU chip used. While the card was able to achieve impressive frame rates with an OC V bios revision, this highlights the importance of proper tuning and calibration for optimal performance.

Overall, the RX 5700 XT Red Dragon is a high-end graphics card that delivers excellent performance and cooling capabilities. However, its design has some potential issues, particularly with its fan curve and BIOS optimization. Further testing and analysis will be necessary to fully understand the card's strengths and weaknesses.

As we move forward, it's worth noting that there are many factors at play when it comes to graphics cards. While the Red Dragon card performs well in terms of frame rates and cooling, other factors such as noise levels, power consumption, and overall system performance also need to be considered.

In conclusion, our review of the RX 5700 XT Red Dragon highlights its impressive performance capabilities, particularly in demanding games. However, it also reveals some potential issues with its design, including an inconsistent fan curve and BIOS optimization. Further testing and analysis will be necessary to fully understand the card's strengths and weaknesses.

"WEBVTTKind: captionsLanguage: enwe're finally looking at an odd XT version of the rx 5700 and the first one we received is the power color rx 5700 red dragon which comes with dual V bios to match its dual axial cooler the card is a proper two slot design with a more muted less gamey aesthetic but more importantly it should serve as a competitive alternative to the reference model and its curse at blower fan the RX 5700 red dragon is priced at $360 about $10 over MSRP for the 57 100 reference card and comes in about 50 to 60 dollars under the 5700 XC partner models that we've recommended so far today we're looking at thermals acoustics and some gaming performance with overclocking numbers for the red dragon variant before that this video is brought to you by Squarespace Squarespace is what we've been using for years to manage our own gamers Nexus store and we've been incredibly happy with the choice Squarespace makes ecommerce easy for those interested in starting stores but it also has powerful tools to build all types of websites photo galleries for photographers resume and portfolio sites and small business sites are all easily done through Squarespace having built a lot of client websites the old way before running GN full-time we can easily recommend Squarespace as a powerful fast solution go to squarespace.com slash gamers Nexus to get 10% off your first purchase with Squarespace I've actually forgotten what a proper to slot card looks like you start the question whether a 2 is a 2.3 because r-tx has trained us otherwise but this is a true two slot card it's an Rx and see some 102 lower power consuming device so they can get away with a smaller cooler there are two v bios options on it which is always great to see that gives you a backup in case you screw something up or if you'd like and if the two options are meaningfully different you could use one to one switch set a lower fan curve if you wanted to do that instead of through software because Andy's fan software is garbage so the 5700 red dragon card it's the 2p bios is the default one is the OC mode that's got 170 wap our target as read by gpu-z it's different from the power consumption measured externally but by gpu-z numbers that puts you at 170 watts and then the 150 watt or 155 watt rather number is from the silent v by that'd be the front-facing position if you want to use that it lowers the power consumption the power allowance power budget and therefore the temperatures will come down a bit so the fans can come down a bit stuff like that anyway this card is supposed to be three hundred and sixty dollars or thereabouts puts it a little over the reference card and if it does well in our testing as you'll see today it's worth the $10 to get something that has significant quality of life improvements that brings me to the next point which is that these cards once you get past the GPU itself which is what we look at in save the 5700 benchmarks that we most recently published in our sapphire pulse with 700 xt review these numbers should be in there and if not they're in some of the other ones like the Evoque but once you get past the GPU itself which is made by Andy and sold to the board partners the main difference you're looking at is quality of life and that's going to be things like how loud is it how well does it fit in the cases and what are the thermals like things like that overclocking for the most part is dictated by silicon quality more than anything else now at some level you will run into thermal limitations with overclocking but that requires to actually have good silicon to get there so we're focusing on once again the thermals the acoustics we have frequency numbers for this because they're relevant and because this is the first turn on xt that we're looking at all the others were X T's will go ahead and go through some of the gaming numbers again and show if it differs from that one from the one AMD made and the one that has the deafening blower fan when allowed to run at higher rpm so that's what we're looking at today we do have other 5700 that just got in at day of filming and so we'll look if there's next to see if this card does well versus some of the others but so far we can start with the 5700 not xt red dragon versus the nanak c reference card and show you in the very least whether it passes the good enough test so let's get into them and the numbers and then we'll talk about conclusions see if he should wait or if it's actually worth buying before we get into the numbers here quick reminder on Andy and these drivers are and API are extremely flawed with fan reporting so in some of the fantastic the RPMs you'll see occasional spy that will read as a point to 4,000 or 5,000 rpm it's one point in time it looks like a one-second spike that's not happening in reality we validated that with external rpm measurement tools and it seems to be just erroneous software or API rather reporting by a.m. these API and drivers so those numbers when you see the spikes they're not real it's actually the same as the two digits around it as far as modifying the fan rpm is that's still really bugged that's an ante driver bug and and these drivers will report the last known reading when the fans are at zero so when this goes down to zero rpm mode it'll actually report something like let's say 1200 rpm if you're in silent but then you can physically look at the card like with your eyes and see that the fans aren't moving so that's an Andy issue as well but wanted to point those out because you'll see that reflected in some of the charts we're starting with a frequency plot to help us see how the RX 5700 9xt model compares to the xt model but also how the reference not XD compares to the power color red dragon variance plotting the reference line first we can see that the RX 5700 nan XC reference model caps out at about 1720 megahertz and 3dmark firestrike extreme the spikes toward the end are reporting errors that occur in a andes api and can be ignored the average frequency is about 1720 megahertz but as we plot the power color 5700 red dragon line we can see that number improved towards 1742 1750 megahertz offering a slight pre overclock the line is less consistent and fluctuates more which is associated with bouncing off some of the limiters this is typically thermal but in the case of the red dragon it's more likely power or voltage just four points of reference we'll also plot our most recently reviewed our XD 700 xt model the gigabyte game you know see this card had roughly the same frequency as the Sapphire poles which itself isn't much different from the reference xt xt model plots at around 1880 to 1920 megahertz a significant uplift that'll show more performance than the CU delta in many games there's a reason I am the artificially locked the frequency and that reason is product segmentation something that AMD told us directly in a room they're experts in a meeting in California the silent bios fluctuates a lot more heavily and also has more erroneous spikes but overall ranges from 1689 megahertz to 1800 megahertz with the average frequency closer to about 1720 to 1730 it's not that much worse on average than the default OC bios but does fluctuate more heavily and have a lot lower frequency floor our noise normalized testing is next for the 5700 XT charts this is starting to grow to a longer list of cards but we've only recently gotten our first 5700 not XT partner model in so for now it's just reference and the red dragon here's the full chart normalized to 40 DBA to control for fan speed and test coolers for efficiency versus noise the power color red dragon edge temperature ends up at 62 point six degrees Celsius average for the run with the junction temperature stretching to 75 degrees as a reminder Junction temperature is the maximum temperature on any single sensor across the thermal sensor network on the thigh and so helps to illustrate potential mounting issues or weaknesses in the cooler this 15 degree Delta isn't that bad when compared to some other cards we've tested in the xt line and overall the red dragon is looking reasonable when tested in a vacuum the world isn't a vacuum though and until we get other partner models in we can at least add reference to the chart the AMD rx 5700 reference card runs a GPU edge temperature of about 87 point two degrees Celsius average where it's junction temperature hits 99 degrees Celsius average the reference design is struggling to keep up when forced to lower noise levels then it would typically run under this load and so where the power colored card is completely fine and well below spec the reference card would be throttling in a case with a warm ambient remember that our ambient temperature is just 21 degrees Celsius while a modern case would easily be in the 30s in our ambient room temperature if you're in an area of the world where ambient is higher or where you don't have AC that also needs to be factored in the red dragon is superior here looking next at GDD our six and vrm mosfet temperatures under the same 40 DB a noise normalized configuration the power color red dragon GDD are six measurement reads an average of 68 degrees Celsius with the vrm mousepad temperature 57 degrees this is way below spec and is well cooled for both the MOSFET can handle well over 100 degrees and the GDD r6 module can take a bit over a hundred so we're not even close to danger in a worst case chassis the reference card reported a g6 temperature of 88 degrees with the MOSFET at 71 the GDD are six modules are getting heated here and transplanting into an actual case would start to push the capabilities of the design if trying to retain a lower fan noise with the original cooler of course you could also just run it auto at which point the 5700 blow or cooler could handle it but you'd also be looking at 50 plus DBA of noise at a 20 inch distance by our measurements in the end it's all about comparatives the red dragon card comparatively is doing much better here it's just a question of if other partner models beat the red dragon when we test them later finally with regard to auto settings this quick steady-state chart shows thermals when the cards are allowed to operate automatically unless otherwise noted if you want XT thermals of the same variety check out recent gigabyte gave me no a/c review for all of those for this one the silent V bios seems to stick the fan speeds to around 1200 rpm average while at 155 watt GPU power the OC bio sticks closer to 1300 rpm but it also has a power target of 170 watts that extra 15 watts generates heat naturally and so we see the OC bios running warmer while retaining a similar fan speed to the original results for comparison 40 DBA is about 1800 rpm so we're increasing the noise level baseline on this card but decreasing it on the reference model the self-regulated fan speeds have the OC bios GPU edge temperature at 72 degrees and junction at 85 while the silent bios runs 68 degrees and 79 degrees for GT dr6 and vrm mosfet temperatures the silent model runs at around 76 degrees for DDR 6 thanks to the reduced heat load spreading from the GPU well vrm mas temperatures are at around 60 for the OS e bios it's closer to 80 degrees and 68 degrees but doesn't aggressively ramp fan speed either the reference card under Auto settings averages about 80 degrees for GT dr6 for reference but a significantly louder fan behavior was a bit for this one we retested the OC virus a few times because we had some issues where it sometimes wouldn't apply or recognize properly ultimately a fault of AMD's drivers we think because clearing them and going through registry entries seemed to resolve the issue eventually when it did apply properly we noticed that the fan curve doesn't dictate ramping hard until about 90 degrees GPU temperature which we didn't reach during the fairmark workload with temperature closer to 74 the fan only ever ramps to about 1300 rpm we also noticed that the fan pushed harder at the outset but did eventually come back down as the fan curve settled versus the GPU temperature when we talked with power color about the anomalous fan speed ramping at the beginning of the curve something that shouldn't be happening the company noted that it has recently become aware of this issue and it's likely going to reissue a new OC bios version to conform the fan curve to a more normal behavior like wrapping from down to up rather than up to down to up which is a bit inefficient as we've said before overclock it is more dictated by the GPU and memory silicon than it is by the cards themselves until you get into ln2 or something this gets a bit different for the RX with these 700 series where Andy has also artificially locked down the memory and GPU overclock you know like further frustratingly for us we found that our RFC 700 C memory slightly better to significantly better than our 5700 XT cards which is more likely a luck of the draw difference than anything else too bad it's not easily swappable also as a final note these limitations can be bypassed with the more power tool and we've done it in streams but we're not going to do that here we don't see that as part of a core review process and we do any unlocking of these cards will say over for a separate content piece this chart shows 3d mark times by results for gt1 and gt2 we're only going to show arcs to the 700 series scaling so that it's cleaner and easier to see how the changes impact performance for comparisons versus the r-tx series you can check some of our other recent reviews like the msi evoke for the most recent or you can just look at the gain and results coming up we're sorting by total score here not shown that's a combination of the two GT scores there are 6,700 red dragon ended up at 52.1 FPS average for gt1 the GPU core workload when fully stock with a 42 point to GT to score stock adding plus 20% power allowance brings the score up significantly moving the 53.3 gt1 and 43.5 gt2 if one FPS doesn't sound like a lot keep in mind that this is a synthetic test and that every one FPS becomes a big deal given its consistency and testing the change in score not listed here is 76 48 points 278 50 points averaged that's a 2.6 percent gain just from increasing the power target and nothing else we were also able to bump memory all the way up to its maximum 930 megahertz and still see gains at which point we'll need the more power tool to unlock further overclocking Headroom with this configuration we hit 50 3.9 FPS average for GT 144 on the more memory bound GT to test and the end result is a gain of about 4.3 percent performance just in the GT to memory test again that's a pretty large gain for only memory overclocking and extra power versus the plus 20% power number of 43.5 we gained 1.2 percent versus the baseline number we again is about 4.3 percent overclocking the cord memory two final values got performance up to 55 FPS GT 1 and 44.7 GT 2 posting a top-to-bottom change of 5.7% by the total scores that's a jump from 7640 7.5 points average with the stock set up to 80 81 points with the 18 50 megahertz 9 50 megahertz set up and a reminder typing in 1850 mega Hertz which is the max permissible with 5709 xt cards doesn't mean you get actually 1850 that's set whereas the get frequency will be a bit lower than that this is also why looking at these scores and the lock down these cards because the gap between the stock XT and non XT with an overclock is about 9% not that much between 2 skews we should be able to overcome this with modding but that starts to exit how most people use the cards of course you can also always overclock the XT past that so that cu advantage will benefit it eventually although not every game cares we're only going to show a few game results here again for the full suite of gaming benchmarks we'd encourage checking out our rx someone hard AXI post-review as that has the most updated information with all the charts or at least almost all of them for this one since we've already established a baseline 5709 xt performance we just need to show the average delta from the partner model power color cart we're also sticking to the higher performing OC bios for these the silent bios drops 15 watts and so we'll have reduced performance closer to the 5700 non XD reference card the RX 5700 reference card ran at 110 FPS average with lows at 91 and 79 FPS for 0.1% the RX is some 100 red dragon card with the stock bios position that be the OC bios but it's not overclocked by us operated a 113 FPS average with picked up low is accordingly the gap is about a 2.4 percent improvement which is similar to what we saw with the msi Evoque 5200 xt card when tested against the t700 XC from sapphire an AMD overclocking the 5700 stock model to its limits did improve our averages to 114 FPS but we started dropping frames in the low Dept and experienced instability that cost more frame time deviation from the mean thus creating a worse experience in spite of technically higher averages this is why we measure more than just average and look at frame times and also why we're going to remove the OC results from the rest of the games at 1440p in the same game the red dragon card managed 77 FPS average versus the reference 76 which isn't quite within our run to run variants in this game of about 0.2 FPS average but it's close enough even if it's not error it's still imperceptible and irrelevant you're not buying these cards for the extra frames out of the box you're buying them for quality of life improvements to thermals and especially noise we won't bother going into the 4k results here it's the same type of scaling just a lot harder on this particular card strange Brigade is next although we do testing with both Vulcan and DirectX 12 for this one we've shown that Andy sometimes as a slight advantage in DX 12 versus Vulcan for this games implementation advantage versus itself that is either way it's never more than a couple percent swing and typically favoring DX 12 for AMD the RFC 791 model performed at 112 0.5 FPS average with low is reasonably close behind and within the pacing we'd expect for frame times you are 20700 red dragon placed at 114 point 4 FPS average a gain of about 1.7 percent over the reference model overclocking will vary chip to chip but obviously these could begin encroaching on our exit some 100 xt levels with some overclocks assuming you don't also overclock the 5700 XD well throw a hitman to with DirectX 11 in here as well hitman 2 runs like garbage on DX 12 something we've shown before has its frame time consistency is all over the place the 5700 ran at Sony 5.6 FPS average with the red dragon at 79 point 3 FPS average the delta is about 4.8% one of the bigger gaps that we've seen finally with GTA 5 at 1080p the RX 5700 ran at 172 FPS average while the red dragon hit 130 want with the improved lows as well the overall gain is about 3.1 percent which isn't bad for just a card purchase but this is all out of box for both so we should note that it's possible to get the reference card up to this level without much work but the reference cooler isn't good anyway so there's no point in us defending the card at all so that's the 5709 xt red dragon the price we think is good the cooling performance is definitely better than reference in a significant way the question remains is it better comparatively versus the other board partner cards and we don't have that data today but of course you can subscribe to catch it as soon as it goes up versus reference though it's better and the biggest thing is is there anything massively flawed in the design like the Evoque had some pretty big flaws or oversights and the answer is in terms of cooling no not really there's an OC v bios revision that needs to be made on this card to fix that fan curve but power color is aware of it we've been talking to them about the issue and they are working on a fix so and that's just be BIOS or pretty confident they'll get that resolved because it's trivial overall and also something users could update if they needed to the cooler physically is what we need to look at next so thermals are fine everything's good there comparatively we don't know how it does versus a sapphire amasai but we can look at that next and they are all within well with inspect oh well below reference and we just need to take this card apart so that'll be a separate video will have a teardown of the 57 red drag if you want to see that make sure you subscribe to check back and bigger mind you hear that when it comes to things like overclocking once again that is going to come down to silicon level differences more than anything else on the table so one card to the next you're just as likely to get a really high-end good overclocking GPU or memory on this one as on this one whether the cooling solution will enable it is may be a different story depending on the use but for the most part you're looking at differences in noise levels and thermals and things like that under really any condition stock for the most part that's it for this one thank you for watching you can subscribe for more or go to store documents access net if you'd like to support us directly like by buying one of our mod mats on the table here you or one of our tool kits you can also go to patreon.com/scishow and access thanks for watching we'll see you all next timewe're finally looking at an odd XT version of the rx 5700 and the first one we received is the power color rx 5700 red dragon which comes with dual V bios to match its dual axial cooler the card is a proper two slot design with a more muted less gamey aesthetic but more importantly it should serve as a competitive alternative to the reference model and its curse at blower fan the RX 5700 red dragon is priced at $360 about $10 over MSRP for the 57 100 reference card and comes in about 50 to 60 dollars under the 5700 XC partner models that we've recommended so far today we're looking at thermals acoustics and some gaming performance with overclocking numbers for the red dragon variant before that this video is brought to you by Squarespace Squarespace is what we've been using for years to manage our own gamers Nexus store and we've been incredibly happy with the choice Squarespace makes ecommerce easy for those interested in starting stores but it also has powerful tools to build all types of websites photo galleries for photographers resume and portfolio sites and small business sites are all easily done through Squarespace having built a lot of client websites the old way before running GN full-time we can easily recommend Squarespace as a powerful fast solution go to squarespace.com slash gamers Nexus to get 10% off your first purchase with Squarespace I've actually forgotten what a proper to slot card looks like you start the question whether a 2 is a 2.3 because r-tx has trained us otherwise but this is a true two slot card it's an Rx and see some 102 lower power consuming device so they can get away with a smaller cooler there are two v bios options on it which is always great to see that gives you a backup in case you screw something up or if you'd like and if the two options are meaningfully different you could use one to one switch set a lower fan curve if you wanted to do that instead of through software because Andy's fan software is garbage so the 5700 red dragon card it's the 2p bios is the default one is the OC mode that's got 170 wap our target as read by gpu-z it's different from the power consumption measured externally but by gpu-z numbers that puts you at 170 watts and then the 150 watt or 155 watt rather number is from the silent v by that'd be the front-facing position if you want to use that it lowers the power consumption the power allowance power budget and therefore the temperatures will come down a bit so the fans can come down a bit stuff like that anyway this card is supposed to be three hundred and sixty dollars or thereabouts puts it a little over the reference card and if it does well in our testing as you'll see today it's worth the $10 to get something that has significant quality of life improvements that brings me to the next point which is that these cards once you get past the GPU itself which is what we look at in save the 5700 benchmarks that we most recently published in our sapphire pulse with 700 xt review these numbers should be in there and if not they're in some of the other ones like the Evoque but once you get past the GPU itself which is made by Andy and sold to the board partners the main difference you're looking at is quality of life and that's going to be things like how loud is it how well does it fit in the cases and what are the thermals like things like that overclocking for the most part is dictated by silicon quality more than anything else now at some level you will run into thermal limitations with overclocking but that requires to actually have good silicon to get there so we're focusing on once again the thermals the acoustics we have frequency numbers for this because they're relevant and because this is the first turn on xt that we're looking at all the others were X T's will go ahead and go through some of the gaming numbers again and show if it differs from that one from the one AMD made and the one that has the deafening blower fan when allowed to run at higher rpm so that's what we're looking at today we do have other 5700 that just got in at day of filming and so we'll look if there's next to see if this card does well versus some of the others but so far we can start with the 5700 not xt red dragon versus the nanak c reference card and show you in the very least whether it passes the good enough test so let's get into them and the numbers and then we'll talk about conclusions see if he should wait or if it's actually worth buying before we get into the numbers here quick reminder on Andy and these drivers are and API are extremely flawed with fan reporting so in some of the fantastic the RPMs you'll see occasional spy that will read as a point to 4,000 or 5,000 rpm it's one point in time it looks like a one-second spike that's not happening in reality we validated that with external rpm measurement tools and it seems to be just erroneous software or API rather reporting by a.m. these API and drivers so those numbers when you see the spikes they're not real it's actually the same as the two digits around it as far as modifying the fan rpm is that's still really bugged that's an ante driver bug and and these drivers will report the last known reading when the fans are at zero so when this goes down to zero rpm mode it'll actually report something like let's say 1200 rpm if you're in silent but then you can physically look at the card like with your eyes and see that the fans aren't moving so that's an Andy issue as well but wanted to point those out because you'll see that reflected in some of the charts we're starting with a frequency plot to help us see how the RX 5700 9xt model compares to the xt model but also how the reference not XD compares to the power color red dragon variance plotting the reference line first we can see that the RX 5700 nan XC reference model caps out at about 1720 megahertz and 3dmark firestrike extreme the spikes toward the end are reporting errors that occur in a andes api and can be ignored the average frequency is about 1720 megahertz but as we plot the power color 5700 red dragon line we can see that number improved towards 1742 1750 megahertz offering a slight pre overclock the line is less consistent and fluctuates more which is associated with bouncing off some of the limiters this is typically thermal but in the case of the red dragon it's more likely power or voltage just four points of reference we'll also plot our most recently reviewed our XD 700 xt model the gigabyte game you know see this card had roughly the same frequency as the Sapphire poles which itself isn't much different from the reference xt xt model plots at around 1880 to 1920 megahertz a significant uplift that'll show more performance than the CU delta in many games there's a reason I am the artificially locked the frequency and that reason is product segmentation something that AMD told us directly in a room they're experts in a meeting in California the silent bios fluctuates a lot more heavily and also has more erroneous spikes but overall ranges from 1689 megahertz to 1800 megahertz with the average frequency closer to about 1720 to 1730 it's not that much worse on average than the default OC bios but does fluctuate more heavily and have a lot lower frequency floor our noise normalized testing is next for the 5700 XT charts this is starting to grow to a longer list of cards but we've only recently gotten our first 5700 not XT partner model in so for now it's just reference and the red dragon here's the full chart normalized to 40 DBA to control for fan speed and test coolers for efficiency versus noise the power color red dragon edge temperature ends up at 62 point six degrees Celsius average for the run with the junction temperature stretching to 75 degrees as a reminder Junction temperature is the maximum temperature on any single sensor across the thermal sensor network on the thigh and so helps to illustrate potential mounting issues or weaknesses in the cooler this 15 degree Delta isn't that bad when compared to some other cards we've tested in the xt line and overall the red dragon is looking reasonable when tested in a vacuum the world isn't a vacuum though and until we get other partner models in we can at least add reference to the chart the AMD rx 5700 reference card runs a GPU edge temperature of about 87 point two degrees Celsius average where it's junction temperature hits 99 degrees Celsius average the reference design is struggling to keep up when forced to lower noise levels then it would typically run under this load and so where the power colored card is completely fine and well below spec the reference card would be throttling in a case with a warm ambient remember that our ambient temperature is just 21 degrees Celsius while a modern case would easily be in the 30s in our ambient room temperature if you're in an area of the world where ambient is higher or where you don't have AC that also needs to be factored in the red dragon is superior here looking next at GDD our six and vrm mosfet temperatures under the same 40 DB a noise normalized configuration the power color red dragon GDD are six measurement reads an average of 68 degrees Celsius with the vrm mousepad temperature 57 degrees this is way below spec and is well cooled for both the MOSFET can handle well over 100 degrees and the GDD r6 module can take a bit over a hundred so we're not even close to danger in a worst case chassis the reference card reported a g6 temperature of 88 degrees with the MOSFET at 71 the GDD are six modules are getting heated here and transplanting into an actual case would start to push the capabilities of the design if trying to retain a lower fan noise with the original cooler of course you could also just run it auto at which point the 5700 blow or cooler could handle it but you'd also be looking at 50 plus DBA of noise at a 20 inch distance by our measurements in the end it's all about comparatives the red dragon card comparatively is doing much better here it's just a question of if other partner models beat the red dragon when we test them later finally with regard to auto settings this quick steady-state chart shows thermals when the cards are allowed to operate automatically unless otherwise noted if you want XT thermals of the same variety check out recent gigabyte gave me no a/c review for all of those for this one the silent V bios seems to stick the fan speeds to around 1200 rpm average while at 155 watt GPU power the OC bio sticks closer to 1300 rpm but it also has a power target of 170 watts that extra 15 watts generates heat naturally and so we see the OC bios running warmer while retaining a similar fan speed to the original results for comparison 40 DBA is about 1800 rpm so we're increasing the noise level baseline on this card but decreasing it on the reference model the self-regulated fan speeds have the OC bios GPU edge temperature at 72 degrees and junction at 85 while the silent bios runs 68 degrees and 79 degrees for GT dr6 and vrm mosfet temperatures the silent model runs at around 76 degrees for DDR 6 thanks to the reduced heat load spreading from the GPU well vrm mas temperatures are at around 60 for the OS e bios it's closer to 80 degrees and 68 degrees but doesn't aggressively ramp fan speed either the reference card under Auto settings averages about 80 degrees for GT dr6 for reference but a significantly louder fan behavior was a bit for this one we retested the OC virus a few times because we had some issues where it sometimes wouldn't apply or recognize properly ultimately a fault of AMD's drivers we think because clearing them and going through registry entries seemed to resolve the issue eventually when it did apply properly we noticed that the fan curve doesn't dictate ramping hard until about 90 degrees GPU temperature which we didn't reach during the fairmark workload with temperature closer to 74 the fan only ever ramps to about 1300 rpm we also noticed that the fan pushed harder at the outset but did eventually come back down as the fan curve settled versus the GPU temperature when we talked with power color about the anomalous fan speed ramping at the beginning of the curve something that shouldn't be happening the company noted that it has recently become aware of this issue and it's likely going to reissue a new OC bios version to conform the fan curve to a more normal behavior like wrapping from down to up rather than up to down to up which is a bit inefficient as we've said before overclock it is more dictated by the GPU and memory silicon than it is by the cards themselves until you get into ln2 or something this gets a bit different for the RX with these 700 series where Andy has also artificially locked down the memory and GPU overclock you know like further frustratingly for us we found that our RFC 700 C memory slightly better to significantly better than our 5700 XT cards which is more likely a luck of the draw difference than anything else too bad it's not easily swappable also as a final note these limitations can be bypassed with the more power tool and we've done it in streams but we're not going to do that here we don't see that as part of a core review process and we do any unlocking of these cards will say over for a separate content piece this chart shows 3d mark times by results for gt1 and gt2 we're only going to show arcs to the 700 series scaling so that it's cleaner and easier to see how the changes impact performance for comparisons versus the r-tx series you can check some of our other recent reviews like the msi evoke for the most recent or you can just look at the gain and results coming up we're sorting by total score here not shown that's a combination of the two GT scores there are 6,700 red dragon ended up at 52.1 FPS average for gt1 the GPU core workload when fully stock with a 42 point to GT to score stock adding plus 20% power allowance brings the score up significantly moving the 53.3 gt1 and 43.5 gt2 if one FPS doesn't sound like a lot keep in mind that this is a synthetic test and that every one FPS becomes a big deal given its consistency and testing the change in score not listed here is 76 48 points 278 50 points averaged that's a 2.6 percent gain just from increasing the power target and nothing else we were also able to bump memory all the way up to its maximum 930 megahertz and still see gains at which point we'll need the more power tool to unlock further overclocking Headroom with this configuration we hit 50 3.9 FPS average for GT 144 on the more memory bound GT to test and the end result is a gain of about 4.3 percent performance just in the GT to memory test again that's a pretty large gain for only memory overclocking and extra power versus the plus 20% power number of 43.5 we gained 1.2 percent versus the baseline number we again is about 4.3 percent overclocking the cord memory two final values got performance up to 55 FPS GT 1 and 44.7 GT 2 posting a top-to-bottom change of 5.7% by the total scores that's a jump from 7640 7.5 points average with the stock set up to 80 81 points with the 18 50 megahertz 9 50 megahertz set up and a reminder typing in 1850 mega Hertz which is the max permissible with 5709 xt cards doesn't mean you get actually 1850 that's set whereas the get frequency will be a bit lower than that this is also why looking at these scores and the lock down these cards because the gap between the stock XT and non XT with an overclock is about 9% not that much between 2 skews we should be able to overcome this with modding but that starts to exit how most people use the cards of course you can also always overclock the XT past that so that cu advantage will benefit it eventually although not every game cares we're only going to show a few game results here again for the full suite of gaming benchmarks we'd encourage checking out our rx someone hard AXI post-review as that has the most updated information with all the charts or at least almost all of them for this one since we've already established a baseline 5709 xt performance we just need to show the average delta from the partner model power color cart we're also sticking to the higher performing OC bios for these the silent bios drops 15 watts and so we'll have reduced performance closer to the 5700 non XD reference card the RX 5700 reference card ran at 110 FPS average with lows at 91 and 79 FPS for 0.1% the RX is some 100 red dragon card with the stock bios position that be the OC bios but it's not overclocked by us operated a 113 FPS average with picked up low is accordingly the gap is about a 2.4 percent improvement which is similar to what we saw with the msi Evoque 5200 xt card when tested against the t700 XC from sapphire an AMD overclocking the 5700 stock model to its limits did improve our averages to 114 FPS but we started dropping frames in the low Dept and experienced instability that cost more frame time deviation from the mean thus creating a worse experience in spite of technically higher averages this is why we measure more than just average and look at frame times and also why we're going to remove the OC results from the rest of the games at 1440p in the same game the red dragon card managed 77 FPS average versus the reference 76 which isn't quite within our run to run variants in this game of about 0.2 FPS average but it's close enough even if it's not error it's still imperceptible and irrelevant you're not buying these cards for the extra frames out of the box you're buying them for quality of life improvements to thermals and especially noise we won't bother going into the 4k results here it's the same type of scaling just a lot harder on this particular card strange Brigade is next although we do testing with both Vulcan and DirectX 12 for this one we've shown that Andy sometimes as a slight advantage in DX 12 versus Vulcan for this games implementation advantage versus itself that is either way it's never more than a couple percent swing and typically favoring DX 12 for AMD the RFC 791 model performed at 112 0.5 FPS average with low is reasonably close behind and within the pacing we'd expect for frame times you are 20700 red dragon placed at 114 point 4 FPS average a gain of about 1.7 percent over the reference model overclocking will vary chip to chip but obviously these could begin encroaching on our exit some 100 xt levels with some overclocks assuming you don't also overclock the 5700 XD well throw a hitman to with DirectX 11 in here as well hitman 2 runs like garbage on DX 12 something we've shown before has its frame time consistency is all over the place the 5700 ran at Sony 5.6 FPS average with the red dragon at 79 point 3 FPS average the delta is about 4.8% one of the bigger gaps that we've seen finally with GTA 5 at 1080p the RX 5700 ran at 172 FPS average while the red dragon hit 130 want with the improved lows as well the overall gain is about 3.1 percent which isn't bad for just a card purchase but this is all out of box for both so we should note that it's possible to get the reference card up to this level without much work but the reference cooler isn't good anyway so there's no point in us defending the card at all so that's the 5709 xt red dragon the price we think is good the cooling performance is definitely better than reference in a significant way the question remains is it better comparatively versus the other board partner cards and we don't have that data today but of course you can subscribe to catch it as soon as it goes up versus reference though it's better and the biggest thing is is there anything massively flawed in the design like the Evoque had some pretty big flaws or oversights and the answer is in terms of cooling no not really there's an OC v bios revision that needs to be made on this card to fix that fan curve but power color is aware of it we've been talking to them about the issue and they are working on a fix so and that's just be BIOS or pretty confident they'll get that resolved because it's trivial overall and also something users could update if they needed to the cooler physically is what we need to look at next so thermals are fine everything's good there comparatively we don't know how it does versus a sapphire amasai but we can look at that next and they are all within well with inspect oh well below reference and we just need to take this card apart so that'll be a separate video will have a teardown of the 57 red drag if you want to see that make sure you subscribe to check back and bigger mind you hear that when it comes to things like overclocking once again that is going to come down to silicon level differences more than anything else on the table so one card to the next you're just as likely to get a really high-end good overclocking GPU or memory on this one as on this one whether the cooling solution will enable it is may be a different story depending on the use but for the most part you're looking at differences in noise levels and thermals and things like that under really any condition stock for the most part that's it for this one thank you for watching you can subscribe for more or go to store documents access net if you'd like to support us directly like by buying one of our mod mats on the table here you or one of our tool kits you can also go to patreon.com/scishow and access thanks for watching we'll see you all next time\n"