Radeon RX 5500 XT Review vs. GTX 1650 Super, RX 570 and RX 580: Goodbye Polaris! | Video Cards | Reviews. The lines of Nvidia video cards based on the Turing architecture and AMD Navi have been focused on the struggle in the segment of high-performance solutions for quite a long time. If you needed a card for QuadHD, 4K and / or uncompromising FullHD graphics settings, you could easily find a suitable solution among the new products from both manufacturers.
But if you needed a relatively inexpensive video card capable of playing games in FullHD at compromise graphics settings, but at the same time not worth it like all other components of the system unit at once, you would have to choose from the cards of the previous generation. More specifically, of the three representatives of the Polaris architecture (RX 570, RX 580, RX 590) and two versions of the GTX 1060 (GTX 1050 Ti, it was difficult to refer to gaming accelerators even at the time of its release).
At the moment, of course, new items are presented in this segment as well. At AMD, gaming video cards start with the Radeon RX 5500 XT in the 4GB version, and Nvidia – with the GeForce GTX 1650 Super, which is equipped with only 4GB of onboard memory.
It will be equally interesting to compare these cards with each other and with solutions of previous generations.
Meet the participants – Radeon RX 5500 XT, GTX 1650 Super, RX 570 and RX 580
Moreover, a paradoxical situation has developed recently in the segment of inexpensive video cards: prices for Radeon RX 570 and RX 580 have jumped up, and now they ask for “age” cards no less, and even more , than for new items!
Moreover – regardless of the amount of onboard memory.
And these are by no means isolated cases for the cards selected for the test.
So, the test version of the RX 570 today will cost 14.5 thousand rubles, which is only slightly less than the cost of the RX 5500 XT. But, if we consider all the RX 570s sold at the time of the publication of the video, it turns out that 14,500 is the bottom line, and the maximum price is a fantastic 17,000.
The GTX 1650 Super will cost the same 14,500, despite the fact that it is much newer and potentially faster than the RX 570. However, it is noteworthy that the lower price limit for this GPU is 13,600 rubles. Yes, these will be the most budgetary versions with one fan and the most simple radiators, but still. The upper bar, however, is the same 17 thousand that are asked for the expensive versions of the RX 570.
And the price of the RX 580 is completely amazing – almost 18.5 thousand, and at the same time it is just starting from 18 thousand, the maximum price is 21 900, for which you can already consider the RX 5600 XT in budget versions (and expensive versions, as the previous one showed review , he doesn’t really need it).
Simply put, a potential buyer has to choose between older and more expensive video cards, and more affordable new products.
It will be all the more interesting to find out what the budget Radeon and GeForce of the current generation can offer.
RX 5500 XT and PCI-e version
Before moving on to the review of the test card, it is worth mentioning the nuances concerning the model itself.
If for the RX 5600 XT the actual topic was the BIOS update , which legally increases the operating frequencies, then any discussion regarding the RX 5500 XT will certainly touch upon the effect of the PCI-e interface bandwidth on the card’s performance.
The relevant material has already been published on the pages of the Club , but its main theses should be repeated here.
Yes, RX 5500 XT, like other graphics cards from the Navi family, supports both PCI-express version 4.0 and previous interface versions. Yes, unlike the older models based on GPU Navi 10, only 8 PCI-e lanes are available to it instead of 16. But in practice, this circumstance does not affect the performance of the video card itself:
The performance of the younger version of the RX 5500 XT may be affected by insufficient on-board memory. However, the performance drop in this case is also noted when using PCI-e 4.0 and PCI-e 3.0.
For the older version of the card with 8 gigabytes of on-board memory, changing the version of the PCI-e interface does absolutely nothing – and this is also natural, since the volume of its own storage is enough for it anyway, and there is no reason to access the RAM resources.
For clarity, in today’s testing, the RX 5500 XT will operate exclusively in PCI-e 3.0 mode. Moreover, both in synthetic tests and in games:
Option of the article for those who cannot read
Sapphire Radeon RX 5500 XT PULSE
The test card, like the previously reviewed RX 5600 XT, is made by Sapphire and belongs to the budget Pulse line. And, like the older model, there is absolutely nothing to say about its packaging and configuration: the box contains only documentation.
Appearance and design
But the card, as is usually the case with the Pulse series, is much more interesting than its packaging.
So, at first glance, you can see that the cooling system clearly continues the canons set by the Navi 10 models. And the point is not only in the decorative plastic casing, but also in the proprietary 100-mm fans. Yes, these are the same turntables that are installed on the older models of the line. And yes: here they are also removable.
The back of the card is, as expected, covered with a metal backplate, but here the canons have already been violated: it does not participate in cooling. However, there is no need to talk about the “air bag” here, since there are slots for ventilation in the backplate.
On the side, you can see other similarities with the RX 5600 XT. The card also has a bios switch and two possible modes of operation – although for the 5500 XT the difference between “power efficient” and “performance” modes is ridiculously small.
But the additional power connector remained eight-pin – which is already strange, given the power consumption of the card and the lack of a corresponding adapter in the kit. Typically, the GTX 1650 Super in the KFA 2 version, which uses a 6-pin auxiliary power connector, has an adapter from two Molex connectors.
The set of interface connectors is also similar to the RX 5600 XT: three DisplayPort and one HDMI. But DVI is absent, and this is a minus for a budget solution. Again, there is no adapter included in the package.
Cooling system and PCB – RX 5600 XT / 5700/5700 XT
From a first glance at the back side of the board, you can understand that its design is very different from the older Navi 10 models, if only because the GPU is literally in the center of the board.
Therefore, for the sake of curiosity the RX 5500 XT will also be disassembled and reviewed.
As mentioned above, the fans here are removable, as on the RX 5600 XT / 5700/5700 XT. This means that if you need to clean the radiator from dust, you will not have to contact the service center or say goodbye to the warranty by disassembling the entire card.
But a complete analysis of the card is impossible here without damaging the warranty seals:
At first it may seem that 4 spring-loaded screws around the perimeter are securing the backplate, but this is not the case. They attach a plastic fan cover to the backplate. However, if you unscrew them, the shroud will not separate from the rest of the structure as it did with the RX 5600 XT. In this case, it is screwed to the radiator with 4 more screws, which cannot be reached from the outside of the card.
The board, by the way, is also attached to the backplate with 4 more screws, and also from the inside.
Thus, to disassemble the card, you need to remove the entire cooling system. That is, to unscrew the screws securing the cross on the back of the chip, damaging the warranty seals.
By unscrewing eight screws, you can remove the radiator along with the shroud and fans. And note that, unlike Navi 10-based maps, here all the elements that need cooling are grouped on the front side of the board.
After removing the radiator, the backplate can also be dismantled, but there is no practical sense in this: it does not participate in cooling in any way, and there is nothing to cool from the rear side.
As expected, the RX 5500 XT board is very different from the older models, and it’s not just about dimensions and arrangement of elements.
The younger Navi 14 did not get the power stages used in the VRM of boards based on Navi 10 chips. The usual scheme is used here, in which the mosfets and their drivers are separate elements.
But you still can’t blame Sapphire for saving: VRM has as many as 8 phases, and 6 of them are allocated to power the GPU. Considering the power consumption of the Navi 14, that’s a lot.
But the heatsink of the cooling system – on the contrary, is very similar to similar parts in older cards. Of course, the location of the heat sinks is completely different here, but the design itself is quite consistent:
The heat from the GPU is removed by a copper plate that distributes the load through the heat pipes. A heat sink is also provided for memory microcircuits, which is in direct contact with the fins of the radiator. The VRM of the video card is cooled in the same way.
There are three heat pipes in the design of the radiator – just like in the RX 5600 XT, which obviously has higher power consumption. They have a diameter of 6 mm, are evenly spaced along the body of the radiator and are soldered both to the heat sink and to the fins themselves.
The plastic shroud is attached to the radiator with 4 screws. Removing it reveals that the heatsink is a simple array of ribs. Of the aerodynamic optimizations, only slots in the zone of heat pipes can be noted here.
Installation in the case – RX 5500 XT
Sapphire Pulse is traditionally one of the most compact nerefs, but in the case of the RX 5500 XT it is most noticeable. With a compact printed circuit board, the card acquires its dimensions due to the cooling system, but even in this case, it turns out to be even slightly smaller than a standard ATX motherboard:
Naturally, in this case, there is no question of a lack of space for a video card in a standard case. The card occupies exactly two slots in height, so there should be no questions with expansion boards either.
Rivals – Radeon RX 5500 XT Review vs. GTX 1650 Super, RX 570 and RX 580
In this case, the competitors for the test card were selected solely based on the criterion of 4 gigabytes of onboard memory and availability on sale at the time of the tests. As a result, we got the following selection:
GTX 1650 Super is represented by the version from KFA 2 – EX (1-Click OC) . By the way, this is one of the most adequate versions of cards based on this GPU: the price is still low, and the cooling system is quite serious.
RX 570 – MSI Armor OC , one of the most budget and least popular versions of cards based on Polaris chips. Nevertheless, for the RX 570, the cooling system capabilities of this card should be sufficient, especially if we bear in mind that the nominal GPU power consumption limit here remains at around 120 watts.
RX 580 – Asus DUAL OC , is no longer a budget version, but rather a mid-range version that can be compared with Sapphire Pulse and Gigabyte Gaming.
Test stand and testing methodology
The configuration of the test bench is already familiar to readers from previous materials:
- CPU: AMD Ryzen 9 3900X
- CPU cooling system: ID-Cooling SE-224-XT Basic;
- Thermal interface: Arctic MX-4;
- Motherboard: Gigabyte X570 Aorus Master
- Disk subsystem: SSD Gigabyte GP-ASM2NE6200TTTD + HDD Western Digital WD10EZRX-00A8LB0;
- RAM: G.Skill SniperX F4-3400C16D-16GSXW, 2x8gb;
- Case : Corsair Carbide 270R
- Power supply : Cougar GX-F 750.
During the tests, the frequency of the central processor was fixed at around 4200 MHz, and the RAM was overclocked to 3800 MHz while maintaining the standard timings. Of course, with the simultaneous overclocking of the Infinity Fabric bus to 1900 MHz, thus maintaining a 1: 1 ratio between the frequency of the RAM and its controller.
Tests were conducted using Windows 10 Professional build 1909 with the latest updates as of August 28, 2020.
For games, the FullHD resolution was chosen, since it was clear in advance that higher resolutions would either be too tough for the test cards, or would force us to significantly lower the graphics settings.
But the settings in some games still had to be reduced. And – not because of the capabilities of the GPU, but because otherwise the cards simply did not have enough memory. In particular, in Red Dead Redemption, the settings were reduced to medium and all additional effects were disabled, and in Resident Evil 3 the volume of textures was reduced from 3 to 1 gigabyte. Otherwise, the graphics settings are identical to those used in the RX 5600 XT benchmark.
Frequency model and overclocking …
Radeon RX 570
The card, although it is supplied with the OC suffix in the model name, actually boasts a very conditional factory overclocking: 1268 MHz by the GPU, the memory frequency is standard 1750 MHz.
It is noteworthy that when passing the Unigine Superposition benchmark (and in games, as you can see in the video above), the GPU frequency is noticeably lower than the maximum possible, and is in the range of 1200-1210 MHz:
The peak power consumption of the GPU, according to MSI Afterburner monitoring, is 123.2 watts, according to Hwinfo64 – 121.9 watts. The peak fan speed is 1548 (1550) rpm, and the GPU temperature indicators in both utilities converge – 73 degrees at the peak.
The overclocking of the MSI card was not the most impressive. So, the standard 2250 MHz memory did not obey – color flashes appeared literally in the first scene of the Superposition benchmark. The stability limit turned out to be 2200 MHz – most likely, the point here is in the individual potential of the chips, and not in overheating.
The GPU agreed to pass synthetic tests at 1350 MHz (although, of course, the real frequency was also lower), but in this case artifacts already appeared in games. Only 1330 MHz turned out to be stable:
To do this, it was necessary to raise the power consumption limit by 45% at once, which in this case was accompanied by an automatic increase in the voltage on the GPU and a strong increase in the actual power consumption. At the same time, it was not possible to achieve a stable GPU frequency: it still fluctuated within the range from 1260 to 1320 MHz:
The power consumption of the GPU increased to 175 watts, which immediately affected the fan speed of 2819 (2811) rpm automatically making the MSI card the loudest in today’s test. The GPU temperature, however, was kept within 78-79 degrees due to the high speed.
=Radeon RX 580
The Asus card is also slightly overclocked: 1360 MHz on the GPU with the nominal 1750 MHz on the memory:
However, in this case, although the power consumption limit of the GPU is also left standard for the RX 580, the frequency fluctuates within smaller limits: from 1310 to the maximum possible 1360:
The maximum temperature of the GPU is 71 degrees, the peak fan speed is 1879 rpm. In practice, this means that the card is slightly noisier than the RX 570 in its normal mode, but remains relatively comfortable.
Overclocking also turns out to be more effective: the memory agreed to work only at 2210 MHz, but the GPU obeyed 1440 MHz, which can be considered quite a decent result for a 14-nm Polaris chip that does not bear the XTR suffix.
At the same time, the frequency is almost always kept at the maximum set point, and power consumption does not increase as much as in the case of the RX 570:
According to MSI Afterburner, the maximum power consumption of the GPU is 189.2 watts, while the Hwinfo64 shows 184.7 watts. The peak temperature is 83 (82) degrees, but this is a consequence of the fact that the fan is in no hurry to raise the speed. Peak 1987 rpm makes this version of the RX 580 relatively quiet even when accelerating.
GeForce GTX 1650 Super
The base frequency of the GPU of the test card is standard 1530 MHz, the peak value in dynamic overclocking is already slightly higher than the nominal one: 1740 MHz instead of 1725. its actual temperature:
In this case, the technology behaves exactly as it should. The chip starts briskly from 1920 MHz, but as it heats up, the frequency drops: when approaching 70 degrees, relatively modest for Turing 1875-1850 MHz are obtained:
Nvidia cards do not show actual power consumption via MSI Afterburner, preferring to show a percentage of the nominal limit. For the GTX 1650 Super, it is 100 watts, and the KFA 2 version retains the nominal value here.
Also, this version of the card lacks a tachometer on the fans, the rotation speed can be monitored exclusively as a percentage of the maximum. It, according to third-party reviews, is 3500 rpm, so 55% should be about 1925 rpm.
The peak GPU temperature in the normal mode was 72 degrees – not an outstanding indicator against the background of the rest of the participants, but the card is noticeably quieter.
As in the case of the Palit RTX 2060, the overclocking of the test card is limited by the fact that the power consumption limit here simply cannot be raised above the nominal value. By the way, this is a typical picture for the GTX 1650 Super: most versions are limited to the same 100 watts.
In this case, the memory agreed to work stably by adding 1230 MHz to the nominal frequency. 1250 – already accompanied by colored dots literally in the very first scenes of the Unigine Superposition.
The GPU, on the other hand, was able to add only 170 MHz, which in practice started from 2055 MHz and dropped to 2010 MHz when warming up to 70 degrees:
The peak power consumption of the card naturally practically did not change, as did the temperature of the GPU with the rotational speed of the fans.
Radeon RX 5500 XT
The test version of the card offers a base frequency of 1737 MHz and a maximum dynamic overclocking frequency of 1845 MHz:
In practice, the GPU frequency during the Unigine Superposition ranges from 1780 to 1800 MHz, and Navi does not have the same obvious temperature dependence as Turing:
However, the temperatures are lower here: only 70 degrees at the peak. The merit is, of course, primarily the Navi 14 itself, whose peak power consumption is only 107 watts, which is only 8 watts higher than that of the GTX 1650 Super.
Of course, considering how efficient the cooling system is on the Sapphire card, we should have expected low fan speeds – and 1035 (or 1056 according to Hwinfo64) rpm at peak is not a surprise. And, of course, there can be no talk of any noise here.
But overclocking, as in the case of the RX 5600 XT, brings an unpleasant surprise – albeit in a slightly different vein.
If the older model perceived the changes made by the MorePowerTool utility, but the attempt to apply them was limited at the driver level, then here the SPPT edit is simply not displayed in the driver. And the point is not in the software part. As soon as the RX 5500 XT was replaced with the RX 5700 XT, the changes began to apply to the same MPT and driver versions.
However, it turned out to be possible for the test card to flash the BIOS from its press version, which can be considered a hint of the possibility of overclocking by modifying the standard BIOS.
Due to this, the GPU frequency limit was raised to 2200 MHz, but this did not lead to any result: the limit of stable operation for the test specimen was just 2000 MHz at 1.15 volts. Further acceleration rests simultaneously on the impossibility of raising the voltage above this mark, and on the inability of the GPU itself to operate at a reduced voltage.
As a result, in synthetic benchmarks we get a frequency of about 1940 MHz, and in games – most often 1960, although this value can fluctuate within 10-15 MHz in both directions.
Power consumption with such overclocking does not change much: 117 (119) watts at its peak. The GPU temperature rises to 72 degrees, and the fans practically do not change their speed: the maximum value is 1077 (1079) rpm.
Tests in games – Radeon RX 5500 XT, GTX 1650 Super, RX 570 and RX 580
Assassin’s Creed: Odyssey
DeusEx: Mankind Divided
Gta v
Horizon zero dawn
Metro: Exodus
Project CARS 2
Red dead redemption 2
Resident evil 3 remake
Star Wars: Jedi Fallen Order
The outer worlds
Total war: three kingdoms
The Witcher 3
War thunder
World of tanks
Conclusion
Since it was said at the beginning of the article that it is primarily interesting to compare the RX 5500 XT with its direct predecessors, then the conclusion should start with the most obvious things.
Both the RX 5500 XT and the GTX 1650 Super literally put an end to the solutions of previous generations.
Yes, this does not mean that if your computer is still running Radeon RX 580 or RX 590, then you should change them for new items: you will benefit from such an “upgrade” much less than you will spend money.
The same should be borne in mind if you are going to buy a video card on the secondary market. The RX 580 can be found at very interesting prices, and this option can be much more profitable than purchasing new cards in the store. Yes, here you have to look for a card with a preserved warranty and be as confident as possible in its history – and yet, it is possible.
But to buy the same RX 580, RX 590 – and GTX 1060 too! – in the form of a new product is absolutely meaningless. The RX 5500 XT is often faster and always much less power hungry than older Polaris models. The same can be said about the GTX 1650 Super, comparing it with the GTX 1060. Moreover, the performance of the latter, according to rumors, is now closer to the RX 570 than to the RX 580.
But the point is not even the best performance and power consumption, but the fact that for the RX 5500 XT and GTX 1650 Super today they ask for either the same or even less than for solutions of previous generations. And as for the purchase b About lshie money product that is in all respects worse – it is something from the field in which the author has come to an end censorship metaphors.
As for the comparison of budget new products from AMD and Nvidia – in the end we can say that the RX 5500 XT in the 4 GB version and the GTX 1650 Super … are simply the same!
Yes, in some games the red card is a little faster, in some it is green, but the difference is insignificant anyway, and there are no cases when one card demonstrates comfortable performance, while on the other it is impossible to play in principle.
There is not much difference in power consumption, and the relatively low performance of the GTX 1650 Super in overclocking is the result of the fact that this particular version of the card does not allow increasing the nominal limit by more than 100%. Had a different card been available, the results would have been completely different.
However, if you study the archive of the GTX 1650 Super bios on the same Techpowerup, you will find that only MSI GamingX and Asus ROG Strix have increased power consumption limits and the possibility of their further manual increase … and buying such expensive versions of an initially budget card is also stupid.
If we exclude the “top” options, the price difference between the RX 5500 XT and the GTX 1650 Super will also be small: yes, the Nvidia card has more budget versions – but there coolers are simpler, and the overclocking is, accordingly, worse: the frequencies of the Turing chips are in direct proportion to their actual temperatures.
And the more expensive options, in turn, are comparable in price to the 8 GB versions of the RX 5500 XT, but do not have an advantage in the amount of on-board memory. And, again, it is far from a fact that they will overclock much better than the test card.
Thus, when choosing between these cards, it is not worth asking yourself the question “Which is better: the RX 5500 XT or the GTX 1650 Super?” and the question, “How much do specific versions of the RX 5500 XT and GTX 1650 Super cost, and what do they offer for this money?” Considering any of these cards a priori the best option in their price segment is the shortest and surest way to buy the worst option.
Related posts:
Review & Test: Video card Gigabyte GeForce RTX 3060 Ti Gaming OC Pro 
AMD Radeon RX 5600 XT VS GTX 1660 Ti and RTX 2060: Review| Set-up| Hashrate 
Review of the KFA2 GeForce RTX 3060 Ti Core video card 
GIGABYTE GeForce RTX 3090 Gaming OC video card: Review| Specs | Set-up| 
INNO3D GeForce RTX 3090 iCHILL X4 video card review: Test | Config | Hashrate 
ROG Strix GeForce RTX 3070 OC Review: Test | Specs | Hashrate
