Radeon RX 5500 XT 8gb review and comparison with GTX 1660 and GTX 1660 Super: Specs| Testing| Configuration| Set up| Video Cards | Overclocking| Benchmark. Find out which one of these Video cards is better for you.
In the previous comparative review , the Radeon RX 5500 XT in the younger version, equipped with 4 gigabytes of on-board memory, was already reviewed. The card performed well, becoming an excellent replacement for the already outdated Polaris, and an equal competitor for the GeForce GTX 1650 Super.
However, at the same time, the difference between the RX 5500 XT and the next in seniority, the RX 5600 XT , was also clear . The Navi 14-based card clearly did not pretend to play a bigger role than just a solution for games in FullHD with compromise graphics settings.
Nvidia, meanwhile, has three clearly ranked graphics card models between the GTX 1650 Super and the RTX 2060. And, if the GTX 1660 Ti, due to its current cost, can hardly be called an attractive solution, then the GTX 1660 and GTX 1660 Super do not deserve such a categorical assessment.
AMD in this segment offers only one more version of the RX 5500 XT, which has a doubled memory capacity.
Moreover, it is almost a tradition for AMD video cards – only in the amount of memory the two versions of RX 5500 XT differ from each other. The GPU configuration is completely identical, the frequency model, if you do not take into account the vendor overclocking, is also the same.
As you might guess, the prices do not coincide.
Meet the participants
So, for today’s testing, we took three video cards released by the same vendor and belonging to the same line, namely, MSI Gaming X.
If we compare how much these cards cost at the time of publication of the review, it turns out that the 8 GB RX 5500 XT and GTX 1660 are asking for literally the same amount: 21 thousand rubles. GTX 1660 Super will cost a little more – 25 thousand.
And again, do not think that this ratio is typical only for MSI Gaming X. If you compare all the cards on sale at the time of the review, it turns out that both the RX 5500 XT 8gb and the GTX 1660 will cost you from 18 to 23 thousand …
GTX 1660 Super can be found from 20 and a half thousand for very sad versions with one fan (so it’s better to count from 22 thousand), and up to all the same 25 thousand for top-end options.
Of course, the prices above are given for the shops of the CSN of St. Petersburg, and are relevant as of October 29, 2020 . You can find out the cost of test cards in stores in your region at the time of reading the article below:
One way or another, under the indicated conditions, the RX 5500 XT is in a deliberately disadvantageous position for itself. But maybe full-fledged overclocking of the MSI version will strengthen its position relative to the GTX 1660?
This is what we will check today.
Option of the article for those who cannot read
RX 5500 XT and PCI-e version
Before moving on to the review of the test card, it is worth mentioning once again the nuances concerning the model itself. Almost any discussion regarding the RX 5500 XT will certainly touch upon the effect of PCI-e bandwidth on card 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 difference in performance between the 8- and 4-gigabyte version of the card can really take place, but it has absolutely nothing to do with the version of the interface. And it manifests itself in the same way when using both PCI-e 4.0 and PCI-e 3.0.
If you pay attention to the screenshot below, you will notice that in Horizon Zero Dawn at high graphics settings in FullHD, the consumption of video memory exceeds 4 gigabytes.
In this case, RX 5500 XT 4gb and GTX 1650 Super are forced to use part of the RAM – this is clearly noticeable by the difference in the amount of RAM used. RX 5500 XT 8gb uses only its own buffer.
At the same time, the RX 5500 XT 8gb is noticeably faster than both its 4 GB counterpart, operating at almost similar frequencies, and the GTX 1650 Super, whose performance is close to the 4 GB RX 5500 XT.
Why is this happening? Very simple. It’s not about the version of the interface, but about the amount of memory consumed. While it does not exceed 4 gigabytes, there is no difference between the two versions of the RX 5500 XT and cannot be.
If the card is forced to use part of the RAM – alas, the speed of even overclocked DDR4 is significantly lower than that of gddr6. Which affects the performance of the video card. Even on the performance of the GTX 1650 Super, which in principle does not support PCI-e 4.0, and you will not hear about the influence of the interface version on it.
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:
MSI RX 5500 XT Gaming X
Like other test participants, the RX 5500 XT was released by MSI and belongs to the flagship line. More budgetary versions are presented in the MECH line.
But, in spite of the fact that this card can hardly be called a budget solution, there is again nothing to talk about packaging and configuration: the card comes with only documentation.
Appearance and design
But like Sapphire, MSI decided to put all the resources into the design of the card itself.
On the front side, you can see a monolithic casing of the 7th generation Twinfrozr cooling system and two low-profile proprietary Torx 3.0 fans with characteristic two types of blades. The test card is made in the same design for the entire RX 5000 line – although, of course, it will be structurally different from the flagship RX 5700 XT and other Navi 10 models.
The back of the card is covered with a backplate, and here it directly participates in cooling the card – and, as the further analysis of the design will show, not only due to the ventilation slots.
The BIOS switch is no longer here – for the older version of the card there is only one operating mode. But the card is powered by MSI from the same 8-pin connector as Sapphire Pulse.
The set of interface connectors is also similar to the previous RX 5500 XT: three DisplayPort and one HDMI. But DVI is absent, which is a disadvantage for a budget solution. An adapter to the “outdated” interface will also have to be purchased separately.
Cooling system and PCB
Although outwardly the RX 5500 XT in the Gaming X version does not differ much from the older cards in the same design, it is quite obvious that the design of its cooling system and the board will be different from that of the Navi 10 models. So it will be interesting to disassemble it too.
A quick-detachable fan mount is not used here, remaining a Sapphire trademark. Thus, cleaning and maintenance is not possible here without a complete parsing of the card.
And he, in turn, is impossible without damaging the warranty seals.
Only the backplate can be removed while maintaining the warranty. It is attached here with 8 screws for a Phillips screwdriver, six of which fix the stiffener, and two more – the VRM heat sink. By unscrewing the screws and removing the backplate, you can find that it serves not only to give the card additional rigidity, but also to remove heat from the board in the areas of memory chips and power elements.
But in order to remove the radiator with the fan shroud, you will have to unscrew 4 more screws, one of which is covered with a warranty seal.
Both parts are removed as a single block, and only a stiffener remains on the board, which performs exclusively the function of reinforcement, and does not participate in cooling in any way.
As in the case of Sapphire Pulse, VRM is implemented here according to the usual scheme, where drivers and mosfets are separate elements. But the manufacturer cannot be blamed for excessive savings either: the VRM has 7 power phases, and 6 of them are allocated for the GPU. Again, this is a very large headroom for the Navi 14 with its modest power consumption.
The radiator of the cooling system, which is logical, differs from the unit used in the construction of maps on Navi 10. Nevertheless, by the standards of the RX 5500 XT, this is almost a top solution.
The principle of its structure is almost similar to the radiator from Sapphire Pulse. The heat from the GPU is removed by a nickel-plated copper plate (which some people still confuse with aluminum). A separate heat sink is also provided for memory chips, which is in contact with the radiator fins. And, of course, the heat from the VRM card is also removed by the main radiator.
But there are already four heat pipes in the radiator design. You might think that there are five of them, but in fact, one of the central tubes is S-shaped, and carries heat from the GPU to both sides of the radiator. All tubes have a diameter of 6 mm and are soldered to both the heat sink and the fins themselves.
Aside from more pipes, the heatsink itself is larger than the Sapphire Pulse. Moreover, both the body of the radiator and the number of ribs in it are larger. But the aerodynamic optimizations are again a minimum: these include only the protrusions on the sides of the ribs, which obviously serve to more evenly scatter the air flow.
The fans are manufactured by Power Logic, model PLD09210S12HH. These are low-profile turntables with a blade span of 90 mm. Their distinctive feature is the presence of two types of blades with different angles of attack. The main airflow is provided by knurled blades, while the second type is located at a steep angle and is designed to create more static pressure. Which, by the way, is very useful, given the rather tightly dialed radiator and the height of the turntables of 10 mm.
The obvious disadvantage of the turntables is the use of a sleeve bearing, which is not particularly durable.
Installation in the case
Due to a more massive cooling system than the Sapphire Pulse, this version of the RX 5500 XT turns out to be slightly more overall:
However, it is only about 250 millimeters long, which is only slightly more than the width of a standard ATX or mATX motherboard. Naturally, there can be no talk of any conflicts between the video card and the hard drive cage, even if there is one in your case.
But in thickness, the card occupies more than two slots. Naturally, as well as on other Navi, Crossfire is not supported here, but possible difficulties with installing expansion cards should be borne in mind.
Rivals
MSI Gaming X, like any expensive version of the initially budget cards, falls into an obvious price trap. It cannot be cheap: not in terms of status, and the cost is higher than that of those options where three and a half power phases are adjacent to an aluminum bar as a radiator.
This is natural, but it is also natural that users with a limited budget will choose cheaper options for cards based on the same GPU, and those who have the opportunity to pay extra will rather pay extra for a more powerful chip than for a cooler and power phases.
Thus, the current version of the RX 5500 XT becomes a kind of “thing in itself”. And therefore – and you need to compare it with direct analogs:
In other words, GeForce GTX 1660 and GTX 1660 Super were chosen as competitors in the same versions of Gaming X.
Test stand and testing methodology
The test bench configuration and testing methodology have not changed in any way from previous materials about the RX 5500 XT.
- CPU: AMD Ryzen 9 3900X
- CPU cooling system: ID-Cooling SE-224-XT Basic;
- Thermo 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;
- Housing: Corsair Carbide 270R;
- Power supply: Cougar GX-F 750.
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. Naturally, the frequency of the Infinity Fabric bus was also raised to 1900 MHz.
For games, the FullHD resolution was chosen, the graphics settings fully corresponded to those used in the review of the 4 GB version of the 5500 XT.
All test cards worked exclusively in PCI-express 3.0 mode. Nvidia cards – in view of the fact that they do not support the fourth version of the interface at all, and for the RX 5500 XT this mode was set manually.
Frequency model and overclocking …
…RX 5500 XT
A separate article has already been published about the frequencies and overclocking of the test RX 5500 XT , which can be found here.
Or you can see its visual version:
The card only has one BIOS option, offering a 130 watt power limit. The memory frequency is 1750 MHz, the GPU frequency is limited to 1880 MHz, but in game mode it is kept within 1860 MHz, which is only slightly higher than that of Sapphire Pulse on the “client” firmware version.
The peak power consumption of the GPU (according to HWinfo64) when passing the Superposition test is 104 watts, the peak temperature is 70 degrees, and the fans spin up only up to 1163 rpm, and there is no reason to expect any noticeable noise from low-profile turntables.
In overclocking with MorePowerTool, we managed to bypass the standard limitations and get a stable 2 GHz for the GPU and 2046 MHz for the memory.
As the voltage rises to 1.23 volts, the peak power consumption rises to 130 watts. The GPU temperature rises only up to 71 degrees, but at the cost of a significant increase in the cooler speed. The maximum value is 2079 rpm, while the noise from the cooler of the video card begins to clearly stand out starting from 1700 rpm. The sound cannot be called annoying, but it stands out clearly against the background of the system unit.
…GTX 1660
For the test GTX 1660, a completely standard frequency model is assigned: 1530 MHz in the base and up to 1860 MHz in dynamic overclocking.
But, like all modern Nvidia cards, it overclocks beyond these parameters thanks to GPU Boost technology – and the final frequency depends on the cooling efficiency of the chip.
At the beginning of the test, the card operates at 2010 MHz, but immediately slows down to 1995-1980 MHz, and as it heats up to 63 degrees – which in this case is the fan threshold – the frequency drops to 1965 MHz, and remains at this level at throughout the Superposition test.
Peak power consumption is 112.9 watts, but the cooler spins up to 1349 rpm, so the final GPU temperature is only 65 degrees. At the same time, the noise is recorded by the device, but remains indistinguishable by ear.
Memory of the gddr5 standard does not overclock with such impressive numbers as gddr6. +635 MHz gives a real frequency of 2318, and an effective frequency of 9272 MHz. But the GPU managed to add 140 MHz to the base frequency, which gave an actual frequency of 2100 MHz. Not a record for Turing, but quite worthy.
Peak power consumption has not increased much, especially since the limit can be increased by only 7 percent. The result is 123.6 watts, the temperature rises to 67 degrees and the cooler rpm – up to 1519 per minute. And it’s still very quiet, the fans start making real noise after 1700 rpm.
…GTX 1660 Super
The GTX 1660 Super, despite being a better performance solution, actually relies more on gddr6 memory. At the same time, the GPU frequency is lower than that of the usual 1660: with the base 1530, dynamic overclocking is available only up to 1750 MHz, and the real GPU frequency decreases from 1935 MHz at the start of the test to 1905 when warming up above 63 degrees.
At the same time, power consumption is, oddly enough, higher: 130.9 watts at peak, the temperature on the GPU is 69 degrees, the fans spin up to 1461 rpm – which, however, is still quiet.
We managed to raise the memory frequency by 1280 MHz during overclocking. The final real frequency is 2070 MHz, the effective frequency is 16 560 MHz. But the GPU frequency was increased only by the same 140 MHz, which in reality resulted in 2055-2040 MHz under load. At the same time, it was obvious that 2055 MHz is the ceiling for this GPU, since a further increase in frequency led to a loss of stability even when running diagnostic utilities (CPU-Z hung at the stage of collecting graphics data), not to mention tests.
Peak overclocking power consumption is 140.7 watts. The peak GPU temperature is 71 degrees, the cooler spins up to 1594 rpm. The card becomes faintly audible against the background of the system unit, but nothing more.
For convenience, we will summarize all the data on the performance and parameters of the cards in visual tables:
Tests in games
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
The Witcher 3
Total War: Three Kingdoms
War Thunder
World of Tanks
Conclusion
The conclusion from today’s review, in fact, could have been predicted at the very beginning: the RX 5500 XT in the version with 8 gigabytes of on-board memory is primarily driven by its price.
Its younger brother with 4 gigabytes can cost as GTX 1650 Super or slightly more, since it is similar in performance or slightly better. But to fully compete with the GTX 1660, you need to offer something more than just 8 gigabytes of memory.
Yes, full overclocking of the RX 5500 XT allows you to show the same or slightly higher performance, which is shown by the GTX 1660 in its nominal mode. Even if we are talking about the top-end version of the 1660, which stably operates at a frequency of 1965 MHz on the chip. But the 1660 can also be overclocked, and the gain from overclocking is no less noticeable than that of an AMD card!
Of course, even when overclocked, the GTX 1660 does not allow us to talk about a transition to a new level compared to the RX 5500 XT. The difference here is purely psychological, and in fact, you will still be playing in the same screen resolutions and with almost the same graphics settings. But if the price is the same, then why should you give money for a product, which, albeit conditionally, is still slower?
All this, however, does not make the RX 5500 XT in the 8GB version an unambiguously bad card. This necessitates a careful comparison of the prices and characteristics of the cards. If, other things being equal, the RX 5500 XT in your city turns out to be noticeably cheaper, and you can invest this money in other hardware with a greater benefit for performance, functionality or further upgrade, you can consider it.
But, if your situation is not the same as described above, the GTX 1660 will be the preferred purchase.
Of course, like any other budget card, you shouldn’t buy it in the most expensive versions. 21 thousand for the same MSI Gaming X is too much in the presence of adequate versions of the GTX 1660 Super for 22 thousand. But 17-18 thousand is already quite justified.
Exactly the same can be said about the GTX 1660 Super itself. Yes, it is noticeably faster than the other two participants, but it’s still a FullHD solution – just with higher graphics settings. QuadHD or ultra-high settings with anti-aliasing at maximum – obviously not its level.
And if so, the closer the price of the GTX 1660 Super gets to the price of the Radeon RX 5600 XT and GeForce RTX 2060, the less sense it makes in buying this card. And in this case, even 4 thousand rubles cannot be called a significant difference: the performance of the older cards will more than justify such an additional payment.
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
