grafic card

Just for the sake of time, let’s remember when the previous generations of Nvidia mobile graphics were released. The highest Ampere model presented by the RTX 3080 Mobile went on the market in January 2021. The graphics carried 6144 CUDA cores and has 8 GB of GDDR6 memory. Exactly one year later, Nvidia launched an improved version on the GA103 core. It is the current, most powerful mobile graphics RTX 3080 Ti (Mobile). It carries 7424 CUDA cores with 16 GB of GDDR6 memory.

Now we are a year later again, and Nvidia is introducing the successor to the RTX 3080 Ti, but also lower models. On this day, five GPUs intended for mobile computers officially go on the market. As the figure shows, it starts from the RTX 4090 as the most powerful model and ends with the RTX 4050 at the tail of the performance offer.

The position of the GeForce RTX 3080 Ti is taken over by the RTX 4080. It has the same number of CUDA cores (7424), but we have a narrower memory bus (192 bits versus 256 bits) and there is also a smaller memory capacity of 12 GB versus 16 GB for the older version. In terms of performance, it will probably be very similar, but this does not apply to games with DLSS 3. We will get to that later.

The flagship is of course the RTX 4090, its specifications are truly monstrous. It offers 9728 CUDA cores, 16 GB of GDDR6 memory and a frequency of up to 2 GHz. Compared to the mobile RTX 4080, the increase in computing cores is more than 30 percent, which could theoretically mean significantly higher performance. Of course, we know that the limit for similarly powerful GPUs is the processor, only tests will show the reality.

The latest PCIe 5 SSDs to hit the market this year will feature the new Phison E26 chip. And since they won’t be popular, it’s not out of place to think about whether it even makes sense to think about the purchase of brand new SSDs. Not quite, according to tests conducted by YouTube channel Compusemble.

Behind everything is Microsoft DirectStorage 1.1, a technology that, among other things, can reduce recording times by up to 40%. This API works on the principle of a set of compression and decompression algorithms entering the communication between the SSD and the GPU, allowing more data to be transferred than under normal circumstances. This process is also less CPU intensive when processing requests from NVMe, allowing for faster and more efficient transfer of data from storage to the graphics subsystem.

At the same time, it is not just Microsoft’s marketing rhetoric, DirectStorage in practice increases transfer speeds really significantly. Even in the case of SATA SSDs. The question therefore arises as to whether this effect becomes even more pronounced if DirectStorage is applied in a comparison test of SSDs for different generations of the PCI Express interface. And it turned out that in the case of PCIe 5.0 drives, not even the Microsoft panacea can do much here.

Square Enix’s Forspoken may not be a good game, but it’s the first to support DirectStorage, so it was used for the test. Because of this, the recording times are already reduced enough that even if you combine the recording times from seven different levels/scenes in the built-in benchmark, it makes the difference between an advanced SSD controller with a Phison E26 for PCIe Gen5 x4 and a two-year-old Adata XPG SX8200 Pro for PCIe Gen3 x4 barely perceptible three seconds.

In numbers, this means that a PCIe Gen3 SSD is 32% slower than a PCIe Gen5 SSD, but in reality you won’t even notice the difference. Not to mention the effect on FPS. At the same time, the throughput of unprocessed data with PCIe 5.0 x4 is roughly 4x higher than with the PCIe 3.0 x4 interface. It means that DirectStorage not only works, but that it even blurs the difference between SSDs for different generations of PCI Express interfaces.

Let’s add that choosing the latest PCIe interface is a logical and correct choice for building a new PC. In addition, there is only one game on the market that can use DirectStorage. In any case, let’s hope for the widest possible implementation of this technology in newly released games.

Conclusion

Aorus Radeon RX 7900 XT Elite received a large and very powerful cooler from the manufacturer. Paradoxically, the price in stores does not differ from the previously tested and modestly conceived Pulse Radeon RX 7900 XTX from Sapphire with a half cooler without LEDs. The manufacturer used the huge reserve of the cooler to significantly increase the cycle times and thus the performance. Unfortunately, the consumption of the card has also increased with this, and the result is that in terms of noise, the card with a top cooler is practically no different from the basic, minimalistic, inexpensive model from Sapphire.

It’s not really clear to me why Gigabyte didn’t make better use of having two BIOSes on the card. Both have the same power limit set at a high 406W, so the cooler has to spin something to keep the chip temps at the regulator’s liking. With Radeons, it’s even worse because the speed of the fans is controlled by the hottest spot on the chip, the temperature of which is usually around 90 °C. Even though the average temperature of the chip is at a low 62°C, the fans run like hell due to the high temperature of the hotspot.

The consequence is that for a card with an extreme cooler you can (by default driver settings) not choose between silent mode and powerful OC mode, but between two practically equally powerful BIOSes, one of which is only slightly better than average in terms of noise and the other already extremely noisy.

One can surmise that the motivation was to make the card stand out in performance tests in both modes, but from a user perspective it would certainly be better to be able to switch between modes with a quiet 350W card that only does 62 fps in games instead of 65 fps and a noisy and roaring 400W beast for those who need the three extra frames for a better feeling.

The consumption then increased almost to the level of the RTX 4090. The Radeon is only 15-25 W below it, but the same cannot be said about the performance. Where cards are not held back by the processor, the RTX 4090’s advantage is not much smaller than against 350W models. If you want an answer to why Radeon RX 7900 clocks aren’t higher, I wouldn’t look for it so much as some bugs in the chip and problems with reaching frequencies over 3 GHz, but rather that with higher clocks, the power consumption jumps up by tens of watts, while the performance increases by percentage units.

A great choice for Radeon fans

Radeons are mostly bought by gamers who prefer a better price/performance ratio, with the understanding that they will have to take care of their card a bit more than when buying a GeForce, and they have no problem with that. Someone wants to play and not worry about anything around them, and someone else likes to play with the card itself. And it is for such individuals that this card is ideal.

In the default state, due to the extreme consumption, it is rather average in terms of operational characteristics, and even in silent mode it does not give the impression of a top product – next to the 320-340W GeForce RTX 4080, it is at least not enthusiastic about the same large coolers, in terms of operational characteristics, it is at least a class worse.

But the surcharge is not high compared to the basic RX 7900 XTX models, it is less than is usual for premium models and, paradoxically, it is now the cheapest RX 7900 XTX with the second model from the same manufacturer.

At the same time, it offers a huge and powerful cooler with a large reserve, a significantly increased power limit to 406 W (with the possibility of increasing it by another 15%), and thus also gives a lot of room for manipulation both for clocking upwards and for optimizing consumption downwards. Basically, you will get a card whose noise will not excite you, but you just need to fiddle around a bit in the control panel and with your own help you will already get a result that suits a card with a similarly monstrous cooler.

And I still have to mention the still unresolved problem with card consumption on the desktop in less common combinations of resolution and refresh rate or on different combinations of monitors. At 120 Hz at 4K, the memory still runs at maximum 3D clocks and the power consumption of the card itself is around 90 W, at 144 Hz at 900 MHz and the power consumption is around 65 W. But AMD promised a fix a month ago, so hopefully we will see it soon.

The Aorus Radeon RX 7900 XTX Elite 24G is currently one of the cheapest RX 7900 XTX available here. At the same time, he got an excellent cooler with a large reserve. In terms of equipment and the price you pay for it, it is therefore extremely advantageous. However, the operational characteristics suffered because Gigabyte drove the clocks too high and the result is a rather average noisy card with high consumption, so due to the worse operational characteristics we will only give it a silver award. But it offers a lot of manipulation space for further tuning both towards higher performance and towards quieter cooling and higher efficiency, so you can win with it and work towards better results.

Conclusion

The crown went up again, so according to the current exchange rate (even if I take the pessimistic one), the price of the reference RX 7900 XTX comes out to some 27,300 crowns. Pulse is against it in our country currently about two thousand more expensiveand with such an additional fee, it already has a worse price/performance ratio than the reference RX 7900 XT starting at twenty-five thousand.

If I put a reference model against the Pulse RX 7900 XTX at a reference price, I’d say that paying an extra two thousand for a cooler to get an average card isn’t worth it. However, in contrast to the reference version, there should be no risk of problems with overheating with the classic heatpipes of the Pulse due to a defective vapor chamber, and the comparison is still rather hypothetical, because the models with the reference cooler are currently even more expensive in our country. The unavailability of the RX 7900 XTX drove the price higher than it should have been.

It can be assumed that with better availability, the price could drop to the recommended price, but at the same time, there is also the risk that we will not see a drop in prices. After the initial shock of the high prices of the new generation of cards, players have probably come to terms with them and are starting to buy in bulk. Cards are disappearing from stores faster than manufacturers can supply them.

On the other hand, you can get the competitive RTX 4080 for prices corresponding to the recommended ones, i.e. for 32,800 crowns. Even against Radeon with an additional charge, the RTX 4080 still fares worse in terms of price/performance ratio. This of course applies in the ideal scenario for Radeons – i.e. with classic rasterization or if you disable ray tracing and avoid DLSS 2 and DLSS 3. You will pay for it with higher consumption, compared to the RTX 4080 at the standard limit, the load power consumption is about 50 W higher. against the overclocked non-reference models, there is still a difference of around 30 W against the RX 7900 XTX.

If you don’t avoid DLSS 2 and 3, the RTX 4080 already has the upper hand even at a higher price. The older version of DLSS 2.x is supported by several times more games than FSR 2.x, and the first few games already support the generation of DLSS 3 images, for which AMD does not yet have an answer. With active ray tracing, GeForce gains the upper hand the more demanding effects the game uses. It can be assumed that the difficulty of the effects will increase rather than decrease, so the difference will increase over time. The RTX 4080 also often has larger coolers even with lower consumption, so they are better off with operational characteristics.

Sapphire Pulse AMD Radeon RX 7900 XTX, 24 GB

If I had to briefly characterize the card in the Pulse design, it is a model offering the performance of overclocked models and small reserves of the cooler for clocking with average operating characteristics at an above-average price. Overall, it currently belongs to the most advantageous cards with a price tag below or around thirty thousand.

But it’s not for fans of really quiet cooling. The cooler does not even have large reserves for clocking with increasing voltage and thus consumption (average temperatures are not bad, but the temperature of the hotspot rises quickly and the fan speeds also increase significantly). It is known that the regulation setting in combination with the radiator throttles the performance. However, the drop in performance is not so great compared to overclocked models with a larger cooler that it is worth paying a few thousand extra just for it.

Also remember that the Pulse will want a well ventilated case. In the test set installed in the Arc XL big tower with seven fans, the cooler usually balanced on the noise limit for the fans (between 1800 and 1850 rpm), with intensive load the speed even went higher. If you click on the first chapter of the performance test with measurements from the F1 22 without ray tracing, you can see in the first half of the graph with the fan speed that it stays at this limit, but in the second part both the speed and the noise are already increasing. After turning on ray tracing, the speed already goes over 2000 rpm, and it’s even worse in smaller cases. Even with more powerful profiles in the drivers, the noise grew beyond the tolerable limit. I hope to get to the details in a separate article.

And I still have to mention the still unresolved problem with card consumption on the desktop in less common combinations of resolution and refresh rate or on different combinations of monitors. At 120 Hz at 4K, the memory still runs at maximum 3D clocks and the power consumption of the card itself is around 90 W, at 144 Hz at 900 MHz and the power consumption is around 65 W. But AMD promised a fix a month ago, so hopefully we will see it soon.

The Pulse RX 7900 XT from Sapphire is one of the cheapest RX 7900 XTX models. At the same time, it offers solid operating characteristics, does not suffer too much from the disadvantages of the cheapest models, the noise level of the card can be described as average. Therefore, for the mix of an attractive price/performance ratio and average operating characteristics, it receives a silver award, with the condition that, as availability improves, the price will drop close to the recommended 27,300 crowns. If, even with better availability, the surcharge would be two thousand against the cheapest models, it would rather be for bronze.

We thank the online store for lending the card for the test CZC.cz

Why have your own VPN?

If one has one’s own public IP address, one reason for one’s own VPN may be to save money. Why pay commercial fees when I need to connect from other locations so that I can see on the network that I am in the Czech Republic.

However, even if I don’t need to deal with this situation, it has advantages in terms of safety and possibly convenience.

• Security is especially important in a situation where we need to log in from an unsecured Wi-Fi network to services with high security requirements. I would behave the same way in case of movement in any other network where I do not trust the security. In everyday life, it is most often internet banking.
• From the point of view of convenience and also security, it is about secured access to your home, to your own data at any time, when needed. Then we don’t have to worry that we don’t have the exact file we need on our mobile phone or laptop, but it’s enough that it’s available in the home storage.

In case we have a public Ip address, we can have a VPN installed at home. However, a VPS service can also be used.

I will look at the PiVPN project, a program that solves simplified installation and configuration for OpenVPN and WireGuard VPN programs. It is primarily intended for the Raspberry PI microcomputer, but it also works on Ubuntu and Debian distributions. The project website can be found at address pivpn.io.

They can also be downloaded directly from the command line. For this we can use this command which also starts the program.

curl -L https://install.pivpn.io | bash

And we get to the actual installation and configuration of the version with WireGuard.

Time and time again, Unreal Engine 5.1 shows that the time of photorealistic game backgrounds is not far away. Let’s take a look and walk through a sunny oasis in the desert, where the water surface glistens and palm trees gently sway in the wind. An ideal environment for quiet contemplation in the evening, sipping your favorite drink and simply being. At least for that one evening, right? However, to enjoy it properly, we will need at least an RTX 2080 at a resolution of 1920 × 1080, for higher RTX 3080 is recommended. RTX 4080 and 4090 owners can rest easy and continue sipping their daiquiri.

My undervolted RTX 2080 sweats blood in FHD, especially at maximum quality. Reducing the TSR (Temporal Super Resolution, similar to FSR or DLSS) to “balanced” will increase the images to values ​​around 30 to 40. However, to admire it, it is enough to walk around the lagoon, admire the palm trees, and on top of that, you can also record the night view of the oasis. There you will find a nice bonfire where you can think about upgrading the graphics. The demo is available for download here or directly below the video above. In addition to caressing the soul, it is also a solid test of whether your overclocked graphics will be stable.