Hardware is one of Gamer.no’s leading investments. After several years of experimentation, we have adjusted our coverage several times, and 2023 will be no exception.
During the next year, we will focus even more on gambling than we have done so far. There will be fewer individual tests, more feature material and more collective tests that will have a longer lifespan on our front.
Although we hardly have the resources to compete with players such as Digital Foundry or Gamers Nexus, we want to contribute to lifting the Norwegian coverage of game technology – which in practice is non-existent.
We are bringing in a new freelancer who will help the undersigned a bit with hardware, and we will write several articles about the performance of new games.
Our test rig has also received a proper overhaul, which we look at in more detail in this text.
It’s only been a year and a half since we last changed the test rig, but due to a component failing after a lightning strike, we just as easily replaced the whole machine.
We are taking a look at what will (hopefully) become a permanent fixture for testing video cards and new games here with us for the next three years.
READ ALSO: How many graphics mods do you need to make the RTX 4090 sweat in Skyrim?
Flagship from Intel
The heart of our test rig is an Intel Core i9-13900k. At the time of writing, it is the fastest gaming processor on the market.
AMD can soon take back the throne with its new X3D processors, but the difference is unlikely to be all the world.
Mikkjell Lønning/Gamer.no
The 13900K is a 24-core processor with 32 threads. We gave the processor 9 out of 10 possible points in our test last autumn, and he does a very good job of reducing the processor bottleneck when we test video cards.
We have – as usual – chosen to use the processor without overclocking it when testing. The aim is that the test results should be representative of as many people as possible, and the majority will not overclock such a processor. In any case, it would not have been relevant because of how hot the processor gets.
The previous 12900K is still in another computer. We’ll have to think about what nice articles we can use the extra machine for after a while.
New keel
Asus is responsible for several of the components in our rig this year, and has, among other things, supplied a Rog Strix LC II cooler with a 394 millimeter radiator.
If I were to buy something personally, I would go for the cooler I had in our previous rig (Asus Rog Ryujin II), but I’m leaving Ryujin in the other computer for a little experiment I’m running. In any case, the LC II is very capable. The cooler is the most efficient I’ve come across for keeping temperatures down, but makes significantly more noise than the Ryujin II.
Mikkjell Lønning/Gamer.no
At full blast, with over 2,000 revolutions per minute, the fan can produce up to 50 decibels. At the same time, the cooling performance is brilliant. It is also possible to turn down the fan speed and still get almost as good results.
Fortunately, the noise level does not have much to say in a test rig, where low temperatures are more important. We can always take noise measurements of video cards in games that require little of the processor.
The i9-13900k is by no means easy to cool under maximum load, but the LC II takes it easy. This is one of the best all-in-one coolers you can get your hands on.
Storage
Mikkjell Lønning/Gamer.no
The test rig is equipped with four PCIe SSDs. We have moved over two of the SSDs from the previous test rig, Kingston Fury Renegade 2TB, and combined them with a W_D Black SN850X 2TB and a Kioxia Exceria Pro 2TB.
The two newcomers are a bit more expensive than the Kingston SSDs. None of them manage their theoretical speeds in practice, but deliver read and write speeds of around 5 gigabytes per second.
Among other things, I will test all of these against the new generation of SSDs that just came on the market, but there will hardly be any big differences in games, even if the read and write speeds are doubled with the new generation.
Mikkjell Lønning/Gamer.no
SSDs in this class have no significant effect on game performance other than load time, and all of these perform identically in the games we have tested load time.
Of our four SSDs, it is the one from Kingston that gives the most value for money, but they are all excellent products.
Main card
From this year we have two different motherboards for our test rig, one Asus Rog Maximus Z790 Hero and one Gigabyte Aorus Z790 Elite AX.
Mikkjell Lønning/Gamer.no
The Asus card will be in the machine most of the time to test video cards when needed, but I have, among other things, planned an article about performance across price levels of main cards, and have therefore equipped ourselves with two cards this time.
The Gigabyte card has more than enough functionality for most builds, but the Asus card has, among other things, a nifty function that makes it easy to change video cards quickly, which is nice for us who change a few times a week.
The Z790 Hero offers more functions than I will ever use, but it is primarily support for the fifth generation of SSDs that makes me leave the card firmly in the rig. I’ll have to look into that more later.
RAM
In the last year, DDR5 memory has dropped in price at the same time as the speeds of the memory chips have skyrocketed.
Our previous test rig had 32 GB of Kingston Fury Renegade chips with a speed of 5200 MT/s. Those pieces have halved in price since then.
Mikkjell Lønning/Gamer.no
This time I have been sent Kingston Fury chips of 6400 MT/s (CL32), together with a set of G. Skill Trident Z5 DDR5-6800 chips (CL34).
The Kingston chips cost almost a thousand Swedish kroner less than G. Skill’s, and are generally a bit faster than G. Skill’s chips. But it is only a matter of a few percent, and the results vary from game to game.
Mikkjell Lønning/Gamer.no
The G. Skill chips will be in the test rig (primarily because I got these first and have tested a bunch of video cards on them), and the others in our extra machine.
RAM development will probably take a couple of years before we have enough timing for it to be worth spending money on the most expensive chips. For now, 6000 MT/s is more than enough for gaming.
You can even buy chips at 8000 MHz now, but the price level is so high that you almost pay one kroner per MHz.
Cabinet
As before, we have gone for a normal cabinet instead of an open test bench, as it gives more representative results for how most people use their machines.
Our case this time is an NZXT H7 Elite. This is a bit smaller than what I have used previously, but more space-efficient under the desk.
NZXT is known for costing a bit more due to its design style and piles of RGB lights, and this product is no exception.
Mikkjell Lønning/Gamer.no
Large parts of the cabinet are covered with glass. The downside to that is that you get less air in than with more traditional mesh in the front, but NZXT has managed to insert three 140mm fans in the front.
The fans make a lot of noise out of the box, but luckily that can be fixed completely by adjusting the fan speed down in the NZXT Cap software.
On average, we achieve around three degrees higher temperatures in this cabinet compared to our predecessor (Cooler Master Mastercase H500P), which was a colossus in comparison.
NZXT H7 Elite is otherwise absolutely brilliant to build in. Neat cabling is easy thanks to all the good solutions from NZXT.
Power supply
Computer components use more and more current, and this means that larger power supplies are needed than before.
Support for running two or more video cards in SLI is on the other hand, so 1200 watts is about the most you can expect to use if you overclock enthusiast components.
Here I have received the second generation of the Asus Rog Thor, which is a platinum-certified, modular power supply. This is the new version of the one I had in my previous rig.
Mikkjell Lønning/Gamer.no
Thor II is basically a power supply from Seasonic that Asus has tweaked a bit with RGB, an OLED screen and a more resilient design. The power supply is practically silent, and comes with an input for PCIe cables with 16 pins.
I’ve been using this for quite a while now, and can’t find anything negative to say about the product. It is right in the middle of nowhere for those who want the best of the best.
Display card
The test rig is basically designed to test video cards in particular, but the manufacturers have slightly different practices on how long they lend out video cards. This means that I have a lot lying around.
AMD has unfortunately been difficult when it comes to lending out cards, but I have the vast majority of products from Nvidia available.
For example, I will test one MSI RTX 4080 Suprim X next week, and this will probably get me on a long-term loan. It will probably be a nice addition to our fabric mix going forward.
I will write several articles about the performance of new games, somewhat in the same style as TotalBiscuit’s good, old “port report” videos. There have been quite a few such cases with us in recent years, and almost all of them have been good reading.
It’s something I want to do much more of. Me is primarily a website for games, and therefore our aim is to turn the hardware coverage towards games to the greatest extent possible.
In order to show the breadth of performance from different budget classes, without spending too much time on testing, I will probably test the new games with two to four graphics cards, depending on the type of game.
Playtesting
It is not easy to find a representative selection of games to test video cards.
Both AMD, Intel and Nvidia are interested in making exclusive agreements with as many developers as possible. This means that the game gets specific settings and functions that favor one of the manufacturers.
A good example of this is how games draw hair on humans and animals, where Nvidia and AMD have their own technologies that perform clearly best with their own video cards. So Nvidia HairWorks and AMD TressFX.
Sometimes this type of function can be turned off, other times not. I mostly like to run games at a preset level such as “ultra” to compare cards, unless there are gimmick settings that give far higher performance to one manufacturer.
Ray tracing has basically been for Nvidia, but AMD has taken the technology into the heat in recent years, so now I’m testing ray tracing with both.
At the same time, I have tried to combine both slightly older and newer games in different genres.
We have then selected 15 games in which all graphics cards are tested:
- Assassin’s Creed Valhalla
- Call of Duty: Modern Warfare II (2022)
- Counter-Strike: Global Offensive
- Cyberpunk 2077 (with and without ray tracing)
- F1 22
- Far Cry 6
- Forza Horizon 5 (with ray tracing)
- Marvel’s Guardians of the Galaxy
- Horizon Zero Dawn
- Metro Exodus Enhanced Edition (with ray tracing)
- Middle-earth: Shadow of War
- Overwatch 2
- Rainbow Six Siege
- Red Dead Redemption 2
- Shadow of the Tomb Raider
Synthetic tests stop spending time on. It basically has little value when we are already testing games.
Each game is tested in both 1080p, 1440p and 2160p, and we run each resolution five times and calculate the average from that.
More games will probably appear on the list, but we won’t include completely new games until we know that there won’t be anything special with updates. Then don’t let me re-test older video cards so often. Benchmarking is very time-consuming (and terribly messy) work.
The entire component list in the new rig will be as follows:
