The Nvidia GeForce RTX 4080 launch started off contentiously with a wallet-screaming asking price of $2,219, a $1,000 increase over the company’s previous generation RTX 3080 series. It didn’t help that on paper, the RTX 4080 looked significantly pared back compared to the company’s flagship $3,299 RTX 4090 GPU.
However, after almost a month of testing out Nvidia’s Founders Edition RTX 4080 with various games on my 144Hz 4K monitor, I’m seeing this card in a different light. It’s a significant leap over the previous generation, especially when you factor in the addition of AI-generated frames with Nvidia’s new DLSS3 tech. Plus, it runs much quieter and cooler while consuming less power than the RTX 4090.
RTX 4080 Design
There’s no getting around it – the RTX 4080 is a big boy. In fact, it’s the exact same size as the gargantuan 4090.
While it’s unlikely to be a problem for those rocking a full tower case, anyone with a mid tower or less will need to make sure they have enough clearance to accommodate one. The RTX 4080 uses a three-slot design, which is a bump up over the two-slot design of the RTX 3080 series, and we’re already seeing third-party cards from the likes of Asus and Gigabyte approaching four slots.
While it’s slightly shorter in size than the RTX 3090, it is quite a bit thicker. The thickness of the card made the other PCIe slot located directly beneath it on my motherboard unusable due to the lack of space between the vacant slot and the GPU.
The benefit of retaining the large RTX 4090 shroud is that it uses the same cooling system and fans which should result in vastly improved thermals and greatly reduced noise levels.
The Founders Edition retains the slick black industrial look of previous cards from Nvidia and it looks great. It’s a refreshing departure from the typically garish designs of third-party card makers. It’s just a shame that Nvidia isn’t making the Founders Edition cards of the RTX 4080 available for sale in Australia.
The RTX 4080 uses a single 12-pin PCIe5 power connector and Nvidia includes an adapter in the box so you can connect it to your existing power supply. The nice thing is that the RTX 4080 has a maximum of 320 watts of power which is the same as the RTX 3080 and substantially less than the 450 watts of the RTX 3090Ti and 4090 cards. This means that you won’t need to upgrade your power supply so long as you’re using at least a 750-watt-rated PSU.
In terms of video outputs, you get three DisplayPort 1.4a and a single HDMI 2.1 port, all with Display Stream Compression (DSC) support. Those can handle up to 4K at 240 Hz or 8K at 60 Hz, which should be sufficient for the time being. That said, it’s a shame not to see DisplayPort 2.1 for future proofing with the next generation of displays that are promising 240Hz or higher refresh rates at 4K or 8K resolution without compression.
Cool, quiet and efficient
Usually, with a high-end air-cooled graphics card, you would expect some serious fan noise when running at full tilt but the RTX 4080 was barely audible inside my glass case.
The RTX 4080 FE completely turns off the fans during idling or light loads such as 4K decode. Meanwhile, under full load, the card wouldn’t exceed 34db according to my decibel meter which was barely audible inside an enclosed case. Additionally, the GPU temperature never exceeded 70 degrees even with some overclocking applied.
The results are unsurprising given that Nvidia is effectively using the same large cooler and fans as its flagship GPU on an RTX 4080 chip that is both smaller and has a lower transistor count but the results are impressive nonetheless.
It’s also worth noting that the RTX 4080 actually draws less power than the RTX 3080 with the card never exceeding 300w while gaming. The only time I could get it to hit 320w was after overclocking and pushing the GPU to 100 per cent load while running 3D Mark synthetic benchmarks.
Test machine
I tested the RTX 4080 in my dedicated gaming rig which includes the following components:
- AMD Ryzen 9 5900X 12-core/24 threads Max Freq 4.9GHz CPU
- Gigabyte B550 Vision D ATX AM4 Motherboard
- Seagate FireCuda 520 1TB M.2 PCIe Gen 4 SSD
- 32GB Corsair Vengeance RGB Pro DDR4-3600 CL16 RAM
- Corsair H100i RGB Platinum SE 240mm Liquid AIO CPU Cooler
- Corsair RM850x 850W Fully modular 80+ Gold Certified ATX PSU
- Corsair iCUE 4000X RGB ATX Mid Tower Case
- Gigabyte Aorus FV43U 4K 144Hz Monitor
I used a fresh installation of Windows 11 Pro together with the latest Game Ready Driver 526.98.
RTX 4080 Benchmarks
For benchmarking, I focused on testing games at 4K resolution at the highest possible graphics settings including maxed-out ray tracing where supported in titles. Even though the RTX 4080 is the second fastest graphics card right now, it hits CPU bottlenecks running at lower resolutions like 1440p and 1080p, leaving plenty of untapped performance on the table. This means that if you plan to game on resolutions lower than 4K, then you’re better off saving some cash and buying a 3070 or 3080 card. The 4080 is a card that simply demands a 4K high refresh rate display to really show what it’s capable of.
First off, let’s start with some synthetic benchmarks. While I don’t like to put too much weight into synthetic benchmarks, it does at least provide a broader picture of the performance leap from the 3080 series.
In 3D Mark’s Time Spy, the RTX 4080 achieved close to a 30% improvement over the 3080 and only 20% behind the RTX 4090. Moving across to Port Royal, which focuses on real-time ray tracing performance, the RTX 4080 achieves a whopping 45 per cent improvement over the 3080 while falling 30 per cent behind the RTX 4090.
| 3D Mark | RTX 4080 | RTX 3080 |
|---|---|---|
| Time Spy | 25,482 | 18,182 |
| Port Royal | 17,982 | 12,310 |
On the gaming front, expect to maintain above 60 frames per second (fps) on the vast majority of titles while pushing beyond 100 fps in games like God of War, and that’s before using Nvidia’s fantastic AI upscaling tech in DLSS. The only exceptions are Dying Light 2, Control and Cyberpunk 2077 which are three of the most graphically demanding titles available on PC. However, kicking on DLSS for any of these games gets you back up to 60 fps and beyond which is something you couldn’t achieve with any previous-generation card.
I generally prefer to play games with DLSS 2 enabled in quality mode if the game supports it as the image quality tends to look better than native 4K while also delivering a big boost to performance.
On average, there’s a 45 per cent increase in performance over the RTX 3080 in all the games that I tested below, amounting to a very impressive generational improvement.
Note that all the games listed below are running at 4K with all their graphical features set to the maximum including demanding effects such as ray tracing. The only exception is God of War which doesn’t support ray tracing. Cyberpunk‘s ray tracing was set to ‘psycho’ which is a notch above ultra and Forza Horizon 5 newly released in-game ray tracing feature was also enabled.
In Deathloop, all settings were set to ultra and all ray traced effects supported in the game – namely ambient occlusion and sun shadows – were set to ‘Quality’. Post-processing Anti Aliasing was set to ‘Nvidia DLAA’, which is the highest quality setting available.
| Game | Native 4K | DLSS 2 Quality | DLSS 2 Performance |
|---|---|---|---|
| Forza Horizon 5 | 98 | 106 | 110 |
| Control | 51 | 79 | 92 |
| Dying Light 2 | 28 | 72 | 94 |
| God of War | 105 | 132 | 158 |
| Shadow of the Tomb Raider | 85 | 128 | 149 |
| Deathloop | 92 | 112 | 122 |
The RTX 4080 has another trick up its sleeve with Nvidia’s third iteration of Deep Learning Super Sampling (DLSS). DLSS 3 takes the super resolution that’s available to all RTX GPUs and then adds Frame Generation to the mix which uses AI to generate a new frame, effectively doubling your frame rate without any performance penalty. As the frame generation feature leverages specifically optimised RT cores built into the GPU, DLSS 3 is exclusive to the 40-series GPUs.
The performance leap with DLSS 3 enabled is very impressive, doubling the frame rate from DLSS 2 pretty much across the board which you can see in the nine titles tested below. Again, all titles below were benchmarked with all their graphical settings switched to maximum including ray tracing.
| Game | Native 4K | DLSS2 Quality | DLSS 3 Quality | DLSS 3 Performance |
|---|---|---|---|---|
| Spider-Man: Miles Morales | 55 | 61 | 103 | 112 |
| Cyberpunk 2077 | 34 | 56 | 79 | 108 |
| A Plague Tale Requiem | 61 | 83 | 116 | 137 |
| Destroy All Humans 2: Reprobed | 94 | 98 | 143 | 178 |
| F.I.S.T: Forged in Shadow Torch | 79 | 113 | – | 186 |
| Loopmancer | 50 | 79 | 107 | 138 |
| Bright Memory Infinite | 68 | 75 | 110 | 127 |
| Microsoft Flight Simulator | 61 | 66 | 121 | 137 |
| 3D Mark (DLSS feature test) | 39 | – | 97 | 132 |
However, DLSS 3’s frame generation isn’t perfect and image quality can suffer depending on the game. For example, in a fast-moving title like Spider-Man: Miles Morales, I preferred to keep DLSS 3 in Quality mode as switching to the other performance modes produced a noisier image.
An example of this is in the menu select scene of Miles on the subway where the edges of the fur on the collar of his jacket had a distracting amount of noise in DLSS 3 Performance and Ultra Performance modes but looked acceptable when dropping it down to DLSS 3 Quality. Switching from DLSS 3 Quality to Performance also only netted a 9 fps boost to the average frame rate which made the decision easier.
Meanwhile, slower-moving titles such as A Plague Tale Requiem and Microsoft Flight Simulator looked fantastic in DLSS 3 Performance mode and being able to more than double the average frame rate over native 4K was transformative.
The same rings true in Cyberpunk 2077, going from 34fps to a whopping 108fps – more than a 217 per cent increase over native resolution.
The image quality and performance impacts of DLSS 3 vary between titles. Some games exhibit little to no visible noise in motion and around fine objects, so my recommendation is to experiment with DLSS 3’s quality and performance modes to see what looks good to you.
There is also a slight hit to input lag when enabling frame generation so it’s unlikely DLSS 3 will find its way to competitive online shooters anytime soon. However, in all the games that I tested, the slight increase in latency was imperceptible during gameplay and if you mainly game on PC using a controller, you’re unlikely to notice any difference at all.
GadgetGuy’s take
The RTX 4080 is the second-most powerful GPU on the market that comfortably outperforms 30-series GPUs while the inclusion of DLSS 3 widens the performance gap significantly. More importantly, it pulls off this impressive performance feat by running far cooler, quieter and consuming less power when compared to the previous generation of cards as well as Nvidia’s flagship RTX 4090. A card this powerful has no right to run whisper-quiet and at sub-70 degrees under full load but the RTX 4080 pulls it off with ease.
The RTX 4080 is overall 30 per cent slower than the RTX 4090 which is a wider performance delta than what we had previously with the RTX 3080 vs the 3090. However, the RTX 4080 costs $1,000 less than the RTX 4090, so you’re effectively getting a 30 per cent slower card that costs 45 per cent less which is a price-to-performance tradeoff that I suspect most would be willing to make.
Later this month, AMD is launching the 7900XTX, pitched as a direct competitor to the RTX 4080 and will hit a similar $2,000 local price point. While we will have to wait and see how team Red’s card performs versus the RTX 4080, early numbers released by AMD suggest that it won’t come close to matching RTX 4080’s ray tracing performance. It does, however, use a regular 3 x 8-pin plug so you won’t have to use an adapter for connecting to your existing power supply as well as DisplayPort 2.1 which is missing on the RTX 4080.
The post Nvidia GeForce RTX 4080 GPU – a cool and quiet 4K gaming monster (review) appeared first on GadgetGuy.
0 comments:
Post a Comment