It’s time to evaluate the brand-new GeForce RTX 4080, Nvidia’s most current $1,200 GPU, which is a substantial 71% increase over the RTX 3080’s MSRP, though obviously Nvidia would certainly like we fail to remember that was ever before a point and also instead concentrate on the $1,200 3080 Ti.
Given, the marketplace has actually transformed a lot ever since, so it will be interesting to see what the RTX 4080 needs to provide. There’s plenty to go over as well as benchmarks to look into, but initially allow’s touch on the requirements, take a look at the Nvidia Founders Version design, and after that we’ll get to heaven bar graphs. In time for this evaluation we likewise received partner models from PNY, Asus, and MSI, which we’ll display in the direction of the end of the testimonial.
The GeForce RTX 4080 is based upon the AD103 silicon which measures 378.6 mm2, that’s a 38% smaller than the AD102 used by the RTX 4090 which sees a 40% reduction in transistors. The 4080 still loads 62% even more transistors than the previous-gen 3090 Ti flagship though. Offered those figures it ought to be no surprise to discover that the 4080 packs 41% less CUDA cores than the 4090, the memory bus size has actually been lowered by 33% and there’s 11% much less L2 cache.
When it comes to core clocks, there’s little distinction between the RTX 4090 and 4080, we’re talking 2505 MHz versus 2520 MHz for the boost clocks, with a small advantage to the 4090. The RTX 4080 has been granted much faster 22.4 Gbps GDDR6X memory, a 7% boost there from the series front runner, though the smaller sized 256-bit large memory bus implies that total memory transmission capacity has actually been cut down by 29% to 717 GB/s.
Nvidia declares a complete graphics power rating of 320 watts for the RTX 4080. That’s the same ranking offered to the 3080, though the maximum GPU temperature level has actually been reduced from 93c to 90c. The minimum power supply demand is 750 watts and also for testing we’ll be utilizing an 850w system.
In regards to design, the Founders Edition RTX 4080 is identical to the 4090 on the outside, the only distinction we can represent thus far is the weight, the 4080 FE version is 3% lighter. Undoubtedly there are some physical adjustments under the skin, we visualize power distribution has actually been reduced, the memory arrangement has actually changed, as well as the GPU die is smaller. It’s additionally possible Nvidia made some changes to the air conditioning, however we didn’t have time to tear it down for this evaluation. Factor is though, they physically appear like the exact same item.
This indicates the GeForce RTX 4080 is still using the controversial 16-pin 12VHPWR power port. A great deal has actually been claimed regarding this connector over the past month, however what the real story is– in our viewpoint– it’s unidentified at this point. So we’re not going to feed right into that tale as we believe a great deal more investigation needs to be done prior to drawing any type of verdicts.
A single PCIe 5.0 power port can deliver up to 600 watts and previously that would require 4 8-pin power ports. Nevertheless, you won’t need a new PCIe 5.0 certified power supply as RTX 4080 includes a 3x 8-pin to single 16-pin adapter, comparable to the 3x 8-pin to 16-pin model provided with the 3090 Ti.
Along with the boost in cores, there’s the inclusion of 4th-gen Tensor cores as well as 3rd-gen RT cores, the GeForce 40 collection also presents DLSS 3, a function that in the meantime is unique to the brand-new collection.
Considering that we tested Nvidia’s DLSS 3 innovation not a great deal has altered, apart from the launch of a few more video games supporting it. Our examination came after the launch of the RTX 4090 and stays the very best place for a comprehensive malfunction of the technology. In our screening, DLSS 3 mainly executes in the same manner on the RTX 4080. Nvidia understands the issues we brought up and is working on a few fixes for points such as broken UI elements as well as transitions.
As for our basic ideas on DLSS 3, the innovation does have some benefit, however it’s extra restricted compared to DLSS 2. Gamers will benefit most from DLSS 3’s Frame Generation technology when video games are currently being performed at a good framework rate: somewhere around 100 FPS before DLSS 3 framework generation is made it possible for, with DLSS 3 then increasing FPS to around 180-200 FPS. This improves aesthetic level of smoothness with high refresh rate displays and functions fairly well under these problems as it conceals visual artifacts as well as latency troubles are decreased. We would not suggest it for video gaming below 120 fps, so it’s more of a niche function that adds to Nvidia’s strong function set.
For this evaluation we won’t be including any DLSS 3 standards though we prepare to revisit it when Nvidia addresses the concerns discovered in our first investigation. Another big concern with DLSS 3 standards is reporting structure generation as an “FPS” metric, as 120 fps making use of DLSS 3 structure generation isn’t truly 120 fps– rather it’s the smoothness of 120 fps with the input of about half that structure rate, plus a few visual issues.
This is a serious concern for comparing benchmark results and we would suggest anyone benchmarking with frame generation enabled should report the numbers as “120 DLSS 3.0 fps,” as an example.
For testing all GPUs were running at the official clock specifications with no factory overclocking. The CPU used is the Ryzen 7 5800X3D with 32GB of dual-rank, dual-channel DDR4-3200 CL14 memory on the MSI MPG X570S Carbon Max Wi-Fi motherboard. In total we tested 13 games at 1440p and 4K, so let’s get into the data…
Benchmarks
To begin with we have Watch Dogs: Myriad and also at 1440p we are CPU limited with these greater end GPUs. Still the data is relevant for any person with a 5800X3D or a CPU that’s slower. But for unrestrained GPU performance let’s relocate to 4K …
At 4K we get a better concept of what the RTX 4080 can. 110 fps on average implied it was 22% slower than the RTX 4090, yet additionally 25% faster than the 6950 XT, 42% faster than the 3080 Ti and 57% faster than the 3080. That’s some pretty outstanding generational enhancements.
Rainbow Six Extraction isn’t as CPU limited at 1440p as what we saw with Watch Dogs: Legion and as a result the 4080 was 40% faster than the 6950 XT and 30% faster than the 3090 Ti. That said it was just 18% slower than the 4090, so let’s see if that margin grows at 4K.
As expected it does as below the 4080 was 25% slower than the 4090, though it did still manage a remarkable 158 fps at 4K. That made it a large 82% faster than the 3080 Ti, 96% faster than the 3080 as well as 102% faster than the 6950 XT.
For screening Far Cry 6 we use real in-game performance rather than the built-in standard tool. Unsurprisingly, the information is CPU restricted at 1440p and therefore the 4080 was simply 3% slower than the 4090 with 181 fps usually.
Moving to 4K changes the primary efficiency limitation onto the GPU and now the RTX 4080 is 43% faster than the 6950 XT, 50% faster than the 3080 Ti as well as 66% faster than the 3080.
Next we have Assassin’s Creed Valhalla and also at 1440p the RTX 4080 was good for 161 fps which saw it route the 4090 by a 12% margin, while defeating the 6950 XT by a 12% margin, however let’s move to 4K to see how those margins transform.
Also at 4K the RTX 4080 was just 14% slower than the 4090 and also just 15% faster than the 6950 XT. The Radeon GPU is aided along by excellent support for Resizable BAR, so in general a disappointing result for the brand-new GeForce GPU.
The GeForce RTX 4090 really amazed us in Search Showdown as well as while the 4080 isn’t as outstanding, it’s still well ahead of previous generation components, routing the 4090 by a 22% margin however leading the 6950 XT by 34%, the 3080 Ti by 37% as well as the 3080 by a huge 58%.
We can see that the 4K margins are similar. The RTX 4080 was 24% slower than the 4090 with 119 fps generally, making it 27% faster than the 6950 XT and 51% faster than the 3080.
We’re seeing a clear pattern arise. This time testing with The Outer Globes at 1440p saw the 4080 just 21% slower than the 4090, yet 22% faster than the 3090 Ti, 27% faster than the 3080 Ti, 34% faster than the 6950 XT, as well as a large 46% faster than the 3080.
The margins continued to be the exact same at 4K though there are some weird adjustments. We see that the RTX 4080 was good for 119 fps making it 25% slower than the 4090, however simply 13% faster than the 3090 Ti despite still defeating the 3080 Ti by a 28% margin, so the 3090 Ti does a little far better than you may expect.
Gunman 3 is an extremely CPU extensive video game and also the 5800X3D isn’t the fastest CPU here, not even gather the release of Zen 4 and also Raptor Lake.
CPU restrictions aren’t a concern at 4K nevertheless and here we see that the 4080 is the expected 22% slower than the 4090 with 142 fps generally. This did make it 26% faster than the 6950 XT and 69% faster than the RTX 3080.
The 5800X3D does perform well in Horizon Absolutely no Dawn and also despite that we still face a tough CPU traffic jam with the brand-new GeForce 40 series GPU as both the 4080 and 4090 were limited to just over 210 fps.
Lifting to 4K removes CPU efficiency constraints and also supplies the expected margins. As we have actually seen over and over again, the 4080 is a little over 20% slower than the 4090 but faster than the rest of the pack.
Next we have F1 2021 as well as at 1440p the 4080 was 16% slower than the 4090 with 209 fps, which suggested it was 34% faster than the 3090 Ti and a huge 48% faster than the 6950 XT.
Then at 4K we see predictable margins as currently the 4080 is 25% slower than the 4090, and also 52% faster than the 6950 XT. It was simply the 4090 that was running into system limitations at the lower 1440p resolution.
Carrying on to Cyberpunk 2077 we find that the 4080 was good for 128 fps typically with the 5800X3D pairing, that made it 12% slower than the 4090, and just 17% faster than the 6950 XT.
Raising the resolution to 4K results in fairly common margins based on what we have actually seen thus far, the 4080 is currently 23% slower than the 4090, however 33% faster than the 3080 Ti as well as nearly 50% faster than the initial 3080.
Dying Light 2 Stay Human isn’t extremely CPU demanding and as a result we’re not limited at 1440p, enabling the RTX 4080 to produce 151 fps typically, leading to a typical 21% deficit to the 4090.
That margin did expand slightly to 25% at 4K as well as remarkably the 4080 was simply 11% faster than the 6950 XT, because of an abnormally solid result for the Radeon GPU.
In Halo Infinite we are clearly encountering a CPU bottleneck at 1440p as the RTX 4080 was just 6% slower than the 4090, yet impressively still 42% faster than the 3090 Ti and 51% faster than the 6950 XT.
At 4K the RTX 4080 is now only 32% faster than the 3090 Ti which is still a large margin, but smaller than the 42% margin seen at 1440p.
In Shadow of the Tomb Raider at 1440p, the RTX 4080 provided a remarkable 206 fps, which indicated it was 13% slower than the 4090, however 35% faster than the 6950 XT.
Bumping up the resolution to 4K saw an average of 116 fps, though this did make the 4080 28% slower than the 4090, so that is just one of the bigger margins we have actually seen. It was additionally only 23% faster than the 3090 Ti and also 33% faster than the 6950 XT.
13 Game Average
Checking out typical frame rates at 1440p, the RTX 4080 generated 189 fps, making it 12% slower than the RTX 4090 at 1440p. It was additionally 24% faster than the Radeon 6950 XT, and also 25% faster than the RTX 3090 Ti, while we’re looking at a 38% boost over the 3080 Ti as well as 51% over the RTX 3080.
At 4K where the CPU traffic jam is removed, the RTX 4080 got on typical 23% slower than the RTX 4090, making the front runner Ada Lovelace GPU 30% faster.
The RTX 4080 was additionally 22% faster than the 3090 Ti, 31% faster than the Radeon 6950 XT, 37% faster than the 3080 Ti as well as 52% faster than the 3080.
Just because we understand many of you will certainly intend to see the 1080p data, here’s the 13 game average at the lower resolution. Obviously, we were generally greatly CPU limited in this testing, which is why the RTX 4080 is just 16% faster than the 3090 Ti and just 7% slower than the 4090.
Ray Tracing + DLSS Performance
Time for a consider ray mapping and also upscaling performance, starting with F1 22. At 1440p the 4080 and also 4090 are limited to around 300 fps with the RT results impaired. Enabling RT + DLSS resulted in approximately the same level of efficiency with 181 fps for the 4080 and 189 fps for the 4090, this made the RTX 4080 simply 12% faster than the 3090 Ti.
After that with upscaling handicapped, the RTX 4080 was 32% slower than the RTX 4090 and also 22% faster than the 3090 Ti, while it squashed the 6950 XT by a substantial 83% margin.
Transferring to 4K aids to remove the CPU traffic jam and after that the 4080 is 27% slower than the 4090 prior to RT results are also allowed. The margin in fact grew to 36% with RT + DLSS allowed and then 38% with just RT enabled with no upscaling. When contrasted to the 3090 Ti the 4080 was 19% faster and then 100% faster than the 6950 XT, yikes.
Next up we have Watch Canines: Myriad as well as here the 4080 as well as 4090 delivered comparable performance under all three examination problems. The 3090 Ti was additionally able to match the 4080 with DLSS made it possible for as this seems the limit of what the 5800X3D can render with RT made it possible for.
Moving to 4K provides us a much better image of where the RTX 4080 stands as well as below it is 27% slower than the 4090 with RT + DLSS enabled, and 26% faster than the 3090 Ti. It was additionally an enormous 115% faster than the 6950 XT which was using FSR.
Marvel’s Guardians of the Galaxy is likewise rather CPU restricted at 1440p, obviously the efficiency overall is excellent with structure prices up around 200 fps without RT enabled, but it also does not reveal us how the 4080 genuinely accumulates, so let’s move to 4K.
At the 4K resolution, the 4080 is 22% slower than the 4090 with RT + DLSS allowed, and also 21% faster than the 3090 Ti. It was also an enormous 81% faster than the 6950 XT which was once more using FSR.
Passing away Light 2 shows that at 1440p the RTX 4080 was 26% slower than the 4090 when comparing the RT + DLSS outcomes. This likewise made it 47% faster than the 3090 Ti, which is a massive margin as well as we saw formerly that the GeForce 40 collection does extremely well in this title.
It’s a comparable tale at 4K, where the 4080 was 31% slower than the 4090, however 46% faster than the 3090 Ti and also an enormous 100% faster than the Radeon 6950 XT which was again making use of FSR.
The last ray tracing/DLSS benchmark that we have for you checks out Cyberpunk 2077. Here the RTX 4080 provided 102 fps at 1440p which RT + DLSS made it possible for, making it just 10% slower than the 4090 and also 23% faster than the 3090 Ti.
At 4K those margins change as well as currently the 4080 is 27% slower than the 4090 when both are running RT results with upscaling enabled. This additionally saw the 4080 loss the 3090 Ti by a 26% margin as well as the 6950 XT by an unbelievable 115% margin.
Power Usage
The GeForce RTX 4080 is remarkably good in terms of power usage as we see in Dying Light 2, pushing use to just 404 watts, which is similar to that of the RTX 3070 Ti and Radeon 6800 XT, which is very manageable. It’s additionally a 23% reduction in overall system usage when contrasted to the 4090.
Power usage did creep up a little in Doom, hitting 471 watts as well as hence the RTX 4080 discovers itself positioned between the 3080 and 3090, however it’s a decrease in overall system use by 22% when contrasted to the RTX 4090 flagship.
Ultimately in Hitman 3, we discover even more of the exact same, power usage is still lower than the 3090 and also is really equivalent to the 3080. Given the performance simply seen, we would certainly say the RTX 4080 is very effective, however let’s lock the framework rate for an intriguing consider power efficiency …
With all 4 GPUs locked to 90 fps, we get to see simply exactly how power effective the RTX 4080 is.
The RTX 4090 really impressed in this examination prior to and also we see that the smaller sized AD103 die is much more effective once again, going down power usage by ~ 6%. This suggests that under the very same operating conditions, the RTX 4080 makes use of half as much power as the RTX 3090 Ti– super impressive things.
Cost per Frame
For our cost per structure analysis we’ll start with MSRP prices, just to get a sensation for where prices over the life of each item must have been. Beginning with the expense per structure based upon the 1440p information, we see that the RTX 4080 slots in between the 3070 Ti as well as 6900 XT, neither of which were particularly fantastic worth in their particular series.
When contrasted to the RTX 3080, cost to efficiency has actually regressed by 13%, as well as 27% when contrasted to the Radeon RX 6800 XT. Still, Nvidia would be quick to explain that when contrasted to the 3080 Ti, the RTX 4080 is a substantial renovation in value, minimizing the price per structure by 28%.
The 4K MSRP information is a little more beneficial, yet nevertheless the RTX 4080 is still 13% more expensive per framework than the RTX 3080, so in that sense even based on MSRP prices, it’s not wonderful value. However when contrasted to the RTX 4090 it is a minor enhancement, though we believe several who are delighting the concept of investing $1,200 on a graphics card would probably simply opt to go the complete hog with the 4090.
The larger concern with the RTX 4080 when talking about worth is the current GPU market and in this chart using the 1440p data it slips to one of the weakest choices you can acquire now, just about matching the awful worth RTX 3080 Ti. But offered exactly how CPU restricted a lot of the 1440p results were, allow’s make our final value analysis based on the 4K information.
The expense per frame results based on the 4K data utilizing current retail pricing. Here the RTX 4080 is a little far better value than the 4090, reducing the expense per frame by just 3%.
Value-wise the RTX 4080 is additionally similar to the 3080 Ti at its existing asking cost of $860, which frankly isn’t excellent considered that makes it 11% even more pricey per structure than the 3080, which is even worse value than AMD’s 6900 XT and also worlds worse worth than the 6800 XT.
Obviously, none of those GPUs are also remotely in the exact same league as the RTX 4080 when it involves performance, so it will interest see what RDNA 3 can do below.
Extra RTX 4080 AIB Models
Prior to we’re done today, allow’s take a look at the three companion designs we have on hand before looking into the operating behavior of each of them and Nvidia’s FE version.
Starting with MSI’s RTX 4080 Suprim X, this one is rather straightforward if you have actually seen our RTX 4090 Suprim X review as it’s the specific very same cooler. The 4080 variation considers just 60 grams much less and also like the FE model, that difference can be credited to the smaller sized GPU, less memory chips, cut-down power delivery and less elements on the PCB as a whole. Still at 2360 grams this thing is still hefty as well as enormous.
Offered the RTX 3090 Ti version of the Suprim X is smaller, the dimension and weight of the RTX 4080 is unneeded, however it seems as though it was just easier for Nvidia’s partners to standardize these monstrosities and also stick them on all current generation premium graphics cards.
No matter exactly how you really feel regarding these comically huge graphics cards, it appears they’re below to stay. The Suprim X is an excellent high quality board, it looks great, and also as we’re about to see it functions very well.
The Asus RTX 4080 TUF Gaming is a carbon copy of the RTX 4090 version. This thing is wrapped in steel, with really couple of plastic elements to mention. Nevertheless, while it appears to be the very same cooler, there are some notable changes which will certainly lower production expenses.
Whereas the 4090 TUF Gaming tipped the scales at 2320 grams, the 4080 version is 17% lighter at 1920 grams. The bulk of that difference can be attributed to the colder which is 308 grams lighter at an overall weight of simply 772 grams.
Whereas the RTX 4090 version used 8 nickel-plated copper heatpipes to call the large vapor chamber, the 4080 gets rid of the vapor chamber totally and although there’s still eight heatpipes, simply 2 of them determine 8mm in size with the rest of the 6mm variety. The 4090 packed 5 8mm heatpipes as well as this allowed for a dramatically bigger contact with the base plate.
Speaking of the base plate, the RTX 4080 variation just makes contact with the GPU while a light weight aluminum spreader contacts the memory chips, with the aid of thermal pads obviously. After that over on the PCB there are less power provide parts, the much smaller GPU as well as just eight memory chips opposed to the lots you’ll locate on the 4090.
After that we have the PNY RTX 4080 which wins the honor for the worst name, are you prepared for this? It’s the PNY RTX 4080 XLR8 Pc Gaming Verto Epic-X, so let’s just call it the Verto.
This is a relatively typical looking RTX 4080, it measures 71mm large, so it will use up 4 ports, it’s 332mm long as well as stands 136mm tall from the PCI adapter. All up it weighs 1860 grams making it simply 3% lighter than the RTX 4090 version.
There are some real layout modifications in between the 4080 and also 4090 variation, regardless of them looking really comparable overall. The RTX 4090 has an enhancing architectural bracket that runs the entire criterion of the card, whereas the 4080 gets rid of the bracket for a modified heatsink design.
The only issue we have with this model is the absence of a dual-BIOS function which for numerous can be a little a deal breaker, specifically at this price point. So keeping that allowed’s take a look at thermals, operating clocks, and power.
Cooling
Here’s a check out how the Nvidia Founders Edition runs after a hr of Hitman 3 at the 4K resolution utilizing max setups. These temperatures were taped in a 21c area set up inside an ATX instance with the doors shut. The GPU hotspot came to a head at 78c while the memory got to 74c, and we saw an ordinary core clock regularity of 2760 MHz, a power draw of 300 watts as well as a fan rate of 1450 RPM. Good outcomes however nothing fantastic.
The MSI Suprim X outcomes were remarkably similar with the hotspot temperature level likewise hitting 78c though the memory was slightly cooler at 72c. We’re also taking a look at a similar operating frequency at 2745 MHz with power draw at 308 watts. The fans were spinning slower at 1200 RPM, yet generally a really comparable result to Nvidia’s FE version.
The Asus TUF Gaming model spun its fans a little bit quicker at 1550 RPM which limited the peak hotspot temperature level to 77c and also the memory to 72c, so the exact same result as the Suprim X as well as FE designs. The typical power draw was higher at 314 watts as well as this allowed for a slightly higher continual 2775 MHz operating regularity.
Ultimately, the PNY Verto, as well as you thought it, primarily the exact same outcomes once more. Below we have a hotspot temperature level of 78c though the memory was substantially cooler at 62c, to ensure that’s a remarkable result. The typical power draw was likewise rather typical at 312 watts and we have a typical 2730 MHz core clock regularity. All this at a 1300 RPM follower speed, so the differentiating factors here are the card’s look and pricing.
What We Discovered
The new GeForce RTX 4080 is clearly an excellent product, providing usually 111 fps at 4K throughout the 13 video games checked. That’s a 20 to 30% increase over previous generation flagships.
That’s all fantastic other than that it’s likewise ‘adequately’ priced above previous generation front runners. It’s currently 20% a lot more costly than the Radeon 6900 XT as well as just 25% less than the enormously overpriced RTX 3090.
While we absolutely like the efficiency as well as functions of the RTX 4080, it’s hard to obtain as well excited when we know you will need to swallow the $1,200 asking rate for the advantage.
AMD is anticipated to counter-punch next month and also we can prepare for that the awkwardly named Radeon RX 7900 XTX will show up next month at $1,000. This GPU needs to conveniently defeat the RTX 4080 in regards to rasterization performance– yet there’s still the problem of ray tracing, which remains to gain ground and also is a vital efficiency statistics currently. Additionally, while AMD’s FSR remains to enhance, so does DLSS, which stays the premium upscaling modern technology.
So Nvidia would certainly have increased the 80 collection MSRP by around 30%– we understand, “just”– would certainly have seen the GeForce RTX 4080 coming in at $900, or $200 more than the RTX 3080 at launch. In that theoretical circumstance, the 4080 would certainly have been excellent value in today’s market, and also would certainly end up being the best value GeForce GPU on the market, matching the cost per structure of the $510 Radeon 6800. Regretfully however, that’s not to be, because Nvidia knows gamers will certainly pay $1,200, and the efficiency increase appears to takes them there.
If you desire leading efficiency, we currently recognize precisely how the GeForce RTX 4080 does as well as while you exist, the RTX 4090 could be a much better selection for numerous since it’s the quicker (and also much more costly) GPU alternative. But if you’re seeking best worth, it’s challenging to draw any type of verdicts till we have actually seen what the 7900 XTX has to use as well as see what gets you the most value. It’s not a long haul either.