While we’ve reviewed a few Gigabyte graphics cards in the past, this is the first time we’re testing the new design that powers the Gaming OC series. It’s also the same design that powers the RTX 4090 and 4080 models in the same series. So it’s the Gigabyte RTX 4070 Ti Gaming OC that we’ll be reviewing today. This model has made a lot of talk during the announcements of NVIDIA (link) since it was originally supposed to be an RTX 4080 but equipped not with 16 GB but only with 12 GB. We are therefore impatient to discover the performance of this model as well as its positioning against the two AMD graphics cards launched at the end of 2023.
We start this new year with the arrival of this new model which will hopefully balance the current offer of GPUs that are mainly intended for the high-end. No doubt that once again, we will have to pay attention to the price/performance ratio.
Thanks to NVIDIA and Gigabyte for making available this copy of the RTX 4070 Ti Gaming OC.
[nextpage title=”Gigabyte RTX 4070 Ti Gaming OC: “]
As a reminder, there will be no ” Founders Edition ” cards offered by NVIDIA for the RTX 4070 Ti. You can therefore only buy partner cards and therefore, depending on the model, with very different prices.
This RTX 4070 Ti has an AD104 GPU with 7680 CUDA cores, which is 2560 cores less than the RTX 3080 Ti. It is equipped with a manufacturing process in 4nm from TSMC and not the 8 nm manufacturing process by Samsung, which allows the TGP to be lowered to 285 watts.
In terms of memory, there are 12 GB of GDDR6X memory in 192-bit, which translates into a bandwidth of 504.2 GB/s. On page four of our article, we will go into detail on the new features of this new architecture, including support for DLSS 3.
The bundle of this Gaming OC:
So what do we find in this box? We find and identify very easily the color code of Gigabyte with black and orange. The box is quite large but still has standard dimensions. The dimensions of the box are 440 x 235 x 95 mm. The box opens on the side to reveal the RTX 4070 Ti Gaming OC.
Once removed, we find a small box that contains the adapter for the power supply of the card. It is a 16-pin (12+4) adapter to two 8-pins of 150w (300w in total). This adapter is necessary for those who do not have a power supply with a 16-pin port or the adapter for their power supply. There is also a support for the graphics card so that it does not bend when inserted in your motherboard.
Cooling:
This Gigabyte RTX 4070 Ti Gaming OC has the dimensions 336 x 140 x 58 mm and will occupy 2.5 slots on your motherboard. This model is part of the Gigabyte “Gaming OC” series. This means that the card is already factory overclocked. Aesthetically, the card plays on sobriety with touches of black, anthracite and dark gray. It is a model that can be matched with many motherboards.
We are dealing here with a custom card, that is to say with a cooling system specific to the brand. It is labelled “Windforce 3x” and has three 100mm fans. Gigabyte uses its patented “Alternate Spinning” technology for this card. The principle is quite simple and involves the three fans rotating in opposite directions. The goal is to avoid turbulence between them.
The nine fan blades are slightly curved to optimize airflow. Our sample features ” 3D Active Fan ” technology to leave the fans idle if the GPU temperature does not require them to run.
The black plastic casing completely encloses the PCB with a very qualitative finish. The aesthetics, quite neutral, should suit many users. Two BIOS are also present, a Silent mode and an OC mode. By default and for our tests, we will use the OC mode.
Radiator:
The heatsink is composed of a single block. As you can see on the pictures, its role is to dissipate the heat from the GPU, the memory chips and the power supply stage.
The heat sink is crossed by simply seven 6 mm heat pipes. These are in direct contact with the graphics chip. The memory chips, the power supply stage and some of the PCB chips are covered with thermal pads to ensure heat dissipation. This is a very strong cooling system that should keep all the components cool!
Finally, we find at the back a backplate whose role is to stiffen the PCB. This one benefits from a very large cut at its end to let the air pass through the radiator. It is easy to imagine that the PCB is much shorter than the heat sink that covers it.
Yes, some backlighting:
Quite a bit of backlighting with the outline of each fan as well as the Gigabyte logo on the top edge of the card and controllable via the RGB Fusion 2.0 software. There are no less than 9 RGB effects that can be set via the software.
The connectors:
To power the card, we find the new 16-pin mini-connectors. As we mentioned above, NVIDIA provides in its bundle an adapter to connect two 8-pin connectors.
The card has three DisplayPort 1.4 video outputs and an HDMI 2.1 port.
[nextpage title=”Specifications and PCB: “]
The technical specifications:
In terms of frequency, the base frequency is announced at 2310 MHz while the Boost is at 2640 MHz. The 12 GB of G-DDR6X memory run at 1313 MHz (504.2 GB/s) on a 192 bit bus. As a reminder, there are no reference cards on this model proposed by NVIDIA. We will only have the right to partner cards that will offer a factory overclocking more or less consistent depending on the model.
Let’s have a look at GPU-Z to check that the information read is correct. The GPU frequency is 2310 MHz, and the Boost frequency is 2640 MHz. Don’t forget that the latter will be higher during benchmarks depending on the quality of your chip and determined by GPU Boost 4.0. The memory has a frequency of 1313 MHz.
Let’s unpack the PCB:
Let’s now deal with the VRM, Voltage Regulator Module. We find a set of 10 + 2 + 1 phases. The number of phases is often determined by the current requirement of the graphics chip. The more current it needs, the more phases there are.
Then we find the mosfets which are here SIC653A. The mosfets are the elements to keep cool and are often topped by a radiator. They are the ones that will convert the direct current since our GPU does not work in 12 or 5 volts, but on a lower voltage often located around 1 volt. Let’s also note to finish, the presence of voltage points integrated directly on the PCB.
Memory chips:
As far as the memory is concerned, there are six of them. They are memory chips of 2 GB in GDDR6X, D8BZC and manufactured by Micron. They should have a good potential in overclocking.
[nextpage title=”New technological innovations: “]
A new generation automatically means a lot of technological innovations and this RTX 4070 Ti, like its big sisters the RTX 4090 and RTX 4080, is no exception to the rule. We will try to mention the most important ones in this page, which is a bit more theoretical.
New Ada Lovelace architecture:
On September 21st at 5pm, NVIDIA unveiled part of its future range of graphics cards based on the “ADA Lovelace”. Technically, this chip is out of the ordinary with its 76 billion transistors, its N4 engraving by TSMC, its Shader Execution Reordering, its 90 TFLOPS, its 200 RT TFLOPS, its 1400 Tensor TFLOPS and a maximum of 18432 CUDA Cores in a single GPU. This is quite impressive compared to the previous generation Ampere.
The chip is organized into 12 Graphics Processing Clusters (GPC) each containing 72 Texture Processing Clusters (TPC) with 144 Streaming Multiprocessors (SM). The Streaming Multiprocessors benefit from a new technology called Shader Execution Reordering (SER) which allows to multiply by two the Ray Tracing operations. We can also notice that the chip of the RTX 4090 is amputated allowing perhaps to see a future RTX 4090 Ti land.
These new RTX also come with 4th generation Tensor Cores, which will be very useful for DLSS and DLSS 3 AI as well as 3rd generation RT Cores which will allow to double the performance of Ray Tracing. In addition to the evolution of generation, their number also increases compared to the previous generation depending on the model.
DLSS 3 is here and already operational :
One of the announcements of NVIDIA during its conference is the arrival of DLSS 3. Be careful though, this one is only compatible with the RTX 4090 and RTX 4080. Owners of a previous generation card will be happy with the DLSS 2 that NVIDIA will continue to develop in the coming months.
As a reminder, until now, the principle of DLSS (Deep learning super sampling) was to allow a gain in terms of FPS by replacing the antialiasing TAA. The idea is to calculate the rendering in a lower definition than that chosen in the parameters and then to replay it in the definition said using an AI applied via the Tensor Core while maintaining an image quality as perfect as possible.
Today, with the arrival of DLSS 3, things are changing even if the fundamental principle of increasing the number of FPS is still the main reason. The novelty comes from the fact that DLSS 3 allows you to add whole intermediate images and no longer pixels. This increases the number of frames per second and consequently the fluidity. To achieve this, DLSS3 uses an optical flow gas pedal integrated into the new RTX 4000 architecture. This is why previous generations that do not have this will not be able to benefit from DLSS 3. We will have to be careful during our tests and compare what is comparable because yes, of course DLSS 3 will be more powerful but it is difficult to judge it on the Ampere generation which does not support it.
Even if the main function is to increase the number of FPS, NVIDIA also reminds the importance of latency. To summarize simply, the principle is to reduce the latency in games as much as possible, that is to say the time that elapses between the moment you click on the button of your mouse and its manifestation on the screen. In order to be able to perform all the readings, a compatible mouse will be necessary.
The latest advances from NVIDIA could turn the world of e-sports upside down, as we could see 1440p 27″ displays replacing the current 1080p 25″ displays. Finally, NVIDIA has already announced over 35 games that are compatible and ready to use DLSS 3. We will of course add games that support DLSS 3 to our tests. Of course DLSS 3 games are backwards compatible with DLSS 2 technology.
AV1 hardware compression:
NVIDIA now integrates its 8th generation NVENC encoder with AV1 which is 40% more efficient than H.264. The arrival of AV1 will be very useful especially for streaming.
[nextpage title=”The test configuration: “]
As we said in the introduction, we are (again) busy updating all our graphics using the latest drivers from NVIDIA, AMD and of course INTEL. We also took the opportunity to slightly modify our test configuration. So here we go again for a new series of tests and therefore of graphics on this slightly modified configuration:
- Motherboard: ROG Maximus Z690 APEX
- Processor: Intel Alder Lake i9-12900K
- Memory : 32 GB Corsair Vengeance 5200 CL38 DDR5
- Graphics Card : Gigabyte RTX 4070 Ti GamingOC
- Cooling system : ROG Ryujin II 360 from Asus
- SSD : Western Digital Black SN750 2 TB + EKWB radiator
- Power supply: Corsair AX1600i + 12VHPWR 600W adapter
- Operating system : Windows 11 64 bits
For this test, we used a high-end configuration using an i9-12900K processor and DDR5. The use of a 144 Hz 4K display is also a prerequisite. For our tests we use a copy of the ROG PG27UQ.
The drivers used are NVIDIA 517.48 and AMD 22.9.2. For our sample of the day, we used the press drivers in version 527.62.
Our test protocol:
All cards are tested at stock frequencies and without any optimization. For the various benchmarks, whether synthetic or gaming, we will opt for 1080p, 1440p and 2160p resolutions depending on the performance of the card. Here, with this high-end model, we will only use the 2160p definition. We felt it was mandatory to compare the performance against the two AMD models we just tested.
As you can see, we have switched to Windows 11 and we have activated the ” ReSize BAR ” in the BIOS, which is visible in GPU-Z. Now that you know everything about our test configuration, it’s time to get started!
[nextpage title=”Synthetic benchmarks: “]
Our first series of tests will be dedicated to synthetic benchmarks. These are often used for overclocking competitions but are also very interesting to compare different GPUs. So we will focus on the 3DMark benchmarks.
We will test the following benchmarks:
- Fire Strike (1080p)
- Fire Strike Ultra (2160p)
- Time Spy (1440p)
- Time Spy Extreme (2160p)
- 3DMark Speed Way
- DirectX Raytracing
- Intel XeSS
- Mesh Shader
3DMark Fire Strike and Fire Strike Ultra :
Let’s start with 3DMark Fire Strike, which is one of the most used benchmarks today. It consists of two graphics tests, a CPU test and a fourth test that combines GPU and CPU. Don’t forget to deactivate the demo which doesn’t bring anything in the final score and prolongs the benchmark duration (unnecessarily). The version used for these tests is of course the latest one
First benchmark that we realize with this NVIDIA RTX 4070 Ti Gaming OC and the least we can say is that we leave the top of the graph to be in the middle part! The score obtained is 43115 points allowing it to be directly behind the RTX 3090 Ti. As you will notice later, the balance of the graph is maintained thanks to the fact that the benchmark is done in 1080p.
The same is true under Fire Strike Ultra as our copy of the RTX 4070 Ti climbs up the rankings a bit, this time overtaking the RTX 3090.
3DMark Time Spy and Time Spy Extreme:
The second test is 3DMark Time Spy. Although this one is done in 1440p, it has the particularity of using DirectX 12. It consists of two graphical tests and a CPU test. As for Fire Strike, don’t forget to disable the demo. These two first benchmarks are proposed by UL Benchmark.
The performance obtained in Time Spy is often representative of what we will get in games. And here, our Gigabyte RTX 4070 Ti Gaming OC comes in behind the AMD RX 7900 XT model. The gap is quite large, which leads us to imagine that it will be the same in games.
Under 3DMark Time Spy Extreme, it is a score of 10717 points that we get and that places it just behind the RTX 3090 Ti. We should have a card that will be between the RTX 3080 Ti and the RTX 3090 Ti in terms of performance according to these first benchmarks.
3DMark Speed Way
Latest benchmark from Ulbenchmark and it’s nice to be able to say that we’re going to have a good time on this Speed Way. It uses DirectX 12 and is by default in a 1440p definition. We have not modified anything and we use the original benchmark.
Again, our RTX 4070 Ti is right where we thought it would be, between the RTX 3080 Ti and the RTX 3090 Ti with a score of 5501 points. It manages to take the lead over the RX 7900 XT for the first time.
3DMark DirectX Raytracing :
UL Benchmarks has added a new test to its series of benchmarks to measure Ray Tracing performance. The idea will be to use the 3DMark DirectX Raytracing test to compare the performance of Ray Tracing hardware dedicated to AMD and NVIDIA graphics cards but now also to Intel!
The 3DMark DirectX Raytracing test is designed to make Ray Tracing the limiting factor. Instead of relying on traditional rendering, the entire scene is traced and drawn in one pass. The test result will depend entirely on the performance of the Ray Tracing. It will be easy to measure and compare the performance of the different cards between them.
This benchmark has always been dominated by NVIDIA cards and our sample of the day is no exception to the rule as we get a score of 68.8 FPS. The use of Ray Tracing is always to the advantage of the greens. We should see our sample perform better in gaming with Ray Tracing and DLSS active.
3DMark Intel XeSS
UL Benchmarks in collaboration with Intel will add this new benchmark to the 3DMark suite. We had early access to this new test for our testing. This Intel XeSS test is designed to evaluate and compare the performance and image quality of XeSS (Xe Super Sampling). There are four XeSS modes to choose from: Ultra Quality, Quality, Balanced and Performance. The 3DMark inspection tool helps you compare image quality with a side-by-side view of XeSS rendering and native resolution rendering. You need a graphics card that supports Intel XeSS to run this test.
So we’ll have two scores, a FPS number with XeSS disabled and then on the right with XeSS enabled. The mode chosen is “Ultra Quality” which is actually the default mode for the benchmark.
In this new benchmark, all graphics cards without exception take advantage of XeSS but it is Intel that has the best gain. ARC models see a gain of 50% while NVIDIA and AMD models see a gain of more or less 30-35%.
3DMark Mesh Shader:
Here is the new test we added to the synthetic benchmarks: Mesh Shader. We added it because it is managed by the RDNA 2 architecture of AMD and therefore our RX 6000. It allows us to analyze the performance level of GPUs concerning this feature. As a reminder, it allows to exploit more complex calculations at the geometry level, in a more flexible way, for each meshlet . The displayed result is the one with the Mesh Shaders active.
Well, as we conclude this first series of synthetic tests, we are a bit disappointed with this Gigabyte RTX 4070 Ti Gaming OC. Finally, NVIDIA made the right choice to downgrade its RTX 4080 12GB model to RTX 4070 Ti as the difference is so important with the “real” RTX 4080. We’ll be starting the games but we already have a clear idea of the performance we should expect.
[nextpage title=”2160p gaming performance: “]
We decided to select six games because they seem to be representative enough. We have opted for DirectX12 and Vulkan with the following games:
- The Division 2 (DirectX 12)
- Borderlands 3 (DirectX 12)
- Red Dead Redemption II (Vulkan)
- Far Cry 6 (DirectX 12)
- Watch Dogs: Legion (DirectX 12)
- Cyberpunk 2077 (DirectX12)
We have chosen to use the benchmarks integrated in these different games.
Why did we do this? Simply to allow you to compare your results with ours if you also have a copy of one of these games. It’s a thought that seemed relevant to us and that will allow you to compare your results with ours.
As we mentioned on page 4 of this article, we have activated the Resizable Bar in the BIOS, which is now supported by all (recent) GPUs. Finally, we have opted only for 2160p definition and selected the cards that are able to compete at this level. Unfortunately, we don’t have any RTX 3080’s available at the moment to add performance to our graphics.
We also decided to harden the graphics quality of some games since the next generations should be more powerful, both from NVIDIA and from the next AMD graphics cards.
The Division 2:
The Division 2 will be tested under DirectX 12. As for the graphics settings, they will be set to ” ultra ” mode with vertical synchronization disabled.
As a reminder, when we started our gaming tests, we said that the RTX 3080 Ti was the reference card in 2160p for us, before the arrival of the RTX 4000, since it offered an excellent price/performance ratio in this definition. It is clear that in The Division 2, our Gigabyte RTX 4070 Ti Gaming OC is between this RTX 3080 Ti and the RTX 3090, RTX 3090 Ti. This is where it should be for other games as well.
Borderlands 3:
For the second game, we switch to Borderlands 3. The graphic options are set to ” Brutal “. It is important to know that Borderlands 3 is quite greedy and that this choice of parameters has a strong impact on the FPS. It is a game that is often praised in the announcements of new graphics cards and precisely in this Brutal mode.
We continue with similar performance in Borderlands 3 with an average of 70.6 FPS and an identical position in the ranking.
Red Dead Redemption 2:
Third game and not the least, Red Dead Redemption II on PC. The advantage of this opus is to offer the possibility to test the integrated benchmark under Vulkan. We opt for the graphics options in ” ultra “.
Third game and not the least, RDR2 offers the possibility to test the integrated benchmark under Vulkan. Same position in the hierarchy with an average of 76.1 FPS.
Far Cry 6:
The very last game we add to this section is Far Cry 6, since it supports DXR as well as SFR proposed by AMD. We will also find it on the next page. For our benchmarks, we opted for ” ultra ” graphics quality.
Even though our file on this game showed an advantage for the RX 6000 cards, our sample averaged 88 FPS.
Watch Dogs Legion:
Last but not least, Watch Dogs Legion will be used for our tests with Ray Tracing and DLSS. For our benchmarks, we have opted for an “ultra” graphic quality.
In the Watch Dogs Legion benchmark, there’s no change in sight, and it’s hard for the RTX 4070 Ti to reach the top four.
Cyberpunk 2077:
This is a game that we have been wanting to add for a while and now it is done, as it now has an integrated benchmark. We will also use this game for our tests with Ray Tracing. For our benchmarks, we have opted for an ” ultra ” graphic quality.
Last benchmark of this panel of games. This is the one that shows the strongest performance with our graphics cards since few cards pass the 60 FPS mark. This is the case with our RTX 4070 Ti Gaming OC which gets an average of 50.3 FPS.
Average Game:
For the gaming average, the NVIDIA RTX 3080 Ti is the reference card in 2160p. The other cards get a score either higher or lower depending on their performance against this model.
Against the RTX 3080 Ti, the contract is perfectly fulfilled with an increase in performance. Unfortunately, if it is almost at the level of the RTX 3090, the gap with the new generation of NVIDIA and AMD is obvious. This is quite normal since we are here on an RTX 4070 Ti which, let’s remember, on the previous generation with the RTX 3070 Ti was content with the 1440p definition.
Now, and you will have understood from the beginning of this article, it is the price that will determine the interest of the card or not knowing that we will not have a Founders Edition from NVIDIA. NVIDIA announces the availability of custom models at 899 euros in France for the launch. We will have to compare with the availability of the RX 7900 XT and see at what price this card is available. Let’s now tackle Ray Tracing and in particular DLSS 3 which will be there to allow the RTX 4070 Ti to stand out from the crowd.
[nextpage title=”Ray tracing and DLSS/FSR performances:”]
Let’s now take a look at the five games we have selected that support Ray Tracing as well as DLSS and for some of them AMD’s FSR as well as Intel’s new XeSS technology which is supported by Shadow Tomb Raider and Hitman 3. As we have already reminded you, if the DLSS remains the property of NVIDIA, the FSR and the XeSS can be used with all cards from AMD, Intel and even NVIDIA.
The five DirectX 12 games that we will test with Ray Tracing and DLSS, FSR or XeSS are :
- Shadow of the Tomb Raider (DLSS)
- Watch Dogs: Legion (DLSS)
- Metro Exodus PC Enhanced Edition (DLSS)
- Far Cry 6 (DLSS – FSR)
- Cyberpunk 2077 (DLSS – FSR)
- Hitman 3 (DLSS – XeSS)
No change for this new series of games since we remain in the 2160p definition. Depending on the card, we will use DLSS for NVIDIA and either FSR or XeSS for AMD models. If FSR is not available in a game, then we will use XeSS on our AMD card.
So even though our Gigabyte RTX 4070 Ti Gaming OC is intended for DLSS 3 which we will discuss on the next page, we also wanted to measure the gain of this RTX 4070 Ti in DLSS 2 against the previous generation.
Shadow of the Tomb Raider:
First one we add to this comparison that supports Ray Tracing and DLSS 2.0. Three levels of Ray Tracing are offered in the game (medium – high – ultra) and the four levels of DLSS (performance – quality – balanced – ultra performance).
For our benchmarks, we opted for ” ultra ” graphics quality, ” ultra ” Ray Tracing and ” performance ” DLSS.
With Ray Tracing activated in ultra, an average of 76 FPS is already achieved. With DLSS, our average is 143 FPS, which means that we have no problem enjoying 2160p with Ray Tracing enabled. Our RTX 4070 Ti is now back in the top four.
Watch Dogs Legion:
Second game, Watch Dogs Legion supports Ray Tracing and DLSS 2.0. It is one of the most demanding games of the moment. Three levels of Ray Tracing are offered in the game (medium – high – ultra) and four levels of DLSS (performance – quality – balanced – ultra performance).
For our benchmarks, we opted for ultra graphics quality, ” ultra ” Ray Tracing and ” performance ” DLSS.
The Ray Tracing in Watch Dogs Legion makes all the cards suffer except for our sample of the day with an average of 35 FPS. The activation of DLSS 2 performance will boost the FPS to an average of 80 frames per second.
Metro Exodus PC Enhanced Edition:
Third game that supports DLSS 2.0. We have dedicated an article to it. Three levels of Ray Tracing are offered in the game and four levels of DLSS (performance – quality – balanced – ultra performance).
For our benchmarks, we opted for ultra graphics quality, ” high ” Ray Tracing and DLSS in ” performance ” mode.
We continue with the greedy games and an average of 42 FPS. The DLSS allows here a gain of almost 100%!
Far Cry 6:
Fac Cry 6 supports DXR as well as the SFR proposed by AMD. For our benchmarks, we have opted for an Ultra graphics quality, we have activated both DRX options as well as the FSR in ” performance ” mode.
In Far Cry 6, FSR is the only choice available to boost FPS after enabling Ray Tracing. All our cards can benefit from it, regardless of the manufacturer. The FSR is to the advantage of the AMD cards but also benefits our RTX 4070 Ti which thus increases to an average of 112 FPS.
Cyberpunk 2077:
A new game that we have added and that proposes in its settings the activation of DLSS but also FSR.
For our benchmarks, we have chosen a “high” graphic quality, an ” ultra” Ray Tracing and a ” balanced ” DLSS/FSR.
This is the most demanding game where our cards have a hard time keeping up. Here an average of 23.3 FPS is obtained. The DLSS 2 in performance mode allows a considerable gain with an average of 67 FPS. We will see later in the article that it is still possible to achieve a greater gain with DLSS 3.
Hitman 3:
The last game has also been added since it is now compatible with DLSS, FSR and XeSS technology. Depending on the model of the tested card, we choose one of these three technologies.
For our benchmarks, we chose “high” graphics quality, ” enabled ” Ray Tracing and ” performance ” DLSS/Xess.
We find an already usual situation with a fourth place in this last game. The DLSS once again brings a very important gain with an average of 84.5 FPS.
So Ray Tracing with the RTX 4070 Ti:
On our six games tested, our RTX 4070 Ti managed to exceed 60 FPS with Ray Tracing enabled combined with DLSS performance. DLSS does the job perfectly and remains an undeniable asset for NVIDIA. It manages to take the advantage over the RX 7900 XT if the game has neither FSR nor XeSS. In the opposite case, the gap is less marked.
Before starting the tests, we asked ourselves the question of the definition of our tests. Would we choose 1440p or 2160p? We have to admit that our Gigabyte RTX 4070 Ti Gaming OC does perfectly well in 2160p with most of the options pushed to the maximum in our different games!
[nextpage title=”The RTX 4080 and DLSS 3:”]
We’re going to add a new section that will concern DLSS 3 since our sample of the day, the Gigabyte RTX 4070 Ti Gaming OC is compatible. This is the third card we test that is compatible after the MSI RTX 4090 Surpim Liquid X and the NVIDIA RTX 4080 Founders Edition.
We will first test the 3DMark DLSS Feature Test beta which supports DLSS 3. Next, we’ll judge how it looks in the games Cyperpunk 2077 and A plague Tale: Requiem. In order to make sure the FPS count is correct, we installed the latest press version of NVIDIA FrameView 1.4, which allows us to check what the benchmark says and what FrameView says. In particular, we knew that the benchmark version with Cyberpunk 2077 ‘s DLSS 3 is not correct. This should be correct soon after a fix is put in place.
3DMark DLSS Feature Test:
We did not keep this benchmark since it only concerns DLSS compatible cards and was therefore not useful for our comparative tests. Here, we have retested it since it officially supports DLSS 3 since the end of the NDA on the RTX 4090.
So we’re going to run a 2160p benchmark with DLSS disabled first, then with DLSS 2 performance and finally with DLSS 3 performance.
Our Gigabyte RTX 4070 Ti Gaming OC scores 30.3 FPS without DLSS, then 81.8 FPS (+169%) with DLSS 2 enabled. DLSS 3 still provides a 64% gain over DLSS 2, so we can judge how much better DLSS 3 is.
Cyberpunk 2077 :
Let’s start with the game Cyberpunk 2077. Here are the captures of the settings we used during our tests. As far as Ray Tracing is concerned, it is in Ultra mode and then we enabled the options for DLSS which is set to Performance mode.
Here is a capture taken during the integrated benchmark. As you can see, there is an error on the part of the benchmark. We have of course, during all our gaming tests left FrameView open to make sure that what is displayed by the game corresponds to Frameview.
So how about in the game Cyberpunk 2077?
With Ray Tracing Ultra enabled, our average FPS is 23.3 FPS. As soon as we activate DLSS 2 in performance mode, we reach 67 FPS (+187%). Finally, with the new version of the DLSS 3 and still in performance mode, it is an average of 87.1 FPS that is reached!
A Plague Tale: Requiem
The second game that we have installed and that also supports DLSS 3, is the next A Plague Tale: Requiem. We also opted for the settings to use Ray Tracing and DLSS 3.
This one is perfectly active in this new game. We are preparing a file on the DLSS 3 in order to compare the screenshots with the DLSS 2 by using the NVIDIA ICAT software. This is a software that we haven’t used before, so we need some time to get used to it. You should also know that we will add Bright Memory: Infinite as a game!
[nextpage title=”Temperature, consumption and nuisance: “]
Temperature :
In order to measure the temperatures, we opted for the Superposition stress test in Extreme mode. We opted for three 5 minutes runs with 10 minutes interval between each test. We took advantage of this benchmark to measure the power consumption of the graphics card for the table above.
In order to know the temperature in Idle, after having finished the three loops of the stress test, we let the card rest for 20 minutes before reading the temperature. It should be taken into account that most of the cards have the “0 dBa” technology and therefore, the fans turn off below a certain temperature, often between 50 and 55°C. This means that it takes longer to dissipate the heat after a period of stress testing or gaming. A card without this technology will be noisier than the others at rest. Small reminder: our temperature readings are taken in Open-Air. In a closed case, the cards will necessarily be hotter.
As we suspected, the temperatures are excellent due to the large cooling system compared to the size of the PCB. We are here on a heatsink that is twice the size of the PCB. This cooling should allow us to take full advantage of the overclocking potential but also to reduce noise considerably. The fans of our Gigabyte RTX 4070 Ti Gaming OC remain idle when temperatures are cool.
Power consumption (full configuration) :
We use a power meter by reading the power consumption on the output of our PSU. Our modular power supply is a CORSAIR AX1600i. The result we propose is the average of the total consumption of our configuration. Only the graphics card changes between each reading. It is therefore easy to calculate the difference between each graphics card.
This is a good surprise and confirms the work done by NVIDIA on this new generation!
The global consumption of our configuration equipped with our NVIDIA RTX 4070 Ti reaches 341 watts. Be careful, the benchmark used only solicits the GPU. For a gaming use, the global consumption of our configuration will be more important but that with all our graphic cards.
This result is an excellent surprise and is related to the new fineness of the engraving. Our RTX 4070 Ti does much better than the RTX 3070 Ti FE. As soon as we get our hands on the PCAT tool or the new version of the ElmorLabs PMD Power Measurement Device supporting 12+4 pins, we will update this chart by isolating only the graphics card.
[nextpage title=”Overclocking:”]
Overclocking via MSI Afterburner :
Direction the latest version of MSI Afterburner software, version 4.6.4 Beta 3. This is the overclocking software designed by MSI but will be compatible with other GPU brands. The software is really well designed but you will have to go through the advanced mode in order to have access to all the settings possibilities.
To test our overclocking, we are going to concentrate first on the Time Spy benchmark. Let’s go! We choose a maximum fan speed, we set the power limit to the maximum, i.e. to 119% and off we go! At first, we only push the GPU frequency in 50 MHz steps until the benchmark is not running anymore. Once we have a stable GPU frequency, we move on to increasing the memory frequency.
By default, at stock frequencies , our RTX 4070 Gaming OC gets an overall score of 22511 points and 23447 points for the GPU score. The boost frequency reaches a maximum of 2880 MHz with the GPU temperature rising to 59 °C.
In the end, I was able to push the Boost frequency to 3135 MHz and the memory to 1450 MHz. This allows me to obtain excellent scores and thus to gain more points. The limit is around 3150 MHz, if the card goes higher, the benchmark crashes. The global score is now 23659 points and the GPU score is 25001 (+6.6%).
Of course, now that you have the maximum frequencies, you’ll have to test with the fans in AUTO and why not, redo a new ventilation curve in order to hold these new frequencies or close to them. Now, the overclocking performance will depend on your card.
[nextpage title=”The last word:”]
NVIDIA’s smackdown continues?
Well, not really anymore. The faithful of our site certainly remember our article about NVIDIA’s announcements last September. In this brief, which followed the announcement by NVIDIA of the RTX 4090, 4080 16 GB and 4080 12 GB, it is this last model which seemed to us to be the least interesting. Our test of the MSI RTX 4090 Suprim Liquid X confirmed our impressions with performances out of the ordinary and that we had never encountered since we test GPUs. This model has also received a “coup de coeur” from the editors!
Our opinion was a little more mixed on the RTX 4080 Founders Edition. The card is also very powerful but the arrival at the end of the year of the RX 7900 XT and RX 7900 XTX from AMD has changed the situation a bit. The price adjustment by NVIDIA of its RTX 4080 and the very reduced offer of the RX 7900 currently allows team green to once again pull their weight.
Personally, we were expecting the RTX 4070 Ti with performance close to the RX 7900 XT and it was not the case. The gap between the RTX 4070 Ti and its big sister, the RTX 4080, is quite substantial. Now, this is not a bad card, quite the contrary. This new generation is ultra powerful and manages to do the job perfectly in 2160p while we are on a model xx70 Ti. It manages to do almost as well as the RTX 3090 or even the RTX 3090 Ti when Ray Tracing and DLSS are activated!
So why aren’t we completely satisfied? It is of course because of the price, which has been the crucial element for several months now. Without the presence of Founders Edition models, NVIDIA announces in Europe custom cards starting at 899 euros. There are two ways to position yourself in front of this price. Either we can say that our RTX 4070 Ti has performances close to the RTX 3090 sold at its launch at 1599 euros and that it is then a superb offer, or we can note that the RTX 3070 Ti was offered at its launch at 619 euros. No doubt that manufacturing costs have increased in recent months, but for us, what counts today is the price of this new generation against the competition. We will see the prices of the custom models today, but we fear to see a lot of models crossing the 1000 euros mark.
This Gigabyte RTX 4070 Ti Gaming OC:
As for our model of the day, the Gigabyte RTX 4070 Ti Gaming OC, it does the job perfectly thanks to its impressive cooling system. The card doesn’t heat up at all and the fans aren’t heard at all. Finally, when you see the size of the cooling system compared to the PCB and the power consumption of the card, you think that Gigabyte could have come up with a more compact version that would have most certainly saved money.
Despite this, the new design of this “Gaming OC” series is perfectly successful and RGB backlight fans will be delighted with the possibilities offered by the management software.
At the time of concluding this article, we had not yet received any information regarding the price of this version by Gigabyte.
We liked:
- A card to fully enjoy 1440p but also 2160p!
- Performance close to the RTX 3090
- A very well thought out power supply with quality components
- A very good overclocking potential
- The GDDR6X memory is still very docile when overclocking
- Excellent cooling system combined with low noise levels
- Very reliable noise levels
- Excellent power consumption, which has been significantly improved in this new generation
- The quality of the card’s design
- DLSS 3 support
We would have liked :
- A lower price for the launch, which should prove interesting in the face of the competition
- A less bodybuilt version
On paper, this RTX 4070 Ti has everything to please. Performance close to the RTX 3090 of the previous generation, DLSS 3 support, AV1 encoding, excellent power consumption but despite this, this RTX 4070 Ti will have to find its place in terms of price against AMD models that are more powerful than expected but also against the RTX 4080 which could finally have a more interesting price/performance ratio. Let’s hope to see many models available at 899 euros at launch but also in sufficient quantity. Thanks to NVIDIA and Gigabyte for making this sample available.