This is a small article on the AMD Radeon RX 6600 8GB vs Nvidia GeForce GTX 1070 Ti 8GB video cards. I include a hand full of DX 11 and DX 12 games and tests to get an idea of how they compare to each other. I didn’t have much time with the system and didn’t have many games to work with. But I provided what I could.
I built a new computer for my sister in-law recently and got her the MSI Radeon RX 6600 MECH 2X 8G video card with an Intel Core i3-12100f processor. I had an Nvidia GTX 1070 Ti 8GB hanging around so I decided to see how they would compare to each other.
The Video Card’s Specifications
| Video Card | MSI Radeon RX 6600 MECH 2X 8G | MSI GeForce GTX 1070 Ti DUKE 8G |
|---|---|---|
| Core Clock Speed | Boost Clock: 2491 MHz Base Clock: 2044 M Hz | Boost Clock: 1683 MHz Base Clock: 1607 M Hz |
| Streaming Cores | 1792 Unified | 2432 Unified |
| Memory | 8GB GDDR6 | 8GB GDDR5 |
| Memory Clock Speed | 2002 MHz | 1750 MHz |
| Memory Bus Width | 128-bit; 32MB Infinity Cache | 256-bit |
| Memory Bandwidth | 224 GB/s | 256.3 GB/s |
| PCI Express (Bandwidth) | PCIe 4.0 x 8 Lanes (15.75 GB/s) | PCIe 3.0 x 16 Lanes (15.75 GB/s) |
| Release Date | October 13st, 2021 | November 2nd, 2017 |
| My Cost$ | $269 New ($249 after $20 Rebate) | I already had. eBay shows $200 or less Used |
The GPU core’s cannot be directly compared spec for spec do to completely different architectures but included them for reference. The Radeon 6600 only has 8x PCIe lanes but since it uses the v4.0 protocol the interface bandwidth is similar to the GTX 1070ti’s PCIe 3.0 x16 lane interface. If you put the card in an older PCI Express 3.0 system then the interface bandwidth is cut in half to 7.88 GB/s, which could have a small impact on performance. The 6600 has -12.5% less raw memory bandwidth compared to the 1070 Ti. It includes a 32 MB pool of on-die memory AMD calls “Infinity Cache” to help make up for the narrow 128-bit memory bus. The Radeon 6600 is built on TSMC’S 7nm process while the GeForce GTX 1070 Ti uses an older TSMC 16nm process.
One metric I would like to have included in the article but didn’t is the power consumption. I have a Kill-A-Watt meter that can measure total system consumption and would like to have plotted the differences between each card. But a little late for that now. The 6600 would likely have had no problem winning that metric but still would like to have seen the numbers.
Configuration and Test Setup
Below is the setup used for both video cards.
| Component | |
|---|---|
| Processor | Intel Core i3-12100f @ 95 Watt TDP (Otherwise Stock) |
| Motherboard | Asrock B660 Pro RS |
| Memory | 2x8GB Corsair Vengeance DDR4 PC3000 15-17-17-35 |
| System Drive | MSI SPATIUM M390 NVME M.2 1TB |
| Power Supply | Corsair TX650 |
| Operating System | Windows 10 21H2 19044.1949 |
| Video Card Drivers | Radeon 6600: Adrenalin 22.8.2 GTX 1070 Ti: Gameready 516.94 |
The processor is one of the best bang for the buck processors out there which cost about $102 shipped. The motherboard was about $105 shipped. The system uses DDR4 instead of DDR5. I just couldn’t justify spending the extra money (I was shopping this computer in the July time frame) for DDR5 ram and motherboard. While there have been some DDR5 price drops since then, it’s still costs a nice premium over DDR4. Especially for higher speed, lower latency DDR5. Anyways, the processor is an Intel i3-12100f with 4-cores/8-threads using the 10 nm “SuperFin” process with the “Golden Cove” core architecture on the Alder Lake platform with a turbo clock up to 4.3 GHz.
I did increase the processor’s TDP limit to 95 watts from the stock 65 watts to help multi-core performance a bit. The ram settings were set to the XMP profile of 3000 MHz and 15-17-17-17-35 timings. The rest of the system should have no real effect on the video card results.
The benchmark resolution goes no higher than 1920×1080. I didn’t have a monitor higher than 1080p or I would have provided higher resolution results. The cards are better suited for 1080p high preset gaming for the most part anyways.
Benchmark Charts and Results
Below begins the benchmark results. I tried to provide the relevant information on the charts and the benchmark results.
3DMark 11: This is a well known tool made by Futuremark for measuring GPU and CPU performance. It uses multi-threading, tessellation, and the Direct X 11 API’s. I ran the Extreme (1080p) and Performance (720p) presets and plotted the results below.
The RX 6600 has a healthy lead under the “Performance” preset. Almost 20% in the Graphics score and 10% in the total or 3DMark score. Things are very different under the Extreme test (which increases the resolution to 1080p) with the GTX 1070 Ti taking almost a 7% lead in the Graphics test and the total score about 5.5% higher.
So in the more important Extreme test the 1070 Ti was a bit faster. Likely the higher bandwidth being one of the reasons.
3DMark Basic: I used the basic (free) version which uses the Time Spy demo for Direct X 12 performance testing. The full version unlocks more features and demos like Port Royal with Ray Tracing, Wildlife using the Vulcan API’s and so on.
When it comes to this DX 12 test, the RX 6600 wins quite convincingly by almost 20% in the Graphics test and over 15% in the Timespy results.
The newer APIs is where the Radeon 6600 should really shine over the GeForce 1070 Ti.
Borderlands 3: This game supports the Direct X 11 and Direct X 12 graphic APIs. I provided benchmark results from both APIs using the in-game benchmark option.
The first look into an actual game with Borderlands 3 rendering via DX 12 has the RX 6600 with a strong 31% advantage in 1080p Ultra and over 37% in 1080p High versus the GTX 1070 Ti. The gap widens to almost 40% at 720p High and 36% in Ultra.
Rendering via Direct X 11 has the video cards performing much closer to each other. NVidia has always had strong DX 11 drivers and is probably one of the reasons it performs so close to the Radeon in the DX 11 tests. The GTX actually performs a touch better in DX 11 than DX 12 which is totally opposite the Radeon.
The Radeons lead over the GTX drops to only 11% in 1080p High and 5.4% in 1080p Ultra. Using 720p High the lead is 6.6% and to just under 5% using the Ultra preset.
If using the RX 6600 there is no reason not to play the game using the DX 12 code path (bugs/stability aside). DX 12 performs so much better.
Crysis Demo: This is an old, familiar game released around 2007 based on Direct X 9.0. I decided to include this game as an extra data point for running old games on much newer GPU’s. So this can possibly represent to a certain degree what it would be like to play on much older games that uses DX9.0. I set the resolution to the highest that I saw available which was only 1650×1080 and I set the presets to the highest available. I ran the 32-bit binaries to get results from the GPU bench option.
In this game the RX 6600 is a good bit faster than the 1070 Ti. About 15% in average frames, almost 14% in minimums, and 28% in maximum. Really not much more to say. Just included it for those who may be interested in how an old game fairs on these cards.
Final Fantasy XIV: Endwalker: “Endwalker” is a Final Fantasy expansion released in December 2021 although the engine is a bit older. I used the official benchmark utility available on the manufacturer’s website. This engine uses Direct X 11. I ran this benchmark at 1080p with both Desktop (High) and Maximum presets.
There’s not much separating the cards. The minimum fps has both cards performing the same. When it comes to the average fps the RX 6600 is about 6% faster using the maximum preset and 5% faster with the high preset compared to the GTX 1070 Ti.
Shadow of the Tomb Raider – Definitive Edition: A well known title from a very well known game franchise. I used DX 12 for the benchmarks at 1080p and provided results for the highest and high preset.
The Radeon 6600 has no problem outperforming the 1070 Ti. Under the high preset the 6600 has a 32% advantage in average FPS and over 37% in minimum frame rates.
While the Radeon’s advantage is a little less using the highest preset, it still has no problems outperforming NVidia. A 23.8% advantage in average frames, 25% advantage in minimum, and 20% in maximum frames.
So in every metric the 6600 had a solid advantage over the 1070 Ti.
Total War: Warhammer: This game offers an in-game benchmark feature with DX 11 and DX 12 support. I ran DX 11 instead of 12 due to the beta nature of version 12 API support. I used the “DWarfs vs Greenskins” benchmark option at 1080p Ultra settings.
The AMD card again shows a solid lead over NVidia. About 24% in average frames, 12% in minimum, and 28.5% increase in maximum fps.
World of Tanks enCore RT: This is a benchmark demo based on the actual World of Tanks game. The game engine uses Direct X 11. World of Tanks gives the results in points and not in fps. The Ray Tracing option does not use GPU ray tracing features so I didn’t include it.
In the final title the results are pretty close between the two cards, but the 1070 Ti has a 6.7 % lead over the 6600. As with some other Direct X 11 rendering results the 1070 Ti does well.
Overall the Radeon 6600 is a better card especially as you use newer APIs. While it did fine in DX 11 it was DX 12 where is really shined. Although I didn’t have any Vulkan results to share it’s known by other reviews that RDNA 2 does much better in DX 12 and Vulkan.
Even if you currently have a 1070 Ti, choosing a Radeon 6600 could give you some nice gains if you plan on running newer games. If you’re running older games that is DX 11 or older only then the RX 6600 likely wouldn’t be worth the small increase to possibly slightly worse performance. But as always you should use your gaming scenario to determine the games and features you look for in a video card.