قالب وردپرس درنا توس
Home https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ Technology https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ AMD Radeon VII Anti-Aliasing & 5K, 8K Benchmarks vs. RTX 2080 | GamersNexus

AMD Radeon VII Anti-Aliasing & 5K, 8K Benchmarks vs. RTX 2080 | GamersNexus



This piece of content is a functional academic exercise, which means that we occasionally show "real-world" cases in use rather than studying how the card is acting and scales to a basic level. This means looking at unseen situations, such as 8K resolution, just to see if measurement improves AMD over NVIDIA; in theory, based on what AMD told us about performance, it's supposed to be, so we just have to prove it. Anti-aliasing takes multiple samples per pixel, often 2x, 4x, 8x, or 16x, and it has a similar effect on resolution resolution performance. The look of AMD's memory bandwidth really benefits from games, what happens to us will be a shrinking or reversal performance of the delta between Radeon VII and RTX 2080 as resolution and anti-aliasing increase, with a stretch or expanding space, favoring RTX, while resolution decreases. We have shown how Radeon VII can close the higher resolution space when we post our review, and talk about how space like the F1

2018 – will grow to 1080p. Now it's time to see if it reverses or closes in high resolution; however, note that the obvious downside is that we enter the territory where content is not enforced in framerate, so it is called an academic exercise.

Superposition AA Testing

Starting shaders set to "Low," textures set to "High" changes and 4K, 5K, and 8K resolutions, we see that NVIDIA leads 27.4% in the baseline, switching to a 27.9% lead – within the error – in the baseline at 5K, then a lead of 28.1% to 8K. There is absolutely no change here from 4K to 8K, at least about scaling, though these resolutions are obviously getting reduced playable as they grow.

We ran a one-off test following the "Extreme" shaders, in this situation. AMD works at an average of 1566 points, with NVIDIA at 1504 points. We have completed this test 4 times and it's averaged, which specifies that this delta is outside the standard deviation.

After learning that the larger differences emerge from the Extreme shaders, the trials ranging from 1080p to 4K. At 1080p, NVIDIA's RTX 2080 plays about 25.8% better than the Radeon VII card at 100% baseline. At 1440p, the advantage of RTX 2080 is rapidly decaying up to 16.1%. The results of the 4K are very close and create a scattered verse, but the RTX 2080 has ended really outperformed by Radeon VII here, with Radeon VII holding a pair of percentage lead points. Now, in terms of "real" framerate, the benchmark runs in about 11-12FPS AVG. It's not playable, it's clear, and that advantage is not really realistic because these settings are not realistically used, but it's interesting to see how two positions can be flip to another & 39; various test scenarios

GTA V AA Testing [19659004] GTA V has many anti-aliasing options, so we're not doing the same MSAA from 2x to 8x and then enable 8x reflections MSAA. Let's start this with just one AVG FPS chart in different settings, which best illustrates how the space changes each test.

With 4K and Very High and Ultra settings for everything, 2x MSAA puts us at 58.8FPS AVG for 2080 and 51.4FPS AVG for Radeon VII, but the more important indicator is that 2080 has 14.4 % lead over Radeon VII card. At 4x MSAA, this lead does not change significantly and goes down to 14.1%, which is effective in error. At 8x MSAA, we begin to see some movements. AVG FPS is 27.5 at Radeon VII and 31FPS at RTX 2080, with improved RTX 2080 at 13%, from the first 14.4%. By adding an anti-aliasing 8x MSAA reflection, we see the difference in change with a 10.8% advantage for 2080 FE. Looking at the overall line graph, we can see that AMD starts recovering some losses towards the extreme end of the scale, it's just that we're on unplayable territory for other reasons.

For reference, with the test value passed we run, the standard run-to-run deviation is about 0.3FPS AVG. These results are very accurate.

Firestrike AA Testing

Firestrike remains one of the best tools for this kind of man-made work. For these, we look for GT1 and GT2 separately. The GT1 loads heavily on the GPU with polygons and tessellations, performs shadows and crunching lighting, and uses compute shaders for post-processing and particle physics. GT2 is too heavy in calculating the shader workload and greatly increases pixels processed in each frame, but reduces work load by more than 50%. GT2 should run better on Radeon VII than GT1, relatively speaking.

Starting with GT2, we see that stock Firestrike Ultra settings have NVIDIA below AMD performance, with 86.7% of baseline Radeon VII 100% performance. This is unique in this benchmark, hitherto. At 2x MSAA, this interval drops to 90.7% of the baseline performance. 4x MSAA brings us to 94.6% of the baseline, with 8x MSAA finally allowing RTX 2080 to surpass AMD and hold 104.7% of the baseline performance.

It makes sense that AMD is usually the most competitive GT2, where the compute workload and memory are remarkable. As with why NVIDIA starts pulling once MSAA is brought to 8x, we are working with the 3DMark team and others to try and better understand specific behaviors. Our view is that this may be related to memory compression in NVIDIA, or perhaps identifying the side of anti-aliasing in particular implementation of this software, or other implementation level advantage.

For GT1, where tessellation and polygons are heavier, we see two cards starting evenly under the Firestrike Ultra settings, with NVIDIA getting an extra 16% on 2xMSAA, 21% to 4x MSAA, and 27% to 8x MSAA.

At 8K with 8x MSAA, framerate levels, but finally seeing Radeon VII pull far ahead of RTX 2080. This is what it takes to completely deplete the memory of RTX 2080, at this point it started choking as it took and dumped memory.

Far Sumigaw 5 HD Texture + AA Trial

Far Cry 5 is next. For this one, we tried the SMAA, TAA, and HD Textures under various situations.

At the first 4K test and using HD Texture at SMAA, we measured NVIDIA at 62.8FPS AVG and AMD at 60.6FPS AVG, putting NVIDIA at about 3.5% earlier. The 1% lows are similarly spaced, such as 0.1% lows.

To test scaling with different settings, we also ran a test with reduced VRAM consumption but other settings. We can switch to the chart now. Results, disabling HD Textures, are similar in results with HD texts. Not surprisingly, we are in the margin of error, because the textures will not actually affect performance unless VRAM becomes a limitation. On both devices, counts, experience has not changed from the previous chart. Qualitatively, the output is the same for each device. Again, NVIDIA is about 3.3% faster here.

Instead of switching to TAA, performance drops slightly for each device, but NVIDIA still maintains a 3.3% advantage over AVG FPS. HD Textures and anti-aliasing changes to Far Cry 5 are not enough to reduce or change the performance delta from card-to-card.

Conclusion: Not Full What We Expect

Our theory is not completely correct. It seems we have forced it in some instances, but the Firestrike throws us on the path. We have come to the 3DMark team and NVIDIA engineers to test and better understand the specific performance scenario. An opportunity where we see a huge increase for Radeon VII is when we have exhausted the VRAM on RTX 2080, made by operating Firestrike Ultra at 8K resolution and 8x MSAA. It is, of course, totally unobtrusive, and thus we point again to this pronunciation is actually a work-only exercise, not one particularly practical. At 8K / 8xMSAA, we see that Radeon VII operates around 4.7-4.8FPS AVG at 2080 & # 39; s ~ 1-1.7FPS AVG. This is a "big" percentage improvement, and since we've exhausted the framebuffer in 2080, it's clearly not possible to play on any of the devices. Whatever it is, it is interesting data, and it may help to better understand other card features in future tests.

Editorial, Test: Steve Burke
Video: Andrew Coleman


Source link