Andrei

Well. Whatever happens in the GPU market last couple years, it was always end-users (gamers) who suffered. Pascal was the last decent value offering we've had on the market which we could actually get our hands on. Turing was overpriced with RT and no real competition in high-end. Ampere with RDNA 2 were decent, but then you've got the crypto boom and unless you're lucky and living in US, you could never get your hands on one of those at decent price. Lovelace and RDNA 3 are overpriced again with both AMD and Nvidia trying to recover from crypto hangover capitalizing on gamers as well. Life sucks, but have to get on with it

Thanks to AMD naming conventions I've misinterpreted your post. I thought with 7950 you're talking about GPU, while you probably meant CPU

Definitely NOT asking how you got your hands on 7950 ES

Now that I finally got the 4080 I thought I could add a few more cents to the benchmarks. Maybe some people upgrading from the older hardware will find it useful. I've recently upgraded from Zen 3 5800X to 5800X3D and just now from 2070 Super to 4080. So there will be stats for both upgrades. Setup All tests are done in VR (Reverb G2). By no means those figures are scientific, so treat those as ballparks. The frametimes are taken using using OpenXR toolkit that records results every second. Therefore the resolution isn't all too great and might not show details of e.g. 1% lows. For DCS settings, all tests until indicated otherwise are perfomed with the following. Those are my normal settings for flying online and offline. For VR specific settings PD in DCS is 1.0 and Open XR resolution is downscaled to 65% of max for Reverb G2. null CPU Tests Let's start with CPU upgrade. Those are the runs of Plasma's Torture map which is loading CPU very well. Objectively you don't see too much of a difference between 5800X and X3D. Subjectively V-Cache gives you a much smoother experience. The tests were done in DCS version 2.8.0.32937. 5800X Plasma Torture Map 2070S Average FPS: 42 Average CPU frametime: 15 ms Average GPU frametime: 23 ms 5800X3D Plasma Torture Map 2070S Average FPS: 42 Average CPU frametime: 13 ms (-2 ms) Average GPU frametime: 23 ms (same) 3D cache gives you around 15% improvement in CPU frametime, but as mentioned above, the overall experience is subjectively better due to less stutters. GPU Tests Now to GPU tests. Note that compared to CPU test prevoiusly, this one is using DCS 2.8.1.34437, newer graphics driver and CPU is curve optimized to -20 all cores. So don't compare it directly to previous test. NOTE: When you see "low spikes" in GPU frametimes it is where I pointed my helmet up towards the sky, which of course gives it a nice FPS boost / FT decrease. First one is Marianas benchmark used earlier by Th3ChosenOn3 and xoxen. As CPU frametimes there are quite low, I think it's a good indicator of overall graphics performance. 5800X3D Marianas 2070S Average FPS: 31 Average CPU frametime: 8 ms Average GPU frametime: 33 ms 5800X3D Marianas 4080 Average FPS: 68 Average CPU frametime: 6 ms (-2 ms) Average GPU frametime: 15 ms (-18 ms) As you can see, the FPS and GPU frame times doubled with the same graphics settings. Interestingly enough, the CPU frame time has decreased a bit as well, which suggests that 40 series might be working better with CPU then 20 series. You'll see this behavior repeat itself in another test. Next we turn back to Plasma's map to evaluate overall performance increase in CPU bound scenario. 5800X3D Plasma Torture Map 2070S Average FPS: 40 Average CPU frametime: 13 ms Average GPU frametime: 25 ms 5800X3D Plasma Torture Map 4080 Average FPS: 70 Average CPU frametime: 11 ms (-2 ms) Average GPU frametime: 10 ms (-15 ms) Again, we see that performance has pretty much doubled and in this case we're really becoming bottlenecked by CPU occasionally. However, the CPU frametimes have also improved. High Settings Finally, I wanted to see how does RTX 4080 handle higher graphics settings. I switched OpenXR render to 100% (up from 65%) and DCS graphics as on screenshot below. The biggest change was shadows (terrain = default and shadows = high). 5800X3D Marianas 4080 High Settings Average FPS: 47 Average CPU frametime: 7 ms Average GPU frametime: 21 ms Obviously performance is lower than with original graphics settings, but still more than playable for me. In fact, it's giving a better framerates and notably better picture then 2070S at lower settings. Conclusion Overall, the results were pretty much according to my expectations. Huge improvement with same settings and still better performance with better visuals. Also, it was interesting to see better CPU utilization with 4080. This upgrade was due a loooong time ago. I skipped 2020 due to availability, I skipped 2021 because of miners. At the end of the day I paid same price for 4080 what it would effectively cost me to buy a 3080 a year ago. Is it worth it? For me - I think so.

Вставлю свои пять копеек С появлением MSFS все хотелки по поводу транспортников и полётов из точку А в точку Б в ДКС можно смело убирать в чулан. В MSFS это делается интереснее, реалистичнее и самое главное, для этого там есть вся инфраструктура, в том числе и онлайн. Проблемой остается наличие бортов и ФМ, но я думаю что этот вопрос решится куда быстрее чем в ДКС появятся большие карты, высокодетализованый ландшафт и необходимая аэронавигационная инфраструктура. Для ДКС нужен фокус именно на боевой составляющей, в особенности: 1. Поведение ИИ отдельных юнитов и групп юнитов 2. Симуляция театра военных действий и соответствующую обвязку (та самая динамическая кампания) 3. Инструментарий для создания шаблонов ДК для кампаний в сингле и мультиплеере 4. Возможность эффективно использовать ресурсы современных ПК чтобы всё это великолепие выше работало. И будет нам всем счастье. Работы в этом направлении явно ведутся как показывают новости и предыдущие релизы, но довольно медленно. Что и понятно, так как основной фокус уже десятилетия (если брать еще со времен Фланкера) делался на отдельных ЛА, их авионике и ФМ, а не на развитие "мира".

I only have a WQHD monitor, so 1440p is max. However, I fly exclusively in Reverb G2, although at 65% scale resolution with 2070S. At 100% that thing gets pretty close to 4K, so that should be relevant comparison.

Вот это конечно обидно с точки зрения близких ТВД. Та же карта Сирии и Египта просто просится быть одним большим куском - открывает отличные возможности для дополнительного геймплея и создания довольно-таки интересных сценариев.

Fair point. But gentlemen few post above did the F/A-18C on Marianas map, both and high and low level, which is fairly representative of the GPU performance. Granted, Hornet is not as heavy as Apache, but it's still a full fidelity module. Some maps, and modules like Channel map are so poorly optimised, it doesn't make too much sense testing and benchmarking with - not overly representative of general performance. I guess not even 5090 with I9-15900K will make good framerates there

Правда. Особенно если сервер на другом компьютере. Но возни больно много. И только в том случае, если большой затык именно по процессору. Хотя ладно, о чем это я. В ДКС всегда самый большой затык по процессору

Jumped the gun, bought 4080 at last. Will do some VR comparison tests between 2070S and 4080 when it arrives

Thank you, and @xoxen for taking the time to do this. This is really appreciated. It's so damn difficult to get any good figures for DCS, let alone a side-by-side comparison of latest hardware. Your efforts will not be forgotten

4080 is definitely faster than 3080 Whether it's better as in value for money it a completely different topic though. I'm seeing 4080 back to stock in many places after last week's 7900 series 'fiasco'. Putting that in brackets, because it's not a bad card, just having some teething(?) problems and following Nvidia steps in bad value for money. And that new stock is appearing to come at more reasonable prices. My hope is that after Christmas holiday period is over, they will start discounting it even more so that I could grab one at more 'reasonable' 1200-1300 Euro. Still insane, but it beats current ~1500. Alternatively, AMD might pull their s~t together and make that driver brakethough giving back 10-15% performance. We'll see.

What's interesting is that the general Radeon cards VR behavior exhibits itself here as well. The frametimes are all over the place, resulting in a very "thick" line. That also adds to the stutters.

Looking at the comments on YT, everyone and his dog are now asking for VR tests. People must be thinking that you guys now have a bunch of hardware lying idly around

VR Tests using plasma's map benchmark DCS 2.8.0.33006 Average FPS: 42 Average CPU FT: 12.6 Average GPU FT: 23.2 DCS 2.8.1.34437 Average FPS: 38 Average CPU FT: 14.9 Average GPU FT: 25.2 Frametimes - lower is better. So 2.8.1 seems worse in terms of performance. There can be other variables at play since I got Win11 H2 update and newer nVidia drivers, but the issue with GeForce cards on Win11 has been fixed.

Really good effort, thanks guys! I think those benchmarks you did are really valuable as it's one of those rare occasions that allow us to compare more or less apples-to-apples. The results are interesting indeed. I'd say both cards are pretty close overall. Perhaps 7900 XTX works slightly better with CPU thus the better frame rate when flying over the river which is quite action packed. VR will not be possible to test it seems. I'd care a guess based on prior experience that 4080 will give somewhat more consistent frame times, but it's hard to tell for sure. To me then both cards seem to be a decent option ignoring the price I will probably go for the one I can get based on the availability / price (discounts) and board design in terms of cooling and noise.

Thanks @Th3ChosenOn3 for the benchmark. Really good stuff. @xoxen looking forward to your run with 4080. That might be the final validation I need to choose for one or another card I would expect them to be quite close though.

Good to know! I'm sitting on Corsair 650W and having read the reviews thought that it might be just enough. Still, decided to upgrade to 850W to make it a bit more future proof. I have multiple PCs anyway and it's good to have a PSU in reserve

I'd expect some improvement. But I feel that you might be getting CPU-bound indeed. To make this a bit more fair as someone else mentioned, do another benchmarking run on an empty map. That will give you another validation point, on how much you gain purely in terms of graphics performance. On the same settings, you will gain some FPS, probably not by a huge margin. I'd guesstimate around <=30%. However, your new card will allow you to max out graphics settings while keeping the same framerate. More eyecandy, yay

Try this one: https://cdn.discordapp.com/attachments/915347310567915541/1036740701196664875/plazma_torture_map_2022_flanker.miz It was used in at least one Youtube video which was benchmarking 4090 and I've been using it as well for CPU benchmarks. It's pretty well done - short, but packed with action. Requires FC3 or Flanker. P.S. Link to original benchmark by map's author.

Well, there you go. Your CPU is a limiting factor by a rather big margin. At least in CPU-heavy scenario, which DCS:Liberation is. Now I guess we need to come back on topic

With a high likelihood it does. Moving from 7700K to 5800X has raised FPS from 65 -> 80 in 2D scenario I've been benchmarking. And from 5800X to 5800 X3D has improved CPU frame time from ~14.8 to 12.6 (- 15%). That is on the same graphics card 2070S. But the best thing for you would be to get an FPSVR for Stream VR or OpenXR Toolkit if you use Open XR and see where your frametime bottleneck is.

Ouch.. Wonder if the next release of the drivers will do it any good...

Please do share your findings. For me, the biggest question between 7900 XTX and 4080 is now how it's going to perform in DCS / MSFS VR. While we might get some 'big boys' to test in MSFS, for DCS performance it's impossible to rely on anymore, but fellow community members

Having checked each and every possible review of 7900 XTX I must say I'm really on the fence right now. The upgrade from 2070S was long time due, but in 2020 and 2021 I didn't want to pay scalper prices for 3080 and 6800XT. Well, guess what, in 2022 you pay MSRP+ which is somewhere around the scalper prices back in the crypto days. Now between 4080 and 7900 XTX it's a difficult choice. AMD reference card design is not the best one. I hate coil while and the memory junction temperatures are on a high side. On the other hand, I can already imagine non-reference designs (when they come) will probably ask a premium over the reference. Out here in Europe, you can find 4080 for around 1500 Eur. If 7900 XTX non-ref will be approaching 1350-1400 that will be a much less compelling argument. Performance wise it's a hit and miss with XTX vs 4080 it seems in pure raster. On average Radeon is somewhat faster, but not much. And you can give team red some credit for lousy drivers upon release. Add the fact that nVidia has been traditionally somewhat better in VR in terms of frame times. My use case is mainly DCS and MSFS which occasional other 'AAA' titles that may or may not work with ray tracing. So RT is not really something of a major importance, but of course it's better to have it then not. So while 7900 XTX looks kinda favourable, price-performance wise, it will all largely depend upon the asking price for the non-reference models. P.S. I'd buy a 4090 in a blink of an eye if it was available in Europe for the advertised equivalent of $1600. But since we land in the land of socialism with crazy high VAT, at Eur 2200+ that is not really an option