The difficulties of choosing a budget video card on the example of RX 460

Like many of you, I like to play computer games in the evening. Time for this noble occupation, unfortunately, is sorely lacking, but I try as much as I can. Most of them prefer rather old toys, but sometimes you want to please yourself with something new and beautiful. Perhaps the final nail in the lid of the coffin was DOOM, when he saw that the saliva immediately began to flow.

But what to do if you have a GT 240 512Mb on your computer, on which Civilization V works in half with a grief? If even on a tablet with a poor Z8300, this game goes better? In general, the upgrade has been brewing for a long time, and the release of new accelerators in the summer / autumn of last year said: It's time.

image

Why Radeon?


The first reason is the monitor. By a happy coincidence, a copy was purchased with FullHD and FreeSync for ridiculous money. Having had the opportunity to test this technology with a friend before, I was surprised how I used to live then. The second reason is oddly enough work. I am developing an application with a rather high load, part of which is being thrown onto the Intel. But there are some unpleasant nuances. It doesn’t have so many computational units, and only the penultimate generation can do double precision on OpenCL. And adding a video card is much easier than changing a mother + percent on all objects. Nvidia, as you know, cuts the speed of FP64 computing in gaming solutions well, for the best sale of the Quadro series, AMD, in turn, is trying to at least somehow increase sales and cannot afford such ugliness.

image

With the manufacturer decided, what about the price?


Everything is simple. The cheaper the better. The unpleasant buying experience of used components says that we should take the fresh one, and from the fresh and cheap we have only 460th. It would seem that it only remains to choose a suitable caste, but the question remains, what will I actually get? And now you have plunged into dozens and hundreds of reviews of video cards.

What are the bad reviews?


There are a number of grieving factors. First: I like to read the text, I read quickly and with gusto. And the majority of reviews now go out on YouTube and listen to how someone with a nasal voice chews on all the obvious things and diligently pours from empty to empty light. The second factor: Normal reviews go to mid-top iron. Manufacturers want to sell iron more expensive and send the appropriate copies to reviewers. It is difficult to blame them for this, but as a result there are reviews of Rx480 in the summer and now, and it is indicated how much the performance has improved after ~~ garbage collection ~~ in the optimization code by the AMD driver, and I have not found such information on Rx 460. The third factor: Cards are tested in unbelievable conditions. This is partially tied to the previous problem, because each site wants to type a certain base and tests one size fits all. As a result, we have 100,500 fps on the Titan X and 10 on today's hero. But where on the one hand there is standardization, on the other there is anarchy. Each reviewer makes up his own name for the unscrewed graphics settings. High, Ultra, Very High, Max, etc. And the saddest thing is that most of the anti-aliasing is forgotten. Many also do not consider it their duty (or do not know how) to indicate the minimum FPS. And he plays a lot more than some “average” on budget decisions. After all, what's the use of you with 30 medium, if you have 60+ when running ahead, and with a sharp turn 5? But we have to work with what we have, and here we have: that anti-aliasing most specify forget. Many also do not consider it their duty (or do not know how) to indicate the minimum FPS. And he plays a lot more than some “average” on budget decisions. After all, what's the use of you with 30 medium, if you have 60+ when running ahead, and with a sharp turn 5? But we have to work with what we have, and here we have: that anti-aliasing most specify forget. Many also do not consider it their duty (or do not know how) to indicate the minimum FPS. And he plays a lot more than some “average” on budget decisions. After all, what's the use of you with 30 medium, if you have 60+ when running ahead, and with a sharp turn 5? But we have to work with what we have, and here we have:

Enough water, give the numbers.


List of reviewed reviews
www.overclockers.ua/video/sapphire-nitro-rx-460-4g-d5/all
www.overclockers.ua/video/asus-rog-strix-rx460-o4g-gaming/all
ru.gecid.com/video/asus_strix-rx460-o4g-gaming
ru.gecid.com/video/sapphire_nitro_radeon_rx_460_4g_d5/?s=all
www.tomshardware.com/reviews/amd-radeon-rx-460,4707.html
www.techpowerup.com/reviews/ASUS/RX_460_STRIX_OC/10.html
3dnews.ru/938654
itc.ua/articles/obzor-videokartyi-asus-rog-strix-rx460-o4g-gaming
www.thg.ru/graphic/obzor_amd_radeon_rx_460/obzor_amd_radeon_rx_460-01.html
www.ferra.ru/ru/system/review/sapphire-nitro-radeon-rx-460-4g-d5
www.guru3d.com/articles_pages/asus_radeon_rx_460_strix_gaming_4gb_review,10.html
www.pcworld.com/article/3109542/components-graphics/sapphire-nitro-rx-460-oc-review-a-classy-capable-low-cost-graphics-card.html?page=2
occlub.ru/testodrom/2730-obzor-asus-radeon-rx-460-strix-kakoj-on-mladshij-polaris?showall=1
www.ixbt.com/video4/asus-radeon-rx-460-strix.shtml (сдесь замеры в разрешени 1920/1200, нормализованы до 1920/1080)
www.osp.ru/pcworld/2016/09/13050201
root-nation.com/featured/sapphire-nitro-radeon-rx-460
hexus.net/tech/reviews/graphics/101773-asus-gigabyte-sapphire-xfx-radeon-rx-460-compared
и гость сегодняшней программы: обзор с nvidia-ориентированного сайта:
nvplay.ru/videocards/geforce-gtx-1050-protiv-radeon-rx-460-sravnitelnoe-testirovanie

В большинстве своем видеокарты протестированы на топовых процах Intel с бешеным количеством ядер и частотой выше 4.0Ghz. Самый убогий затесался i5-3xxx с частотой 3.3, но у не думаю, что остальные комплектующие хоть как-то влияли на карту этого уровня. Свои кастомы рассылали Sapphire и Asus, частоты которых выше референсных, но итоговые FPS я привел нормализованные до 1200 МГц. Карты я рассматривал с объемом памяти 4Gb, младшие версии в текущих реалиях не интересны.

image

* Settings: Ultra MSAA x2, Ultra, Max, Ultra 4x MSAA * The

peak value of the reviewer with 3dnews, who guessed to remove anti-aliasing, otherwise the difference is not critical and the picture is the same for everyone

image

* Settings: Ultra FXAA, HIGH TSSAA 8TX, and some were just too shy to point out *

The difference is not critical, especially since the picture under the volcano is much better:

image

The first decent scatter. It can be assumed that the peak midpoint is the absence of smoothing, and the authors did not wish to indicate whether it was used. The most recent result is remarkable. Review for January of this year, and the settings are the same as in the penultimate: Ultra TSAA 8x

image

* Settings: Ultra HBAO +, Max (ray medium), TAA 16xFA, Ultra TAA 16xAF HBAO +, Max *

The difference reaches 15 frames. Not bad? The first failure is a review from the site nvplay.ru. In addition to the low average number of frames, they also dropped the minimum to 21. The second failure is the results of ixbt.com, because they measured at a resolution of 1920x1200. The column following the failure is the normalization of this result.

image

Here everyone is more or less the same, and osp.ru just decided to set the settings to Normal. Naturally the rest of the polls are Max, Ultra and so on.

image

20 frames of difference! All reviews on this game are for August / September, so the influence of drivers is minimal. All of them have Max, ultra and a lot of smoothing. The biggest results are at www.tomshardware.com , the smallest are at ixbt.com, although their settings are not the most twisted. To believe is not clear.

image

Another joy is the Witcher difference almost in half, of course the worst result is at nvplay.ru. The highest again at www.tomshardware.com .

Review results


You can not trust anyone. To be honest, after reviewing the clouds of reviews, I no longer strongly believe that someone else honestly chases all the tests. the maximum includes a map to look at heat / consumption and drive off benchmarks. Those games that have a built-in benchmark give out + - everyone has the same results, the rest watch other reviews and randomly throw + a couple of shots. I really hope that this is not so, but the feeling is just that.

The result on the map: the map as a whole is good for its money, and since myopia automatically adds anti-aliasing to the picture, I can let it play even at very high settings, which I did not allow for a long time. And the calculations are still waiting, since AMD drivers under Linux persistently do not get up.

UPD:Updated most graphs from conditionally unplayable 20 frames to the ceiling in most games 50

Also popular now: