Is AMD 390 enough for 1080 gaming?

I highly doubt you can go past 3.5GB @ 1080p in far cry 4 unless you're doing something crazy like maxed out MSAA. In fact in most games the 0.5GB issue is only an issue due to higher resolutions such as 4k and future proofing which quite frankly no card in this price range from any company at the moment is fast enough for 4k gaming so whether it has 8GB or 3.5+0.5GB so unless you're planning to crossfire the 390's and run higher resolutions the 8GB is probably irrelevant.

Don't get me wrong the 390 is probably the better card and the 8GB is a nice to have I'd definitely take a 390 over a 970 if they were out when I bought it however I wouldn't trade a 970 for a 390 if single gpu 1080p gaming was my target.

It would be a complete waste of time for similar performance.

 

Played with a lot of cards (test allot). 290 FOR ME is NOT as fast as 390. The MIS I have is a bit faster then my NV 970.

And if you search and bench your self you find the 390 is better then 290x in allot of games.

So if the OP does buy SOMETHING just EBAY the 290 or keep as back up.

 

All this depends on the NVIDIA driver and the method they have used to mitigate this. I doubt there is anything on the hardware level which tells the memory crossbar to "avoid" certain memory controllers (it would mean that they would have to make different silicon, instead of binning the existing one), therefore their solution to this is a software one.
Since I don't believe they would spend the resources to write a tweak for each game out there, they must be using a generic solution, or a kind of algorithm that compensates for situations like that. Like all systems of this kind, it might have imperfections. There is also the fact that games "see" the whole 4GB of VRAM, and try to use it as streaming texture cache in order to speed up loading. Dying Light had to have a developer patch for that, since it was stuttering with the 970. My question is if that developer sidenote will be kept current with games let's say one and a half year for now.

 

Please provide some proof before stating some non-sense.

The R9 390 is an oc'd non throttling with extra voltage and extra 4gb vram. R9 290 vapor-x or pcs+ cards are the ones you should compare the r9 390 to. The reference ones or even the tri-x fall short. Here is some benches to show how a non-reference r9 290 performs compared to a non-reference r9 390 (pcs+). Please keep in mind the driver versions are different and the extra 150 mhz memory clock on the r9 390 card.

http://pctuning.tyden.cz/hardware/g...0-proti-powercolor-pcs-r9-290-v-testu?start=8

Your usual reviews show or test only the reference r9 290 with many not even bothering to bench them and instead just used older test results to save some time, despite driver improvements in the last year over time. Many site used even the test results from the cat 13.12 and compared it to the results of cat 15.15.

The R9 290 performs exactly like an r9 390 clock to clock. An r9 290 clocked well performs even faster than the r9 390x.

 

Test 290 and 390 in the same PC with same core, Vram clocks and 15.7 drivers for both and you'll see the "light"...

Don't use reference 290/290(X) cards, they can't manage temps like custom cooler cards.

If you search you will find benchs for 290 using 14.12 drivers and benchs for 390 using 15.15, this is nowhere near a fair compare if you consider AMD released this drivers (15.15) only for 300 series for one simple reason: these drivers performs better than older ones and can be used to show series 300 as better performing GPU than series 200 who could not use these drivers.

They had a big success doing this as we see so much people fooled by this simple trick.

Bench it by yourself if you like and want to , serious sites who took this "little detail" in consideration reported clock to clock EQUAL performance for both in latest 300 series reviews.

I would like to see this "new" 390" "turbo-tuned" performing better than any of my 290X in ONE game if you are able to find it, it could be funny!

 

Well my gpu at 1150/1550 beats yours at say 1050/1400 maybe

We can do a duel if you want with those clocks

Name the game and the benchmark settings, and put the cpu back to stock, and ill do the same with mine. You can keep hyperthreading on, i doubt it will make a diff. aside from maybe crysis 3 and fc4, maybe witcher 3 in cities and ACU infront of the notre dame.

 

Sorry my 290Xs run at a conservative OC (for them): 1150/1600 in CFX and at 1200/1600 in single GPU mode.

Good GPU PCB and some hundreds spend in WC custom loop make some difference after all.

 

Well i guess i only have one option left to do, Nitrogen cooling and an oc of 1350/1800 then we can do some benching

Whats your voltage offset?

 

Nothing special, the important thing is the temp:

Spoiler

 

The rebadges are rebadges. It depends on which 290 you compared on which 390. The 390s in general have better cooling units. You can't compare a Vapor-X 390 to a vanilla, or a bad-cooler 290.

http://hardocp.com/article/2015/07/20/xfx_r9_380_double_dissipation_4gb_video_card_review/9

 

Can that really be the case, or is just maybe that they ran into different scenarios or maybe stutter during the testing? If the drivers were more optimized for the r9 300 series..... that would kinda suck

Just noticed that they tested a 4gb r9 380 against a 2gb r9 285, maybe thats why fc4 suffers a bit.

 

Yeah that is nothing special......... what the deuce man? :eek2:

Way too high for everyday use. The temps might be good, but the voltage is over doing it.

 

notice hes in LN2 mode, unless he actually has LN2, those are bogus as it heats up the PCB to counteract contact freezing...

 

He's 41C under load apparently. He has an external radiator, and his cards are water cooled.

 

not sure why hes using the ln2 bios then, that would generate extra heat around the chip.

 

My guess is that since he's cooling the **** out of them, he simply flipped the "high performance" switch. I don't know what are the real differences between the bioses on those MSI cards.

 

290X Lightning is not a reference card who has only 5+2 power phases.

It has 12+3 power phases and they are all enabled on LN2 bios mode when the 2x8 pins and the 6 pin power connectors are connected to supply a max of 450W.

The voltage for GPU is balanced over this 12 dedicated phases while on LN2 mode, in normal mode less phases are used.

 

Hi guys,
I want to upgrade my gpu too and i have to choose between those 2 cards too.
And i see you recommend 390 and its fine
but what about dx12?
does 390 supporting what is needed for future gaming with dx12?

thank you

 

The 390 supports DX12.

 

yes but not fully.Is it important?