ASRock RX 590 Phantom Gaming X 8 GB

[W1zzard]

Priced at $249, ASRock's Radeon RX 590 Phantom Gaming is the most affordable RX 590 at this time. This dual-fan, dual-slot card includes the highly popular idle-fan-off feature many GTX 1660 Ti cards are lacking, but faces strong competition from NVIDIA's latest card otherwise.

Show full review

 

[danbert2000]

Essentially, this card would have been a good deal a year ago if it were just competing against the 1060 and the 1070, but with the 1660 Ti out now and more 16xx cards coming, it's just not worth the money or extra power consumption. I don't think I've ever seen an AIB AMD GPU that got better power consumption than the reference version, so that's pretty impressive. If they dropped the price to $200 I could see a place for this card, at least until Navi comes out.

 

[jabbadap]

No power numbers for 1660ti(No reference model, or?). Overall not a bad card and with that price and game bundle quite reasonable offering(Thought in Europe there's better RX 590s for cheaper). 1660 should be out next week with $219 msrp and will put a little pressure on the prices again, so lovely competition ahead on the mid range.

Essentially, this card would have been a good deal a year ago if it were just competing against the 1060 and the 1070, but with the 1660 Ti out now and more 16xx cards coming, it's just not worth the money or extra power consumption. I don't think I've ever seen an AIB AMD GPU that got better power consumption than the reference version, so that's pretty impressive. If they dropped the price to $200 I could see a place for this card, at least until Navi comes out.

I don't think that is from the reference card, there's no reference RX 590. Looking at the noise graphs "Reference" noise figures are taken from XfX RX 590 Fatboy.

 

[ArbitraryAffection]

This is the 590 i would get, if i was going to buy one. Then i get matching GPU and Mobo, but interestingly Asrock + Sapphire is also special as my Motherboard POST logo says ASROCK|SAPPHIRE when i have my 570- installed and i get a special "Sapphire boost "button XD havent pushed it yet...

Anyway the 1660ti is flat out a better buy, this needs to be $199 to make sense imo.

 

[Xuper]

The board is very similar to RX 480 Reference ( Top = Asrock RX 590 ) , Bottom ( Reference RX 480 )

 

[0x4452]

Looking at these latest performance numbers, I am surprised that the 1060 has gained on the 580 by 3% over the last 1.5 years. So much for the FanWine

Aug 2018, 580 is 1% faster than 1060:

March 2019, 580 is 2% slower than 1060:

 

[ArbitraryAffection]

Looking at these latest performance numbers, I am surprised that the 1060 has gained on the 580 by 3% over the last 1.5 years. So much for the FanWine

Aug 2018, 580 is 1% faster than 1060:

March 2019, 580 is 2% slower than 1060:

Different games I guess

 

[W1zzard]

(No reference model, or?

Yeah that's the problem, I'm checking from time to time if there is any new model that's as close as possible to reference

 

[kastriot]

Nice review but those cards are last train before navi and they should go for 199$ so that would empty their stocks and dont forget myriad of mining rx570/580 card going from 100$ and upward.

 

[xkm1948]

Nah this card aint worth over $200 That horrible efficiency comparing to 1660Ti.

 

[Athlonite]

ASRock Radeon RX 590 Phantom Gaming X OC 2xHDMI 2xDP 8GB $475.47NZD Yeah there's no compelling reason in that price for me to upgrade from my Sapphire Nitro+ RX580 OC 8GB

 

[Deleted member 158293]

Games make it a good value, or take out the games, lower price by +/- $70, new nice price point.

 

[Athlonite]

Games make it a good value, or take out the games, lower price by +/- $70, new nice price point.

The games are only good value if that's the games you want to play otherwise they're not good value and prices here are don't seem to differ even if no games are included

 

[eidairaman1]

Looking at these latest performance numbers, I am surprised that the 1060 has gained on the 580 by 3% over the last 1.5 years. So much for the FanWine

Aug 2018, 580 is 1% faster than 1060:

March 2019, 580 is 2% slower than 1060:

ASRock Radeon RX 590 Phantom Gaming X OC 2xHDMI 2xDP 8GB $475.47NZD Yeah there's no compelling reason in that price for me to upgrade from my Sapphire Nitro+ RX580 OC 8GB

Considering you can oc them and bios mod them

 

[B-Real]

Essentially, this card would have been a good deal a year ago if it were just competing against the 1060 and the 1070, but with the 1660 Ti out now and more 16xx cards coming, it's just not worth the money or extra power consumption. I don't think I've ever seen an AIB AMD GPU that got better power consumption than the reference version, so that's pretty impressive. If they dropped the price to $200 I could see a place for this card, at least until Navi comes out.

Actually it's $10-20 more expensive than a GTX 1060, offering +10% performance and giving 3 AAA games worth of $180. So this is a pretty good deal.

Looking at these latest performance numbers, I am surprised that the 1060 has gained on the 580 by 3% over the last 1.5 years. So much for the FanWine

Aug 2018, 580 is 1% faster than 1060:

March 2019, 580 is 2% slower than 1060:

As others said, different games and the problem with the Metro benchmark draws some attention to this chart. And f. e. if you check Hellblade:

https://tpucdn.com/reviews/ASRock/RX_590_Phantom_Gaming_X/images/hellblade_1920-1080.png
https://www.techpowerup.com/reviews/Performance_Analysis/Hellblade_Senuas_Sacrifice/4.html

RX580 got 4 fps slower than 1,5 years ago?

Nah this card aint worth over $200 That horrible efficiency comparing to 1660Ti.

LOL, so you say the RX580 8GB doesn't worth the $190-200? You green eyed guys can only come up with efficiency. There is a $280 card with 1070 performance with 0 games and a $250 card which has a ~25% disadvantage meaning it would need to cost around $225, but do not forget you get 3 games worth of $180 out of which 2 were released and are acclaimed by critics and players, and an upcoming game which is supposed to be very good - while you get 0 games with the 1660Ti. So you basically get the card for $70 if you are interested in all 3 games, and even you are interested in only 1, you get the card for $190.

 

[IceShroom]

Like to see power draw of RTX cards running raytraced game. Without raytracing on only half(or 40%) die is active.
Performence is poor on games using Unreal engine. Enteresting.

 

[EatingDirt]

Looking at these latest performance numbers, I am surprised that the 1060 has gained on the 580 by 3% over the last 1.5 years. So much for the FanWine

Aug 2018, 580 is 1% faster than 1060:

March 2019, 580 is 2% slower than 1060:

As others have said, different games. An obvious specific example is Dragon Quest XI, where AMD cards perform 1-2 tiers below what they do in basically every other game, other Unreal Engine games included.

https://tpucdn.com/reviews/ASRock/RX_590_Phantom_Gaming_X/images/dragon-quest-xi_2560-1440.png

It's common knowledge that Unreal Engine games perform better with Nvidia cards, but DQXI is an outlier even from an Unreal Engine standpoint. Even Darksiders 3, also an Unreal Engine game included in the benchmark suite, shows the AMD cards at least somewhat close to their price-tier counterparts.

https://tpucdn.com/reviews/ASRock/RX_590_Phantom_Gaming_X/images/darksiders-3_2560-1440.png

I honestly don't think DQXI belongs in a benchmark suite. The way it performs on AMD cards isn't even representative of other Unreal Engine games.

 

[newtekie1]

As others have said, different games.

Also, margin of error.

 

[0x4452]

As others have said, different games.

That argument has the opposite effect, because newer games run better on fine wine, right? So different games are helping the 580 ;-)

 

[ArbitraryAffection]

That argument has the opposite effect, because newer games run better on fine wine, right? So different games are helping the 580 ;-)

Finewine is a bit of a myth these days honestly. It only really happened with Tahiti and Hawaii. This is because, i think, GCN was relatively new at the time of the former and of course it gained a lot as AMD learned to get the most of the architecture. In addition Kepler hasn't done as well in recent titles compared to how well Maxwell+ does. Hawaii and Tahiti both competed vs Kepler mostly. 390/X was better because of the VRAM in many titles these days. The same cannot be said for Fury X, it is terrible aging vs 980 Ti in many games, otherwise it hasn't really gained a whole lot, neither has Vega honestly. Maybe 5%. Since Pascal NVidia cards also do really well, I think mainly because of the update to each price point VRAM. Pascal (non GP100) and Maxwell are very similar in design, updates to PAscal driver should benefit Maxwell too. IDK about Turing but I think they aren't TOO dissimilar. Turing imo, looks very good and future proof in feature set and design.

 

[EatingDirt]

That argument has the opposite effect, because newer games run better on fine wine, right? So different games are helping the 580 ;-)

You take off DQXI, and you probably have 1-3% right there. It's an extreme outlier on being 20%+ faster than the 580, which is the reason I'll continue to suggest that it be taken off the bechmark suite as it's not even representative of must UE4 games, which normally favor Nvidia by around 5-10%.

You can make the AMD cards look even worse if you benchmark nothing but UE4 engine & gameworks games.

Alternatively you can make the AMD cards look much better by benchmarking a bunch of idtech/frostbyte/CoD engine games.

 

[W1zzard]

I honestly don't think DQXI belongs in a benchmark suite.

it'll be replaced by another title in the next round of rebenches

Right is without DQXI

You can make the AMD cards look even worse if you benchmark nothing but UE4 engine & gameworks games.

All we need is from gamedevs to stop using those engines, so that new games don't use them, so that I don't have to use them in my benchmarks.

 

[mtcn77]

W1zzard, can you include some -lv1, -lv2 testing? I normally assume the worst, but have since used your frequency monitor charts in the past.
FineWine is not working just because developers don't want their eggs in one basket. Horizon Dawn is one example, they just dropped flipquad just so they wouldn't need programmable rops which one major developer lacks, lol.

 

[W1zzard]

can you include some -lv1, -lv2 testing?

what's that?

 

[mtcn77]

I'm not so sure, at first I thought it was memory timing strap shift.
Reddit
I think it is in Wattman memory section.