XFX Radeon R9 285 Black Edition Video Card Review @ [H]

XFX Radeon R9 285 Black Edition Video Card Review - Today we are reviewing the new XFX Radeon R9 285 Black Edition video card. We will compare it to a pair of GeForce GTX 760 based GPUs to determine the best at the sub-$250 price point. XFX states that it is faster than the GTX 760, but that is based on a single synthetic benchmark, let's see how it holds up in real world gaming.

I hope XFX has fixed their fan issues.

In my mining farm, i've got about 30 XFX DD 290 cards and more than a few have dying fans (running @ 82% 24/7) that are in their terminal "death wobble" phase. A couple sets of fans i've already had to outrightly replace.

WTH with the GPU-Z readings? 64 ROPS with a Pixel fillrate of 62.4GP/s??



Overclocked its even better =O what a monster.. jokes apart.. whats wrong with GPU-Z?

^^^ Tonga only has 32 ROPs (like Tahiti). GPU-Z smokin' something.

IDK where the power saving features are =S.. the overclocked R9 285 consume more than a R9 280 even taking into consideration differences on hardware from the reviews.. David's system without card = 62W, grady's system without card 95W. Overclocked 285 used in David's system = 394W, overclocked 280 used in Grady's system = 337W but also a 285 in the same grady's system where he test the 280 at stock the 285 consume more power... a big woooot there... Kill-A-Watt precision on float?

David what method do you use to push the cards and take the power consumption?. Grady explain he use real world gaming but you don't...

Araxie said:

IDK where the power saving features are =S.. the overclocked R9 285 consume more than a R9 280 even taking into consideration differences on hardware from the reviews.. David's system without card = 62W, grady's system without card 95W. Overclocked 285 used in David's system = 394W, overclocked 280 used in Grady's system = 337W but also a 285 in the same grady's system where he test the 280 at stock the 285 consume more power... a big woooot there... Kill-A-Watt precision on float?

David what method do you use to push the cards and take the power consumption?. Grady explain he use real world gaming but you don't...

Click to expand...

could simply be 2 different cards at 2 different voltages, cpu could be at different voltages throwing it off, one PSU might be running at a lower efficiency than the other.. there are a lot of different variables that could cause the numbers to be different.

sirmonkey1985 said:

could simply be 2 different cards at 2 different voltages, cpu could be at different voltages throwing it off, one PSU might be running at a lower efficiency than the other.. there are a lot of different variables that could cause the numbers to be different.

Click to expand...

im well aware of that... thats why I mentioned about the 62W system without GPU (more power consuming 285)... vs the 95W system without GPU (lower consuming 280 overclocked but also in the same system Higher consuming 285 vs 280 in the same system at stock speeds) the 62W system have a 33W of advantage and even with that advantage the total system consumption its greater there...

Araxie said:

David what method do you use to push the cards and take the power consumption?. Grady explain he use real world gaming but you don't...

Click to expand...

Power numbers are observed on my kill-a-watt during the course of doing the runthrough data collection for the evaluation. Basically, as I'm playing each game, I'm glancing at the kill-a-watt (it is positioned right next to the screen) so I can see the power draw.

I was a bit surprised to see the numbers as high as I did with this card - given the messaging around increased efficiency as a benefit of the Tonga GPU, I was expecting something like the 270X/270 level of power, but this felt more like the 280/7950 for the wattage being observed.

Power numbers are difficult to do as the CPU usage within games can swing them up to you 100w in one direction or another. There's also the purported variances related to the kill-a-watt itself, efficiency of power supplies (which can even vary at different load levels) and probably a few pinches of magic smoke that can cause these numbers to differ. I would say they're going to be most accurate if you let each power number stand on its own within a review, as cross comparing against reviews (especially by different authors) will lead to your head getting permanently tilted sideways trying to figure it out.

atp1916 said:

^^^ Tonga only has 32 ROPs (like Tahiti). GPU-Z smokin' something.

Click to expand...

It probably was the culprit that finished the bottle of two buck chuck that I was working on during the course of the evaluation ;-).

Thanks for the answer David, the fact that those numbers even attracted your attention are a good interesting point about power consumption of Tonga cards, specially if the rumored 285X its coming i don't see the point more than bridgeless Xfire and all of those minimal features..

Do you need to increase core and memory in parallel or in relation to each other. IE is there is a problem if you only increase one or the other - because of introducing something like wait states that might actually degrade performance? or is that an unfounded concern?

Archaea said:

Do you need to increase core and memory in parallel or in relation to each other. IE is there is a problem if you only increase one or the other - because of introducing something like wait states that might actually degrade performance? or is that an unfounded concern?

Click to expand...

Wow. Epic thread necro. Well played.

Generally, core clock makes more of an overall performance difference than memory clock does, thus, we typically increase core clock to the highest level first, and then boost memory to the extent that it will not negatively impact the core clock.