Asus R9 290x DCII (Guru3D Review)

[Imouto]

Because its running at higher clocks out of the box and doesn't throttle. R290X throttles down to ~850mhz in some games and still consumes so much power.

There's toms and a few other sites that looked at power use with aftermarket coolers, it drop big time, google it I am too busy watching Doge fly to find it now.

Then it comes down to how important that power consumption is for you. I'd trade efficiency for acoustics any day of the week. Same with efficiency for performance. What I can't stand is noise. Anyway this is what I found with a quick Google query:

http://www.tomshardware.com/reviews/radeon-r9-290x-hawaii-review,3650-29.html

But it is hindered by the crappy reference cooler:



So I really need your help here since all the sites let the card throttle. I'm not saying that efficiency isn't worse at higher temps, ofc it is. I just want to know how much.

 

[KompuKare]

So I really need your help here since all the sites let the card throttle. I'm not saying that efficiency isn't worse at higher temps, ofc it is. I just want to know how much.

HT4U.net tried an R9 290 with both the AC Xtreme III and the Prolimatech KM 26:

http://ht4u.net/reviews/2013/arctic_accelero_xtreme_iii_meets_amd_r9_290/index6.php
http://ht4u.net/reviews/2013/r9_290_im_griff_prolimatech_mk-26_black_im_test/index7.php

In summary their results were:
Reference: 289.4W
MK26: 260.5W
Xtreme III: 257.8W

That's with their temps being:
Reference: 94.0°C
MK26: 58.0°C
Xtreme III: 60.0°C

The Tom's article was this one:
http://www.tomshardware.com/reviews/r9-290-accelero-xtreme-290,3671-4.html
But I don't see any power consumption figures there.

Anyway, the Sapphire R9 290 Tri-X OC and Asus R9 290X DirectCU II OC reviews don't show any savings but then they are overclocked and seem to have different power delivery etc.

 

[jj109]

Seems like they believe their results show most of the heat does exhaust out the back. Interesting!

Having owned their hybrid fan DCU II (see above), I can say with confidence that it does not exhaust that much hot air out the back. The hot air that went through the fins mostly bounces off the lower IO panel and back into the case.

 

[Silverforce11]

Then it comes down to how important that power consumption is for you. I'd trade efficiency for acoustics any day of the week.

Cooled with AIO, with NZXT G10 mod. Lots of users use the zip tie method or dwood brackets to same effect for dirt cheap.

http://www.legitreviews.com/nzxt-kr...oler-review-on-an-amd-radeon-r9-290x_130344/5

They inquired AMD about the power use drop and the theory is less leakage at low temps.. or rather, what AMD didn't admit, is by letting the cards run at 95C, its causing excessive leakage.

 

[jj109]

Excessive leakage, electrical, and physical wear. High temperatures are just bad news for ICs in general, "high" power ICs like GPUs in even more so.

If I recall correctly from my VLSI course, a 20C increase doubles the electromigration rate of aluminum interconnects. There are layout tricks of mitigate electromigration, but non-fatal defects in the die could be pushed over the edge and become fatal over time. It's just a good idea to run cooler if long-term reliability is a concern.

 

[Imouto]

Anyway, the Sapphire R9 290 Tri-X OC and Asus R9 290X DirectCU II OC reviews don't show any savings but then they are overclocked and seem to have different power delivery etc.

That's the whole point of reference temps being useless. Factory overclock, temp targets and such make it. Temp is only a factor in the whole equation. Temps at what price? Oversized coolers? Noise? Closed loop?

I'm still pretty sure that the sweet spot (for me) between temps, power efficiency and noise with this DCU 290X is with temps a tad higher hence the 90ºC figure. But ASUS should remove the 20% cooler sticker.

They inquired AMD about the power use drop and the theory is less leakage at low temps.. or rather, what AMD didn't admit, is by letting the cards run at 95C, its causing excessive leakage.

Do you realize that this closed loop is keeping it at 44ºC? Same goes for the Accelero or the MK-26. Those temps are out of reach for any custom card at reasonable noise levels. Of course such temp deltas lead to massive power usage drops but that's not what I'm talking about.

If you go for lower stock temps you have to sacrifice noise levels. If you go for higher temps efficiency is worse. Just take a look at your own data. When should we stop? 50ºC at 70% fan? 45ºC at 100%?

The trend is going for lower temps So the temp figures in reviews are useless unless you factor noise and efficiency too. Efficiency figures at different temps would be nice data to have.

 

[Silverforce11]

T
Do you realize that this closed loop is keeping it at 44ºC? Same goes for the Accelero or the MK-26. Those temps are out of reach for any custom card at reasonable noise levels.

Accelero is 60C. Users with GELID@70C see the same result, there's heaps using aftermarket, AIO zip ties on OCnet forums. There was an official review (from memory, swedes or something similar) that analyzed power use at different temps on the R290X, and they found a sudden spike at 90C onwards.

Here's another for you to compare:

Stock R290X, with throttle clocks to ~850mhz:

http://www.pcper.com/reviews/Graphi...ges-Performance-Fan-Speeds-R9-290X-and-R9-290

Compare to power use versus Asus card running at 1050mhz power use:

http://www.pcper.com/reviews/Graphi...Graphics-Card-Review/Cooling-and-Clock-Benefi

Moral of story for AMD is: Don't let your GPUs run hotter than 90C, it ruins efficiency and requires more airflow to cool the extra power used.

 

[Imouto]

Accelero is 60C. Users with GELID@70C see the same result, there's heaps using aftermarket, AIO zip ties on OCnet forums.

Still out of reach for custom cards. If we go by ht4u data by cutting the temps by 30% you have a nice 16% drop in the power figure.

By legotreviews data cutting temps by 50% lead to a nice 44 watt drop for the whole system. Pretty much the same as ht4u.

You don't have that much room with a custom card.

Going back to this card my bet is that ASUS could have done a quieter card without throwing the efficiency out the window.

There was an official review (from memory, swedes or something similar) that analyzed power use at different temps on the R290X, and they found a sudden spike at 90C onwards.

I'll give you that. Maybe my 90ºC figure was beyond the spike threshold but sure ain't more than a couple of degrees and far from the 78ºC from HC review.

The point is that temps are totally up to the assembler. Taking a look at some card and saying "look at them temps" is meaningless without taking noise and efficiency into account.

 

[Skurge]

Accelero is 60C. Users with GELID@70C see the same result, there's heaps using aftermarket, AIO zip ties on OCnet forums. There was an official review (from memory, swedes or something similar) that analyzed power use at different temps on the R290X, and they found a sudden spike at 90C onwards.

Here's another for you to compare:

Stock R290X, with throttle clocks to ~850mhz:

http://www.pcper.com/reviews/Graphi...ges-Performance-Fan-Speeds-R9-290X-and-R9-290

Compare to power use versus Asus card running at 1050mhz power use:

http://www.pcper.com/reviews/Graphi...Graphics-Card-Review/Cooling-and-Clock-Benefi

Moral of story for AMD is: Don't let your GPUs run hotter than 90C, it ruins efficiency and requires more airflow to cool the extra power used.

That last part isn't true, the larger the temperature differential between the core and the heatsink/air. The more heat is transfered.

 

[Lonyo]

I think he means it's a compounding problem. The hotter the card runs, the more power it uses, the more heat it makes, the more cooling you need. More power = more heat. Hotter temps = more power.

 

[xodusgenesis]

Anyone know when this is being released? I can't find it in any stores in the US

 

[SlickR12345]

Great card, I like how it blows the GTX780 TI in the water in several games and in the others its pretty much neck and neck with it at $150 cheaper price.

Great value card, now cool and silent.

 

[Lonyo]

Anyone know when this is being released? I can't find it in any stores in the US

Overclockers.co.uk in the UK has the following listings:

Sapphire: 27/12/2013
Gigabyte: 17/1/2014
MSI: 10/1/2014
Asus: 17/1/2014

So probably within the next 3 or 4 weeks at most, unless something goes wrong somewhere.
OC pulled in the Sapphire date because initially I think that was mid Jan as well.

 

[blackened23]

Anyone know when this is being released? I can't find it in any stores in the US

Mid January according to Asus.

 

[Psyside]

Nvidia advocates should stop posting that HardwareCanucks graphs of Crysis 3, because FXAA is used instead of MSAA, when MSAA is used, this is what you get.

 

[Skurge]

Nvidia advocates should stop posting that HardwareCanucks graphs of Crysis 3, because FXAA is used instead of MSAA, when MSAA is used, this is what you get.

Pretty much moot since that isn't playable really.

 

[Psyside]

Pretty much moot since that isn't playable really.

Moot? sure it is, because the Nvidia card didnt win.

For me moot is using useless FXAA on 700$ card.

 

[blastingcap]

NV won't see a dime from me until they make their top three tiers of cards come with at LEAST 3GB standard, preferably 4GB+. Making people pay extra for what will likely become mandatory VRAM as next-gen console ports come out is not cool.

/off-topic

 

[Enigmoid]

Moot? sure it is, because the Nvidia card didnt win.

For me moot is using useless FXAA on 700$ card.

Uh no..... Its because 23 fps is unplayable no matter how you cut it. Useless is paying $700 for 23 fps.

Its like testing igp's at 1080 ultra + 4x MSAA. It doesn't matter if one is 100% faster than the other because 2 vs. 4 fps is completely unplayable.
Highest Playable settings are what matters.

 

[moonbogg]

NV won't see a dime from me until they make their top three tiers of cards come with at LEAST 3GB standard, preferably 4GB+. Making people pay extra for what will likely become mandatory VRAM as next-gen console ports come out is not cool.

/off-topic

They've been doing that for several generations now. The 570's didn't have enough, neither do most of their current lineup. A year from now 3gb will be right on border line, just like 2gb is right now even requiring the lowering of some settings despite having enough GPU power.
Also I wouldn't expect Nvidia to actually release cards that make sense in the future with adequate Vram. I expect more skimping and cheaping out forcing more frequent upgrades. That's why I am going AMD.

 

[Psyside]

Uh no..... Its because 23 fps is unplayable no matter how you cut it. Useless is paying $700 for 23 fps.

Its like testing igp's at 1080 ultra + 4x MSAA. It doesn't matter if one is 100% faster than the other because 2 vs. 4 fps is completely unplayable.
Highest Playable settings are what matters.

Listen dude, i don't even want to think about play monster game with amazing eye candy and using AA which blurring the image instead of improving.

If one card is faster with max settings, its faster in my book, Crysis 3 is game to be pushed to the limits...at least SMAA 4x they should use, not FXAA.

Point being, 780 ti is not really faster in Crysis 3 then 290X, at least not 10% faster. Most people want to max out Crysis games anyway. So ye, it can't be moot point, never.

Its like saying, GTX680 is almost as good as 780 if you don't use max settings, if you use which card is faster?

In 2 tests, from 3 290X DCII is beating the 780 ti when MSAA is applied, so again which card is faster in Crysis 3?

 

[Face2Face]

Moot? sure it is, because the Nvidia card didnt win.

For me moot is using useless FXAA on 700$ card.

It's useless if you are running 23fps @ 2560x1440... None of the current High end cards can play it well with MSAA @ 1440p..
3-4 Fps.. who cares.. Both cards equally suck with MSAA -- 22.9 to 23.0 -- OMG it's faster!! within a margin of error...

 

[Psyside]

It's not useless if you are running 23fps @ 2560x1440... None of the current High end cards can play it well with MSAA @ 1440p.. 3-4 Fps.. who cares..

So just because that settings makes the cards crowl, its useless? but hitting 60 fps with disgusting AA mode is great?

What about 2XMSAA? 4XSMAA? in those scenarios 290X is getting really close, and with MSAA it beats the 780 ti, so the "780 ti is faster in Crysis 3" is the moot point, not using MSAA.

This is clear win for 780 Ti,

FXAA scenario? not really.

 

[Face2Face]

So just because that settings makes the cards crowl, its useless? but hitting 60 fps with disgusting AA mode is great?

Exactly - What's the point if you are watching a slide show..? Not much looks disgusting @ 1440P
Sounds like Linus Tech Tip testing.. let's make it crawl to show you pretty much... nothing relevant..

 

[Silverforce11]

It would make sense to test ultra settings with MSAA or SMAA if it were multi card config where its playable. But I agree, slideshow fps aren't meaningful for single cards.