• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD Radeon R9 290X 4 GB

its a bit too hot and loud for my taste but all I care about is it driving the prices down.:)
 
290X in quiet mode spends most of its gaming time in 650-850 MHz region, and it still beats 780 with a far margin in most games.

I dont see it?

1% and 4% over a stock reference 780. Heres the kicker most 780s on the market beat the Titan as is. On top of that a reference 780 is slower than the aftermarket cooled custom PCB 780s at the same clock speeds. So quiet mode the 290X = 780 margin of error in benchmarks is 3% as such they are tied and with a slight edge in uber mode overall to the 290X

perfrel_1920.gif
perfrel_2560.gif


Lets see where a 780 custom card comes in out of box
perfrel_1920.gif
perfrel_2560.gif


At 1920 the custom 780 is on par with the 290x
At 2560 the custom 780 is 2-3% slower than the 290x which falls into the margin of error.

I fail to see on a broad set of games how the 290X is that amazing. With clock scaling being what it is if the card pegged 1000 mhz during gaming. It would be on average another 5-7% faster giving it a greater lead and one that can be felt not just seen in benchmarks on the first run.

As it stands knowing performance drops over time slightly means in reality Quiet mode = 780 and Uber mode = Titan its not as much faster as its on par and everyone acts like beating titan is some insane feat? Every GTX 780 reviewed on TPU other than reference card beats Titan.

The reference 290x is not impressive. Aftermarket cooled cards will show what the 290x is actually capable of right now the Hawaii silicon looks fantastic. Cooler / Bios / and reference settings not so much.

So looking at the current data a GTX 780 is $100 more expensive.

290X + waterblock = cost of a GTX 780
290X + aftermarket air cooler like an Accellero = $70-80 = $20-30 cheaper than a GTX 780 for same performance without the noise / heat or clock throttle
 
Yeah your situation being described is like comparing a Nascar race to a highway joyride.

There are different levels of overclocking, I run 3DMark that is extrieme, that is the Nascar race. Tuned settings I do not use in day to day.

I play games, I overclock for that, the profiles only kick in when in 3d mode. But it is a day to day overclock(a small overclock, limited to no voltage increase). I want it quiet more than loud. You can look at my specs, I am not arguing to argue, those are the specs I game with. The 290 is damn good, I think it IS a 480. I am excited to see this tech optimized in the future.

I just don't see why you think you can dismiss other people's use case in a thread because it is not your use case.
 
Last edited:
where is this "far margin" you are talking about? are we even looking at the same graph?
at 4 times the amount of noise and 25% more power draw, the 290x is about 1% faster than 780gtx in silent mode at 1920x1080, and 4% more in 2560x1600

http://tpucdn.com/reviews/AMD/R9_290X/images/perfrel_1920.gif

Reviews have also shown that at higher resolutions (like eyefinity) the R9 290X does pretty well and scales to the higher resolutions well. At 1080p, there is a good bet that you don't need a video card this beefy. You're getting this if you're going beyond that, be it with 1440p, 4k, or with 3 displays.
 
I just don't see why you think you can dismiss other people's use case in a thread because it is not your use case

I haven't dismissed anyone's case. I wasn't the one who came in to the thread telling people to "grow up". People can have their case, and not see eye to eye, but that doesn't mean you come saying I have dismissed everyone in this thread...
 
This is AMD's version of Fermi/GF100.

It seems like history keeps repeating itself. Sure, you get a card that is cheaper than its competitors for the same performance but you pay for it in power consumption and noise.

It's the same as the HD 2900 XT and GTX 480 (and more recently the 7970GE, to a lesser extent).

I'll take a slightly higher performing and hotter Fermi equivalent at $550 over a throttling Kepler at $999 anyday. The key here is that the R290x is limited by the temperatures and not the VRMs like the Titan -- and it raped the Titan in its own Nvidia-favoured games at higher res. Ask anybody here what they'd rather have -- a GPU that's cooler/temperature limited or a PCB power limited card and anyone sensible will pick the former every time -- much easier to buy an aftermarket version of the card or a new cooler than voltmod or BIOS flash and risk getting a $1k paperweight. Most people here have 2 PCs or more and a tablet/phone as well so I doubt they will give a shit about a few extra watts here and there (though that Blu-ray/multi-monitor consumption looks quite embarrassing -- curious if it can be fixed with some modest underclocking/custom profiles).

One thing is for sure though. If that 780 Ti is anything less than a fully un-gimped GHz Edition GK110 with all 2880 cores and at the same price, Nvidia can GTFO till next year when Maxwell is ready. No point in them wasting time releasing anything less than that (that rumoured 2490 core GK110 will fail epically against this card, regardless of clocks). Titans are going to collect more dust on retailers' shelves until they do some serious price cuts on it (50%-60%).

This card did a hell of a lot better than I thought it would (I expected it to slot beneath the Titan, not above it). Big props to AMD for bringing the prices back close to sanity levels. And I'm glad this card just took a huge shit on Nvidia's "boutique card" parade to be honest.

I personally will wait for the BF4 benchmarks and aftermarket versions before considering biting the bullet on this card. The beta performance was...inconclusive, to say the least, and I'm curious to know if one 290X will max BF4 or not -- and if not, how far it is from doing so.
 
Last edited:
I haven't dismissed anyone's case. I wasn't the one who came in to the thread telling people to "grow up". People can have their case, and not see eye to eye, but that doesn't mean you come saying I have dismissed everyone in this thread...

I read:
When would you need Uber mode other than extreme situations?
As a way of saying/implying/dismissing anyone who would need Uber mode most of the time, we can drop it.

*Bro hug
 
Hey W1z what if instead of overvolting, you try to undervold? If it's still stable and you manage to lower the temps as well, wouldn't that mean longer time at the max boost state and better performance?
 
i'm looking at doing a PCIe scaling article with 290X in the next weeks.
I'm looking forward to it. It (PCIE3 vs PCIE2) didn't appear to make too much of a difference with Nvidia's GTX Titan. However, I thought that it may have been a different situation with AMD's R9 290X.
 
The key here is that the R290x is limited by the temperatures and not the VRMs like the Titan -- and it raped the Titan in its own Nvidia-favoured games at higher res.

Yeah, my VRM's dont help me game at 1136 core (143 MHz above max stock boost of 993) or bench at 1202Mhz (209MHz above stock max boost). Still......

And your definition of 'rape' is a tad off the mark. Single digit %'s does not cry 'rape'. Though in uber mode you could carry our many nefarious acts and not be heard.

However, I do believe that under proper cooling the card beats Titan in all fair fights. It might use a lot more juice but that's not really a point for enthusiasts.
 
Price... nailed it weeks ago! Performance right what I figured. :toast:

Efficiency not totally sold, but would like to see it over more games and that average of Perf/Watt. As it throttles back we could find the such results noteworthy. I honestly thought AMD would find the efficiency within a rearrangement of the GCN to provide 18% performance, while not raising power usage as was with Bonaire. That said, it's not that bad 20% faster with 5% more power (Uber mode 27% using 13%).

Noise, bad but considering the watt's/heat trying to be shed from a die 30% smaller than a GK110 it’s not unpredicted. Could AMD advanced the blower cooler to do better perhaps? Are we hitting the limit of heat transfer while maintaining a dual slot cooler that blows all the heat out the back... conceivably? Could AMD gone more exotic, imaginably, but at what cost? Would love to know the actual fan differences of the Titan and the AMD uses in in terms of noise, performance and price. I suppose AMD has to displace more heat so if the competing coolers and the fans are basically equal, the R9 290X has to push more air so 2dbA isn't bad. As a release/reference cooler it's not great, however that wouldn't be a big issue, but it becomes one if the AIB's aren't able to move their designs in just a few weeks.

AMD did this, although I don't think they're plan was ever to do it, and the whole reason for the "No New GPUs from AMD for the Bulk of 2013" in early February. I think they thought they could hold-out for 20Nm on risk production Q3 2013, but that went bye-bye with TSMC and they had to counter the GK110 with something. They realized the 28Nm for a high performance design, while staying significantly economical (they don't have the luxury of geldings from the professional line) and maintain power was no "walk in the park" on 28Nm. This is what we get and why neither party really hasn't done exceptionally better than rebranding 28Nm production.

I think AMD had make do with design compromises. Except one place price and on that we have a success.

What will Nvidia do... drop a 780 to $580; and bring the 780Ti in at $650, but other than initial review samples all those 780's will be AIB dual fan style cooler's. Titan will maintain at $1000, but in most terms it's already vanished. Those who need it for DP compute have pretty much snapped one up, those for gaming it it has no importance to pay the extra.

What's next "Non-X" versions for $450 that are 3-4% below a 780, while 10% besting the 770. That might be a good contest by the Christmas season. :D
 
Everyone is freaking out this thing is 2dBa higher noise than the Titan under load... If you enable Uber Mode, which has a slight increase in performance, and which would obviously generates more heat, the card jumps 10dBa... People's expectations are both unrealistic and on the verge of Nvidia fanboy. If you got a problem with noise or cooling, that's exactly why they make aftermarket coolers.

To expect the card to give the same performance with at least similar power consumption, heat output, and noise level as the competition isn't being unrealistic. Especially not when the competition has been on the market for 9 months.

The card has potential for sure, but AMD really should have worked on the stock cooler more. There is no getting around that. The price is the only saving grace, and since rumor has it the 780Ti will just be Titan with 3GB of RAM and likely price the same or slightly lower than 290X and with a stock cooler that doesn't deafen people, the price advantage will be short lived.

the difference between 40decible and 50 decibel is TEN (10) TIMES LOUDER

50dBA is twice as loud as 40dBA, but still that is quite a bit louder.
 
Very much AMD equivalent of "GTX 480" launch. Performance is more than ok but stock cooler is awful and power consumption is all over the place. Price would be nice if it had a proper cooler but as it is, it just makes GTX 780 DCII OC look even better buy.

Btw "A change in power by a factor of 10 is a 10 dB change in level." -wikipedia

50dBA is indeed 10x 40dBA power.
 
the difference between 40decible and 50 decibel is TEN (10) TIMES LOUDER

10dB is about twice as loud to the human ear. Power, and perception of the human ear are different things.
 
The codename fits this card well, just like Fermi fit the GTX 480 well, though both could be used to describe either card, Volcanic-Fermi.
 
It seems like a great card, really nice when considering the price, so hopefully it will bring the prices down. I wonder what affect applying some Arctic Silver/MX paste might have on the temps? I know it's not ideal, nor should it be acceptable that someone has to do that reach acceptable temps/db, but in my experience, reapplying paste can take about 10% off the gpu temperature, which could be enough for the card to reach 1000mhz in quiet mode?

Personally, I prefer AIB cards, simply because they have more finesse (temps/acoustics). But I've often wondered how far a bit of paste might go in terms of making a reference model a much more acceptable purchase.
 
w1zzard; said:
powervstemp.gif
there is the data, linear fit is y = 1.1975x + 215.29

basically for every °C that the card runs hotter it needs 1.2W more power to handle the exact same load

Got that from a thread in xs. I'm pretty sure if watercooled, then power consumption will be much much lower.
 
I'm impressed!! (Price/performance ratio just rocks!!) Just built my entire rig 7 weeks ago. I may have to jump ship, what to do?
 
Last edited:
I'm extremely disappointed with this card. I had high hopes for the 290/290X and was all ready to switch to the red team. The price and performance is awesome, but the power consumption, noise, and thermals are a deal breaker.

Sure, the extra power consumption will have a minimal effect on my power bill, but the cost of having to run the AC to keep my apartment at a reasonable temperature sure will. My GTX 670 already heats up the room during long gaming sessions. Maybe I could overlook this if the card was properly cooled and the fan noise didn't have to be so loud my neighbors would complain to keep it from throttling. Hopefully the 290 has a little bit better power consumption and launches with aftermarket coolers. If so, it'll be a day one purchase for me.
 
The card has potential for sure, but AMD really should have worked on the stock cooler more. There is no getting around that. The price is the only saving grace, and since rumor has it the 780Ti will just be Titan with 3GB of RAM and likely price the same or slightly lower than 290X and with a stock cooler that doesn't deafen people, the price advantage will be short lived.

I doubt the Ti will have a slightly lower price than the 290x, but time will tell.
 
Now we go LIVE! to get a reaction of Nvidia TITAN owners after showing them the R9 290X priced at $549.99


LOL!

But seriously, these jokes are getting kinda old, moving on...


Now now Mr Xzibit, less of the trolling please ;)

You could also rightfully use the same video with the quote,



In all seriousness, it's a brilliant piece of engineering and the price for it is absolutely fantastic. If you already have a loop, then this is a dead cert. But to build a loop for it alone, well, that's adding a couple hundred bucks on top for that heat dissipation.

But I assure you Mr Xzibit, as a Titan owner I am not crying, I am smiling because AMD have the means to make Nvidia rethink their strategy and that for all of us is a pretty good thing.
Also, it was mostly enthusiasts that bought Titans, not green eyed zealots. With that in mind, Titan owners might well end up grabbing these too. We have the cash, we buy wtf we want :D

Yup, you have to recognize this is an amazing piece of hardware, yes, AMD had to save in the cooler, but I don't care, I know I'll put my 290X under water, and the best thing of all, those savings were passed down to the costumers.

What good is a Magnesium/Aluminum alloy cooler if is going to be in a box for the rest of the card's life span? Don't get me wrong, Titan looks like a piece of modern art compared to the 290X (how does the saying go? That's a face only a mother would love :p) but I paid an arm a leg for something I used for a few weeks only.

Haters are gonna say this card is loud, hot and inefficient, who cares? the performance is there, and for a reasonable price too! This goes on and on, Prescott, Bulldozer, GTX480, Titan, and now the 290X... Whatever, people will always find something to complain about.

All I know is this is an awesome card, and credit goes to AMD for pricing it right, and that's alright in my book :)
 
Last edited by a moderator:
Btw "A change in power by a factor of 10 is a 10 dB change in level." -wikipedia

50dBA is indeed 10x 40dBA power.

However, it isn't 10x louder to the human ear.
 
Back
Top