So what's the verdict on R9 290X ?

[2is]

It's amazing how important power consumption is when your brand of choice isn't the clear performance king any longer...

If you need to add another AC unit because of an extra 150 watts being dumped into the room, your original AC was on its way out to begin with anyway.

 

[Fastx]

Is it a great success?

Nice Card
(My opinion)

AMD did a nice job on the performance level (looking at all the BM's) of this card, the only thing it needs is AIB cooler or possible a Arctic Extreme III and should take care of the temps/noise to good/nice degree and less throttling which could/should just improve the performance even a little more than where it is now. If I would have went/ordered a 290x already I would not have been able to resist not ordering a Arctic Extreme III also asap to put on this card and try out imo. So your looking at around 600.00? (AIB cooler) or $$625.00 with a Extreme III which should work well imo. If noise and temps are not a problem for you then $549.00 .

I really thought the card would price around $599.00 -$649.00 so at 549.00 (from some vendors)it is covering the cost just about for a aftermarket cooler and/or cooled AIB card imo.

Edit update Face to Face posted some info on the Arctic Extreme III for the 290x so check this first incase anyone reads the above and thinks about a Extreme III for the 290x

http://forums.anandtech.com/showthread.php?t=2350153

 

[24601]

It's amazing how important power consumption is when your brand of choice isn't the clear performance king any longer...

If you need to add another AC unit because of an extra 150 watts being dumped into the room, your original AC was on its way out to begin with anyway.

I could care less about the power consumed.

Too bad for the laws of thermodynamics though, where I have to dissipate that wattage.

 

[Elfear]

are you suggesting for that 1 fps (sadly only one game benched). you gonna: give up tessellation, give up frame pacing, give up wattage, give up temp, give up noise so you can pocket $100 today only to pay it back in electricity.

Why would you have to give up any of that going with a 290X?

Tesselation - Is the 290X performing particularly poorly in tesselated games?

Frame Pacing - Working very well according to Guru3D and PCPer had good things to say about the new Crossfire experience too.

Wattage - You do give up a little efficiency but it's not a massive difference as some are making it out to be.

780 is 5% more efficient than the 290X in Uber mode and 19% more efficient in Quiet mode.

Temps - Honestly, who really cares about temp as long as the board can handle it (which AMD engineers claim it can)?

Noise - Well you have two choices. 1) Quiet mode is 2db higher than a Titan with it's great factory cooler and 4% faster than a 780, 2) Uber mode is loud at 50db (not jet engine loud but definitely noticeable) and 13% faster than a 780.

So you can still have a card that is faster than the 780 and hardly any noisier.

So yes, I'd say for a lot of people the $100 savings is great.

 

[chimaxi83]

the regular benches are moot regardless if warm up or not. cause there is time for the gpu to cool until the next bench is started.

the throttling bench show a different story. no time for the gpu to cool. constant load. and it shows 290x have a 1fps advantage over 780.

actual gaming will definitely show a different story. inside of a case and in moderate ambient and constant for hours on end.

-----

are you suggesting for that 1 fps (sadly only one game benched). you gonna: give up tessellation, give up frame pacing, give up wattage, give up temp, give up noise so you can pocket $100 today only to pay it back in electricity.

if so. more power to you.

Do you mind showing me how to calculate exactly how long it would take for a 290X to cost you $100 in electricity?

Actually, let me take care of that for you. Let's say this card is using 250W continuously, for 4 hours a day, every single day of the year.

0.25kW x 4 hours = 1kWh

1kWh x $0.16 (or your electricity cost) = $0.16 per day

$100 savings (only to pay it back in electricity, right?) / $0.16 = 625 days

So you'd have to have this card consuming a continuous 250W for four hours, every single day, for 625 days, or 1.7 years, to "pay it back in electricity". You can see the feebleness of that argument, can't you?

 

[UaVaj]

Elfear

instead of jumping to the end like you did. if you took the time to read this thread from the beginning. you would realize that all the support links are posed in post #24. you would also realize that for "real gaming" a 290x is not much better than a 780 in post #69.

as for beating a 780. if trading all that for 1 fps is your cup of tea. more power to you.

 

[UaVaj]

You can see the feebleness of that argument, can't you?

can you see the feebleness of ~1fps difference in real gaming?
http://www.nordichardware.se/Grafik...er-pa-grund-av-ineffektiv-kylare.html#content

 

[Enigmoid]

can you see the feebleness of ~1fps difference in real gaming?
http://www.nordichardware.se/Grafik...er-pa-grund-av-ineffektiv-kylare.html#content

I wonder how much fps the R290X loses in the benchmarks if it had to run for 15+ minutes.

And the card is slightly faster than the 780. Within margin of error (5%) with the titan. An overclocked 780 is going to perform similarily. AMD really borked this card, should have put a proper cooling solution on it.

Impressed with the Hawii chip, not impressed with the R290X reference.

Edit:

Lets look at their Sleeping dogs result.

42 fps when the card really only plays at 38. Now how is the normalized game results going to change when this is taken into account?

 

[24601]

BTW: Am I a prophet or what? Way better than Jebus or Muham madbro?

http://forums.anandtech.com/showthread.php?t=2344677&highlight=

AMD/ATi did exactly what I predicted but I made the mistake of giving AMD/ATi the benefit of the doubt that they would ship a cooler to match the real TDP of the chip.

People are lapping it up even without the cooling solution.

It's alla bout living life 30 seconds at a time . (Before the thermal failsafe kicks in).

 

[iiiankiii]

I really want to see the 290x with proper cooling. Right now, the average is around 1100 MHz with stock cooling and voltage. If it can do around 1250 with good cooling, this is hands down the best bang for the buck. That'll match the performance of the avg. OC of the titan and 780gtx.

Power consumption is a mute point at the high end as long as the performance justifies it.

 

[chimaxi83]

can you see the feebleness of ~1fps difference in real gaming?
http://www.nordichardware.se/Grafik...er-pa-grund-av-ineffektiv-kylare.html#content

In case you hadn't noticed, the purpose of my post was to deconstruct the idiotic notion that this card will, in real world terms, somehow cost more in electricity.

Looks like you don't have a response to that. Got it :awe:

 

[XiandreX]

Dude. I'm going to have to grill you for a second. I apologize in advance but I feel this is warranted:

No better no worse? FFS, Please read the reviews and translate them before you cite them as evidence. That nordic review shows the 290X 11% faster at 1080p and it is of course faster at higher resolutions. It also had 10% higher scaling after being overclocked, so that's 10% higher than Titan with that in consideration. You're focusing on the clockspeed over time graph. "DO NOTE". Like your phrase there. DO NOTE. That the performance is consistently higher over the lifetime of that graph, so your statement "no difference" is incorrect. The first two runs are faster, and then the graph stabilizes. They also manually used the low fan mode - uber mode fixes this. This does not excuse the pathetic cooler on the 290X. I've said a million times I think that cooler is terrible. It's a rather easy fix by setting it to uber or upping the fan speed. Again, that doesn't excuse the state of the cooler. I'm with you on that.

This was their conclusion from the review :



I get it. The cooler on the 290X sucks. I don't disagree. You know what bugs me though? Crap like this. Citing evidence which really isn't evidence, evidence that directly contradicts what you're saying. Nordichardware allows all of their hardware to "warm up" during their benchmarking tests and they do lonog benchmarking loops - despite this , the 290X is more or less even with the Titan. Not "even or wose" like you state. For Frick sake read the review that you cite as evidence, this is pretty pathetic.

Like I said, I agree fully that the cooler on the 290X sucks. I also think the 780 is a better "balanced" GPU ignoring the cost. Hell, if they were the same price, the 780OC is outright better because of the intangible benefits such as noise and efficiency. If you want to say that, i'm right there with you. But once you start spreading garbage like this, that's where it gets annoying. This clockspeed over time BS you're talking about, along with solmeister, doesn't really have a basis in reality. The performance is still relatively even with Titan even after long periods and after the warm up, which is what nordic hardware does in their review suite. The throttling over time graph? If you look at it, the performance is stable for the duration of the graph, with the first two runs being faster - this was also using a manual low fan speed. As said, uber mode should fix this, although it will be noisier because the cooler sucks. They do loops 10-20 times for each benchmarks, it isn't a 30 second type of thing - and still, the reviews show it even with Titan more or less. Come on man. Don't be ridiculous. If you want to trash the 290X, just come out and say it. If you want to trash AMD just come out and say "I PREFER NVIDIA AND DO NOT LIKE AMD blah blah blah blah". Heck, i'll trash the cooler with you. I think it sucks. But don't cite evidence that isn't evidence and don't spread misinformation.

Blackened I concur... like I have said in other threads and this one. We all agree that the reference cooler is crap. However the hardware is insane and with AIB coolers the heat/throttling issue is gone. Check the temps from overclockers... as much as 30 degree difference (celcuis).

 

[BrightCandle]

I don't think power consumption matters much, only its impact on noise. The card is still faster in its quiet mode than a 780 and cheaper. I am disappointed in the price however, its still quite an expensive graphics card compared to the historical pricing structure of these two companies but its better than Nvidia's ludicrous prices. The trade off with the current cards I think is clear, about 6 Db of extra noise over the 780 in quiet mode and about 11.5Db in Uber mode over the 780. But Uber mode competes with the Titan which makes it 9Db louder than its Nvidia's Titan. That is the trade off, you get 50-100% more perceptible noise for $100-450 less. You gain some exclusive features (Audio, mantle) and loose some others (PhysX, gsync).

However compared to a Titan you can get an entire watercooling kit and a 290X and have the card be silent and get to water cool your CPU and still have change in that $1000 to buy dinner for the wife. Considering the watercooling will likely have the card up in the 1200Mhz range with a voltage mod that is quite simply a fantastic deal and will perform quite a bit better than even an overclocked Titan.

The OEM's will almost certainly make higher quality air coolers that will solve the trade off for a little more money but we'll see that if and when they come out. My hope is we'll get Uber mode like performance but around the volume of a 780 and maybe $50 more on the card and that will really be a bargain.

There are still features on the 780 that might be worth that higher price, things like PhysX, FXAA and potentially Gsync in the future. The 290X however counters with a truly fantastic audio addon that in the future promises to bring back decent environment sound mapping like the Aureal Vortex had (that was awesome) and the mantle API which EA games will be using heavily but we still don't know the impact of.

At 4K resolutions the 290X seems to really excel compared to the competitors cards, it on average looses at 1080p compared to the 780 but as the resolution goes up the card looses less performance with each step and by the time it reaches 4k it is winning by 25%. It still isn't really achieving good frame rates at 4k so that isn't a usable result but at 2560x1440 its got a significant edge.

In crossfire the cards scale well. Frame variance is still higher than with NVidia's cards but for the most part its a not an issue and no one is really noticing that level of variance although their are studies showing people do still prefer Nvidia's SLI as its smoother. At 4k resolutions unfortunately some games show significant stutter as before and it appears the drivers still haven't been fixed there so its best to not get CF if you are considering 4k resolutions.

That I think is basically the roundup of the 290X. Its cheaper than the competition, has some cool extra addon features that promise more in the future. The current solution is a bit too loud as the cooler isn't all that great but we all expect that to be fixed with the OEM cards in a months time. Its a decent competitive card that is quite a bit cheaper than its competitor. That is my conclusion. I still think a 290X is a bad buy considering you can get 2x 7970's/2x680's for that money which are going to outperform it in pretty much all games at resolutions people really play at.

 

[XiandreX]

I don't think power consumption matters much, only its impact on noise. The card is still faster in its quiet mode than a 780 and cheaper. I am disappointed in the price however, its still quite an expensive graphics card compared to the historical pricing structure of these two companies but its better than Nvidia's ludicrous prices. The trade off with the current cards I think is clear, about 6 Db of extra noise over the 780 in quiet mode and about 11.5Db in Uber mode over the 780. But Uber mode competes with the Titan which makes it 9Db louder than its Nvidia's Titan. That is the trade off, you get 50-100% more perceptible noise for $100-450 less. You gain some exclusive features (Audio, mantle) and loose some others (PhysX, gsync).

However compared to a Titan you can get an entire watercooling kit and a 290X and have the card be silent and get to water cool your CPU and still have change in that $1000 to buy dinner for the wife. Considering the watercooling will likely have the card up in the 1200Mhz range with a voltage mod that is quite simply a fantastic deal and will perform quite a bit better than even an overclocked Titan.

The OEM's will almost certainly make higher quality air coolers that will solve the trade off for a little more money but we'll see that if and when they come out. My hope is we'll get Uber mode like performance but around the volume of a 780 and maybe $50 more on the card and that will really be a bargain.

There are still features on the 780 that might be worth that higher price, things like PhysX, FXAA and potentially Gsync in the future. The 290X however counters with a truly fantastic audio addon that in the future promises to bring back decent environment sound mapping like the Aureal Vortex had (that was awesome) and the mantle API which EA games will be using heavily but we still don't know the impact of.

At 4K resolutions the 290X seems to really excel compared to the competitors cards, it on average looses at 1080p compared to the 780 but as the resolution goes up the card looses less performance with each step and by the time it reaches 4k it is winning by 25%. It still isn't really achieving good frame rates at 4k so that isn't a usable result but at 2560x1440 its got a significant edge.

In crossfire the cards scale well. Frame variance is still higher than with NVidia's cards but for the most part its a not an issue and no one is really noticing that level of variance although their are studies showing people do still prefer Nvidia's SLI as its smoother. At 4k resolutions unfortunately some games show significant stutter as before and it appears the drivers still haven't been fixed there so its best to not get CF if you are considering 4k resolutions.

That I think is basically the roundup of the 290X. Its cheaper than the competition, has some cool extra addon features that promise more in the future. The current solution is a bit too loud as the cooler isn't all that great but we all expect that to be fixed with the OEM cards in a months time. Its a decent competitive card that is quite a bit cheaper than its competitor. That is my conclusion. I still think a 290X is a bad buy considering you can get 2x 7970's/2x680's for that money which are going to outperform it in pretty much all games at resolutions people really play at.

Well summed up BrightCandle. I couldn't have said it better myself. I honestly believe that once the AIB 290x units get released, depending on the premium will sway a lot of people in the direction of the 290x as the choice to go for.

 

[UaVaj]

In case you hadn't noticed, the purpose of my post was to deconstruct the idiotic notion that this card will, in real world terms, somehow cost more in electricity.

Looks like you don't have a response to that. Got it :awe:

your math is "excellent" with the wattage. to get the complete picture.

now can apply that same math to the 1fps difference, the tessellation difference, the frame pacing difference, the noise difference, the buy price difference.

thanks. will wait for the results. :whiste:

 

[wand3r3r]

Dude. I'm going to have to grill you for a second. I apologize in advance but I feel this is warranted:

No better no worse? FFS, Please read the reviews and translate them before you cite them as evidence. That nordic review shows the 290X 11% faster at 1080p and it is of course faster at higher resolutions. It also had 10% higher scaling after being overclocked, so that's 10% higher than Titan with that in consideration. You're focusing on the clockspeed over time graph. "DO NOTE". Like your phrase there. DO NOTE. That the performance is consistently higher over the lifetime of that graph, so your statement "no difference" is incorrect. The first two runs are faster, and then the graph stabilizes. They also manually used the low fan mode - uber mode fixes this. This does not excuse the pathetic cooler on the 290X. I've said a million times I think that cooler is terrible. It's a rather easy fix by setting it to uber or upping the fan speed. Again, that doesn't excuse the state of the cooler. I'm with you on that.

This was their conclusion from the review :



I get it. The cooler on the 290X sucks. I don't disagree. You know what bugs me though? Crap like this. Citing evidence which really isn't evidence, evidence that directly contradicts what you're saying. Nordichardware allows all of their hardware to "warm up" during their benchmarking tests and they do lonog benchmarking loops - despite this , the 290X is more or less even with the Titan. Not "even or wose" like you state. For Frick sake read the review that you cite as evidence, this is pretty pathetic.

Like I said, I agree fully that the cooler on the 290X sucks. I also think the 780 is a better "balanced" GPU ignoring the cost. Hell, if they were the same price, the 780OC is outright better because of the intangible benefits such as noise and efficiency. If you want to say that, i'm right there with you. But once you start spreading garbage like this, that's where it gets annoying. This clockspeed over time BS you're talking about, along with solmeister, doesn't really have a basis in reality. The performance is still relatively even with Titan even after long periods and after the warm up, which is what nordic hardware does in their review suite. The throttling over time graph? If you look at it, the performance is stable for the duration of the graph, with the first two runs being faster - this was also using a manual low fan speed. As said, uber mode should fix this, although it will be noisier because the cooler sucks. They do loops 10-20 times for each benchmarks, it isn't a 30 second type of thing - and still, the reviews show it even with Titan more or less. Come on man. Don't be ridiculous. If you want to trash the 290X, just come out and say it. If you want to trash AMD just come out and say "I PREFER NVIDIA AND DO NOT LIKE AMD blah blah blah blah". Heck, i'll trash the cooler with you. I think it sucks. But don't cite evidence that isn't evidence and don't spread misinformation.

:thumbsup:

your math is "excellent" with the wattage. to get the complete picture.

now can apply that same math to the 1fps difference, the tessellation difference, the frame pacing difference, the noise difference, the buy price difference.

thanks. will wait for the results. :whiste:

The reviews warmed up the cards, and found a double digit difference greater than the 780 (at least in uber mode and higher resolutions). You have 1 single graph showing a 1 FPS difference in a single game. Resorting to cherry picking simply shows fanboyism. Even in the single cherry picked game, it's faster than the $100 more alternative.

Every point you keep going on like a broken record. The frame pacing is shown to be pretty decent. The price is better than NV. The performance is better than the 780 and around titan (beating it too, barely).

Start showing facts, not just repeating the same slogan over and over (whatever team green is supposedly better in at the moment). The fact you are claiming differently than every review out there is a little disturbing to say the least. :whiste:

 

[OILFIELDTRASH]

The math on the wattage is flawed. The formula is assuming the 780 is not using wattage as well. The 290x is not consuming 250 watts more. More like 50 to 100w more. Factor it that way and you'll get the idea of what the extra consumption costs.

 

[UaVaj]

there is no point in continuing this discussion when you and your followers obviously choose to pick and choose whatever you want to read and whatever you want to believe and ignore the rest.

then have the nerve to call me fan boy ism.

already posted the all the supporting link on post #24. need to make them red and bold too or what?

-----

when something as simple as real gaming (constant 2-3 hours with bf3 or the like) inside a case with moderate ambient cannot even be comprehended.

you and your followers can all go back and praise your 2 minute benchmarks.

 

[UaVaj]

as if post #83 was not clear enough...

 

[blackened23]

If you actually READ your cited article refutes what you state and is not evidence at all. Their testing is prolonged, in fact, BF3 was ran for 3 hours. But did you read that crap? Of course not. You look at one chart, one in which - oddly enough - the performance over time was consistently higher, and THEIR results were consistently higher than Titan, you come to this grand conclusion of "worse than 780". Their results after prolonged testing for hours was, faster than Titan. But did you understand or read that crap? Hell no. To you, you mis-cite the information and link a chart that was used with a manual slow fan speed. As I said, as THEY said, the problem is fixed with a higher fan speed or uber mode.

Like I said, I think the 290X cooler is garbage AND I think the 780OC is a better balanced product. If you want to state this, I will agree with you - i'm right there with you on this. Yet, what you're doing I can only conclude that you're here intentionally skewing the article with your broken English. You simply want to trash AMD while skewing information along the way. That's the only conclusion I can make since you can't even bother to read your so-called evidence, and you link a chart in which you didn't even read the accompanying article to. And their testing was prolonged for hours. But can you be bothered to post anything but FUD? Misinformation and FUD is annoying for both camps. And i'm afriad that's exactly what you're doing right now -it's annoying, especially with the broken grammar. I've called multiple people out on misinformation, both red AND green guys, precisely because it's annoying.

 

[wand3r3r]

How about discussing facts backed with proof? So far I haven't seen any that demonstrate the fallacies you mention (UaVaj).

Short benchmark runs will show impressive numbers while the card is cool, though. Once you start gaming for extended periods of time, the card will get progressively slower as it heats up, and you'll be hit with a 30% performance penalty in the long run. We then made sure the card was at constantly realistic long-term-use temperatures for our benchmarks.

http://www.techpowerup.com/reviews/AMD/R9_290X/33.html

Despite this downclocking symptom you are attempting to describe, the 290x still scores like this:

7-13% faster in uber bios at 1080p/1600p, while hot, and $100 cheaper which helps alleviate the subpar cooler.
http://www.techpowerup.com/reviews/AMD/R9_290X/27.html

 

[Enigmoid]

The reviews warmed up the cards, and found a double digit difference greater than the 780 (at least in uber mode and higher resolutions). You have 1 single graph showing a 1 FPS difference in a single game. Resorting to cherry picking simply shows fanboyism. Even in the single cherry picked game, it's faster than the $100 more alternative.

You realize that sleeping dogs was the only game they bothered to test in that manner and that had they tested other games they would have been used as well.

And that according to his post those results are inaccurate because they do not stress the card long enough for thermal throttling to occur. Even that review shows 42 fps for sleeping dogs when the card will actually only give you 38-39 after 15 minutes.

And really 1 fps is well within the 1) margin of error 2) can't visibly be distinguished.

 

[PingviN]

Cooler is objectively terrible compared with what Nvidia has, but that's gonna change when AIB manufacturers get their hands on the GPUs. Other than that, I have no complaints. It's faster and cheaper than the competition and has a lot of nifty features coming with it.

 

[Imouto]

Oh, looks like ShintaiDK said that stuff about temps and performance enough times to start a new trend. So I'll post this here too:

hardware.fr testing methodology

About that Swedish web there's only 2 possible scenarios.

- They only tested 1 game properly and debunked the rest of the review.
- They did all the testing properly and their results are legit.

They're either incompetent or the 290X pulls ahead. I don't really know why you keep spinning it.

 

[wand3r3r]

How about discussing facts backed with proof? So far I haven't seen any that demonstrate the fallacies you mention (UaVaj).

http://www.techpowerup.com/reviews/AMD/R9_290X/33.html

Despite this downclocking symptom you are attempting to describe, the 290x still scores like this:

7-13% faster in uber bios at 1080p/1600p, while hot, and $100 cheaper which helps alleviate the subpar cooler.
http://www.techpowerup.com/reviews/AMD/R9_290X/27.html

You realize that sleeping dogs was the only game they bothered to test in that manner and that had they tested other games they would have been used as well.

And that according to his post those results are inaccurate because they do not stress the card long enough for thermal throttling to occur. Even that review shows 42 fps for sleeping dogs when the card will actually only give you 38-39 after 15 minutes.

And really 1 fps is well within the 1) margin of error 2) can't visibly be distinguished.

Are you talking about TPU's statement (or the single benchmark that UaVaj included)?

If so, go read their review. I quoted their words above.

Short benchmark runs will show impressive numbers while the card is cool, though. Once you start gaming for extended periods of time, the card will get progressively slower as it heats up, and you'll be hit with a 30% performance penalty in the long run. We then made sure the card was at constantly realistic long-term-use temperatures for our benchmarks.