R9 290 *Complete* review list

[Jaydip]

This was discussed, beginning near the bottom of the third page of this thread.



They used a cooler from a 280X on a 290? Does that mean coolers built for the 280X, 7970, or 7950 should work on a 290/290X despite the 45 degree rotation of the die? If I could use my Accelero 7970 or the Twin Frozr off my 7950 on a 290, I'd be very tempted to order one now . . .

Ahh thanks :thumbsup:

 

[SolMiester]

On price/performance, no doubt.

- GTX780 reference beating performance for $100 less
- R9 290 OC is just a hair behind GTX780 OC for $100 less
- Putting the lights out on the rumored GTX780 Ti at $699 with 2x R9 290s
- Trouncing 770 2-4GB cards by 25-29% for $10-70 more
- Beating NV's Titan in BF4 with a $399 card before Mantle
- Destroying Titan SLI with R9 290X CF by 20-40% in multi-monitor gaming, which means R9 290 CF > Titan SLI under the same usage in both FPS and smoothness.

Ear plugs not required the minute after-market versions come out. HIS ICEQ X2 cooler from R9 280X dropped the temperatures just 63*C under gaming in games like BF4 and Crysis 3.

You also get 4GB of VRAM for no premium at all. It might not matter, but it's "free" anyway.

Aside from the Multi card performance on a newer GPU, all you are measuring is $ perf, not performance period, and AMD has always been the cheaper of the 2, so nothing new there.
4K in my opinion is moot as neither really delivers as a single card, and until the Ti & aftermarket cards are released, THE performance ceiling of 28 node, has yet to be seen.
Its a great chip, poorly delivered IMO, and unless you dont mind finishing the product with an acceptable cooling solution, i believe most will gratefully go with the polished NV alternative at the reduced price, until OEM versions of the 290 with sane cooling and noise levels round the product off.

 

[3DVagabond]

It's rather simple really. This card is not a $400 card. Unless your're deaf or your PC is in the other room, it can't be used to give you the advertised performance due to the noise. So you'll have to invest in a cooler which will bring the cost of the card closer to $500 than the MSRP (while voiding your warranty). Taken these into consideration an aftermarket overclocked $780 doesn't look that bad at $515.

This post is total FUD.

1, It is in fact a $400 card regardless of where you put your PC or hearing impairment. 2, The noise in no way stops it from performing "as advertised". You don't have to invest in a cooler, but if you do you can get a perfectly acceptable cooling solution for $50. It will then be cooler, quieter, and faster than the 780 @ $500 ($50 more). When the aftermarket 290's come out they'll likely still be $100 cheaper than the aftermarket 780's you are comparing the 290 reference card to.

 

[Saylick]

Jeebus, 2 cm away seems awfully close. Although this method would be easy to replicate across multiple cards, the result becomes rather meaningless as no one in the right frame of mind plays PC games with the graphics card 2 cm from their ear. The only useful bit of information that you get from a testing methodology like this is the relative acoustical performance between cards, but not the 'absolute' acoustical performance.

I would agree that a standard needs to be developed, however, but developing that standard would take a lot of coordination and agreement with respect to the measuring device, measuring distance, which case to use, which PC components to use, etc. You'd end up trying to hit a moving target.

Maybe what needs to happen is to start attempting to quantify the amount of sound damping enclosures provide using a standardized sound source, and measure this at various distances away from the enclosure. With each new enclosure review, there would be a section that measures the perceived noise from the standard sound source at various distances away. For video card reviews, you repeat the experiment except you leave the card in an open bench but still record the noise levels at the same distance intervals as the enclosure review. This way, you can get an idea of how much less noise you should be able to expect from a given enclosure + graphics card combination using the data from the enclosure and graphics card reviews. You could even compile the data into a spreadsheet and by permuting through each case + graphics card combination, you can get a gauge of whether or not you will get better or worse acoustics with a given upgrade.

 

[3DVagabond]

True, and in an unprecedented move from NVIDIA for this holiday season, they are bundling three free AAA games too. Assassin's Creed IV: Black Flag and Batman: Arkham Origins alone both sell for $60 each and are brand new games. So with the noise/fan speed/heat situation and lack of free game bundle on R9 290 cards, AMD vs. NVIDIA comparisons on perf. and price are very much apples to oranges right now.

Unprecedented? How could you possibly have missed all of the AMD Never Settle bundles?

 

[hyrule4927]

Unprecedented? How could you possibly have missed all of the AMD Never Settle bundles?

Unprecedented for Nvidia maybe? I hope that's what he meant anyway.

 

[SolMiester]

So the worst case scenario is the $400 card performs in line with what you'd expect vs. the $500 competition. If you don't mind more fan noise you can easily best the more expensive competition in performance.

And if you do the same with the $500 card you will get the $699 Ti stock performance....<shrug>

 

[3DVagabond]

Or it could be as AMD stated, and there may be an issue with the cooler. Too much thermal compound, or not enough, or a heatsink with a mounting surface that is not in spec.

AMD did change the spec on the 290 fan speeds, which will differ from cards that were already shipped. Those cards will require a firmware flash.

Actually I believe it's just the drivers that increase the fan speed.

 

[3DVagabond]

AMD did amazing, even if you're green through and through the massive price drops in response to the 290x does nothing but benefit you. Too much epeen sword fighting when everyone should just enjoy the better prices we've got now.

Amen! Everyone enjoy the lower pricing and free games.

 

[VulgarDisplay]

Have you guys seen the THG review? it was rather disturbing for sure.Looks like AMD wanted to shine in the review benches, I don't endorse that behavior either from AMD or NV.

C'mon. Stop repeating this it's already been proven wrong.

Tom's 1 reference card they bought that was OBVIOUSLY broken has now turned into all 290's are broken.

People sensationalize things to an extreme degree on the internet. Literally Tom's (probably the WORST reviewer in existence) bought one retail card (that AMD tells them is broken and needs to be RMA'd) and they publish this ridiculous article that now everyone has taken and run with. Now all 290's throttle super bad and the review scores are unattainable.

Except for the other reviewers who bought retail cards and are within a 1% margin of error with their review model....

 

[3DVagabond]

Never mind.

 

[Will Robinson]

Really,the people that have called the R9 series a success are NVDA.
They have slashed the prices of their high end cards to compete.
Competition at its finest.

 

[3DVagabond]

Theoreticals, so not even sustained DP, are just that. Hardly much of a selling point in CUDA dominated scientific community.

Furthermore, Hawaii can barely utilize even SP without burning 300W, let alone 1/2 DP,
and those professors, you can call them crazy, but they are proly not too keen on setting fire to their labs and will likely insist on sub 250W solution.

So in conclusion: the imaginary card will have to be clocked at 300MHz or something like that. So you can forget about that 50% DP over K6000.
Yes I am exaggerating, but not much

Actually, you are exaggerating a lot. Besides moving the goalposts and being completely off topic.

The statement was that GK110 was 100mm² larger because it had real estate dedicated to compute that Hawaii was lacking. What you quoted was the response, which showed that isn't true. What you said is completely irrelevant to the discussion. As well as being completely false. 300MHz to get under 250w? Come on now! I'd bet the the 725MHz clocks that Titan runs at with DP enabled would get Hawaii well under 250w.

 

[Jaydip]

C'mon. Stop repeating this it's already been proven wrong.

Tom's 1 reference card they bought that was OBVIOUSLY broken has now turned into all 290's are broken.

People sensationalize things to an extreme degree on the internet. Literally Tom's (probably the WORST reviewer in existence) bought one retail card (that AMD tells them is broken and needs to be RMA'd) and they publish this ridiculous article that now everyone has taken and run with. Now all 290's throttle super bad and the review scores are unattainable.

Except for the other reviewers who bought retail cards and are within a 1% margin of error with their review model....

Funny thing is it depends on the circumstances.I remember THG was quoted before multiple times to prove a point, so now when they are pointing out their observations that makes them incompetent? I get it man :thumbsup:

 

[3DVagabond]

The thing is, AMD does not sell their products cheaper than NV(Intel) because it can, but because it HAS to.

That's because AMD's customers don't drink the koolaid.

Don't put nVidia in the same category as Intel. That's beyond ridiculous.

 

[3DVagabond]

Actually when you equalize noise between R9 290 ref. and GTX 780 ref. (as Anandtech did in their review), the GTX 780 trounces the R9 290 in performance.

As for your fixation on Titan, that graphics card was beaten for a fraction of the price ages ago by GTX 780 SC's (without these alternatives being ridiculously loud or hot either).

Let's face it, until better after-market cards arrive, Anandtech cannot even recommend R9 290, and the only one to blame is AMD for designing such a loud and hot ref. card.

What kind of a qualification is this? I nearly spit my coffee out when I read it.

You know what? When you equalize price and VRAM between them you get a 290 vs. a 770 4gig and the 290 trounces it.

 

[blastingcap]

Does the R9 series come with a game bundle yet?

 

[3DVagabond]

wow, anandtech's review was brutal against amd.

sounds like they were more pissed off at the revised fan curve and took it out on the review.

More than likely just needed to prove that AMD CENTER and all of the money they are taking from AMD advertising hasn't compromised their ethics. It's raised some concern by some. S/A forums

 

[3DVagabond]

http://arstechnica.com/gadgets/2013/11/amd-stomps-nvidia-with-r9-290-at-least-in-reviews/

Another site recapping.

I agree anyone who went beyond kindergarten knows "one" isn't a sample size. But it should be investigated.

Judging by availability of the 290 chips which are salvaged from the X version, this is one leaky architecture.

We've had people on these boards state their 290X's don't throttle. We've seen OCUK take retail units from their inventory and they don't throttle. We've had Sweeclockers(?) obtain a retail sample and it showed no issues being within MOE manufacturing tolerances of their review sample. I'm sure that's enough of an investigation for most reasonable people "who went beyond kindergarten" and don't have an agenda.

 

[OCGuy]

We've had people on these boards state their 290X's don't throttle. We've seen OCUK take retail units from their inventory and they don't throttle. We've had Sweeclockers(?) obtain a retail sample and it showed no issues being within MOE manufacturing tolerances of their review sample. I'm sure that's enough of an investigation for most reasonable people "who went beyond kindergarten" and don't have an agenda.

Edit: This is a separate issue so I will stop discussing it here.

 

[BrightCandle]

I am glad it was investigated by others, because it needed to be. Tomshardware had 2 cards (not one, anyone that has read the review knows its two cards) that performed very poorly due to thermal throttling. Its good that AMD has confirmed it considers that a RMA scenario, makes it easier for people who receive them (hopefully rare) to get them replaced with the companies that sell them.

 

[Erenhardt]

Actually when you equalize noise between R9 290 ref. and GTX 780 ref. (as Anandtech did in their review), the GTX 780 trounces the R9 290 in performance.

Yea, we have a new metric. [Performance*smoothness/equalized_noise*Watt]*%of_industral_design

So in conclusion: the imaginary card will have to be clocked at 300MHz or something like that. So you can forget about that 50% DP over K6000.
Yes I am exaggerating, but not much

"Have to be clocked"... so it is given. Damn, ok my bad. Now give me source for that FACT

 

[GaiaHunter]

Yea, we have a new metric. [Performance*smoothness/equalized_noise*Watt]*%of_industral_design

It is a shame that metric wasn't in place during the GTX 480 debut.

One of these days I'll see a launch where AMD has superior performance without having new metrics introduced.

So far we had texture shimmering in old games (even if one needs to blur all the details to avoid the shimmering), the FCAT, now noise equalization...

 

[SlowSpyder]

And if you do the same with the $500 card you will get the $699 Ti stock performance....<shrug>

Right, so you spend $100 more (25%) over the 290 to get the roughtly same performance as the 780Ti via overclocking the 780. With the 290 you're not overclocking, just trying to keep the boost speeds AMD says the card are made to run at (I think AT said 927MHz in their review). But you can then overclock the 290's, too...

But all of this is a moot point, AMD provided a reference card and a GPU for it's partners. The aftermarket cards will be here soon enough, and they will bring reduced noise, increased clocks, and likely reduced power use.

And for everyone saying that you don't need 4GB, can you explain the benefit of 2GB or 3GB on a GPU of this level? I can tell you the benefits of having more memory, but I'd like to hear how having less is better.

 

[SiliconWars]

Yea, we have a new metric. [Performance*smoothness/equalized_noise*Watt]*%of_industral_design

Christ I laughed, hard. :thumbsup: