AMD Radeon R9 290 Review: Fast And $400, But Is It Consistent?

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
after reading all this back and forth.... i think you all should just be men about it and waterblock and voltmod your gpu's.... all this arguing means nothing if your not willing to push your piece of silicon to its limits. 7970's and 680's at 1.5ghz+ had been pushing performance noticably above stock titan levels at least a year before the titan came out. lets see the benchies play out and see the actual 1.5-1.8ghz of this hawaii chip before we say its faster than a voltmoded waterblocked titan@1.7ghz... and then let the 780ti get its chance.... those are the only benchmarks that matter... nobody cares about 1ghz benchies.
 

mirakul2

Honorable
Oct 8, 2013
5
0
10,510
Tom's hardware owns AMD a public apology for accusing about "golden sample". Worst article ever.

5046
 
Three things strike me about recent 290 / 290x reviews:

1. I find the discarding of heat and noise issues disturbing ..... do not want to see this kinda thing become "acceptable". BTW, It wasn't mentioned in the article but this card almost unseats the H100 for "loudest thing ya can put in a PC".

2. I find the difference in benchmark scores from the various web sites disturbing.

https://www.youtube.com/watch?v=djvZaHHU4I8

3. Boys, the talk about this release crushing nVidia or nVidia crushing AMD is silly talk. It should be about a healthy competitive environment, only if they both succeed does the customer not suffer. And they have more to worry about than each other. Neither makes the most popular GFX card hitting Steam Servers ..... here's the Top Ten.

http://store.steampowered.com/hwsurvey/videocard/

DIRECTX 11 GPUS

1. Intel HD Graphics 4000
2. NVIDIA GeForce GTX 560 Ti
3. NVIDIA GeForce GTX 550 Ti
4. NVIDIA GeForce GTX 660
5. NVIDIA GeForce GTX 670
6. NVIDIA GeForce GTX 560
7. NVIDIA GeForce GTX 450
8. NVIDIA GeForce GTX 460
9. NVIDIA GeForce GTX 650
10. NVIDIA GeForce GTX 660 Ti
 
There may be factors at play, maybe driver related fixes, maybe the 290X needs to be rebenched in identical circumstances or maybe AMD coaxed Nvidia's plans with a Vanilla 290X and will release a 290X GE card after the 780ti is out. Needless to say the 290 at $400 is looking an absolute steal.
 
A very familiar story with that of GTX 670 vs GTX 680 - $100 premium for a halo card. GTX 670's were the best value one could get back then.

The only difference, and the major concern for a user experience centric person is the noise. Noise, the bane of AMD cards. People can say "I have a case 1m away" or "use noise reducing solutions". Remember; even in a case, the noise propagates - especially in a gaming case. Resorting to noise reducing solutions to cover AMD's poor cooler is also not the answer either.

The enthusiasts need to apply pressure to companies to continue innovating and improving; default cooling solutions are not exempt from this either. This, I feel, is the primary reason for Chris to focus so much on user experience.
 

somebodyspecial

Honorable
Sep 20, 2012
1,459
0
11,310


Point me to the posts in that thread showing it different than toms. I see people running 70% (far above the regular 47% fan), 100% etc, but I don't see it running cool and fast without doing something that isn't like the current retail (like adding water, this doesn't count). Toms is showing retail with retail fans causing a slowdown on 2 cards. I don't see people heating them up either and then testing to show any different than toms at retail fan speeds with a retail card's fan, though one guy I saw commented it takes forever for it to cool down after running a test. I don't have time to read the whole thread, but since you're pointing there you must have some posts you know about in there somewhere you can link to directly right? You have to do a little better than go here and read 3000 posts. :) At least give a page a few of these GREAT posts that show toms is wrong or something. And I read over 450, so c'mon spit out some post #'s as examples that defy toms results.

You could be completely wasting my time (like the first 450 posts where everyone just discusses ordering, warranty worthless etc). Tell me where to look that shows a retail card running its fan at 47% and matching Toms PRESS review cards. That is what I expect from retail cards, not 20% slower or AMD cherry picked chips for review cards. That rules out all water-cooling posts so should save you some time ;) It also rules out posts that they changed the fan on, as again that's not showing different than a retail, as you're CHANGING retail in that case and NOT like toms tested retails. Since you're one of the first posts in the thread, surely you can point these people out :) You're post #7 and all over in there ;) Also as some said in that thread, it likely voids the warranty the second you take the thing apart to replace the retail fan/heatsink so this already is a BAD IDEA for most and I double that thought at 95 degress...LOL. Maybe AMD was counting on this. No RMA's if everyone is VOID right ? :) VaporX confirmed it (he's with Sapphire right?)? Like I said, I went through the first 450 for nothing before jumping around wasting more time :) But a few had some with relevant info like this:
"2. Stock Cooler Really has to go, Takes a very long time to cool down after testing. Like every review said its no joke this card runs hot but its good in the winter =P" post 297

https://www.youtube.com/watch?v=QQhqOKKAq7o
Thread does give some interesting NOISE vids though ;) Courtesy Nordichardware site.
$ilent got 63 deg but with fan at 100% and doing who knows what for how long, but gpu clock shows 1150 (but this is max, not where it runs over time). No time to read more of the garbage in that thread and $ilent was having all kinds of driver problems while being the only one posting any real data in the first 450+ (a guy told him to re-install his OS...LOL). I think this sums his card up responding to someone saying it can't be $ilent at 100% fan:
"Honestly, its so loud is untrue and when you put your hand at the back of the card exhaust it actually feels like a hair dryer."
I guess I should have went backwards maybe, assuming the posts are better at the end, but one of the first posts says thanks for cleaning up the thread...I was suckered ;) You could delete the first 450 posts and miss NOTHING but "my card is coming in 3 days" etc.

How is gsync NOT a proven tech? 3 game devs and every review site says it is a sight to behold and is a GAME CHANGER with no re-coding necessary. So far we have ONE monitor at $100 more. Where did you get $150? The card has been mentioned at $100 difference and we have no other models yet. Theoretically this would be cheaper once INSIDE and replacing the chip already in the monitor. As it stands now they'll release a kit for you to do just that. So having it in already without whatever you're replacing should be less than NV's $100 addon.

What piece of evidence do you have that says Gsync doesn't work? Id (carmack), Epic (reigns/sweeney) and EA (andersson from Dice) all say it is awesome. They showed up on stage to declare this. You can watch a bunch of vids from them explaining the tech on youtube. These people, not NV, I mean them personally taking time to tell you why it works and what it does and will allow devs to do. Do devs usually show up to tell you about a product failure? To claim I'm ignorant you have to have some sort of proof someone says it sucks. Find me a developer or review site that said, "I saw that gsync in action, it sucks and is dead, so no point in buying NV for that, it's a failed tech." Heck find anything close to a statement like that. I'll let you off with anything even resembling it was "unimpressive". :) Heck I'll go further and say find any of them that says ANYTHING negative besides "AMD doesn't have it"...LOL. I don't count anything saying it's proprietary either. Because we don't know what they will do with it and can license it to anyone they want (though I wouldn't give it to AMD, it's not a detractor to me as a buyer). We have nothing on that topic and AMD may be in talks right now for licensing (doubtful, but just saying none of us knows).

You are calling me ignorant based on guesswork that goes against all current info. I'm saying I'm not ignorant based on all top devs (EA/Id/Epic so far) and top websites saying it was awesome. We have nothing negative said so far as far as functionality and pricing for stutter free/tear free/lag free is worth more than $100 on something I'll live with for 6-7yrs (my current monitors are both 5 & 6 yrs old). I'll pay you more than $13 a year or so for those 3 features that affect me all day when gaming. I expect both of my monitors to live until next xmas when I plan to buy one of these probably. Most of us (98+%) are on 1920x1200 or less, so if we're going to be upgrading soon anyways, it's really only requiring something I already have on my list of upgrades.

Also I don't even need a card if I have anything 650TI or above. I'm not in that group (have radeon 5850) but that's already a lot of people out from that cost. I'll wait for 20nm maxwell but there are many who are already 1/2 way there or wanting 1440p/1600p right now but haven't bit yet due to 1440p cost ($560+ at newegg if you don't buy from foreign or no name sites/brands on newegg, check newegg ONLY box). Most of us are in a waiting mode for vid cards that can turn on EVERY detail and still stay above 30fps min in all games, and monitors to come down a bit at 1440p. Hence 98% below this res. A ton of people have 1440p+ on their menu of upgrades (heck a lot still don't have 1080p according to steampowered surveys). People like me have BOTH already on our upgrade lists so plans haven't changed, just got gsync included.
 

somebodyspecial

Honorable
Sep 20, 2012
1,459
0
11,310


He's not the only one concentrating on the noise, many sites did it. Anandtech spent some time covering it also and even went so far as to say it was unacceptable to them and not recommended. Every site says something about the noise. And you're right, doing things like asking users to put on a water block to solve it is not the answer (voiding their warranty as the sapphire rep said in slomo4sho's link).

I really think we might have avoided a lot of this talk with $20 more spent on cooling it properly, but you have to judge it as they ship it on debut as there isn't anything else to judge it with (hopes and dreams they fix it with aftermarket stuff? Can't base a review on that). Although in Toms case, they had 2 retails to judge, so those results have to be taken into consideration also. These cards and the noise cause some real red flags to pop up. I've never seen retails run 20% slower than review cards. Very odd. They are usually FASTER.
 

somebodyspecial

Honorable
Sep 20, 2012
1,459
0
11,310


So then NV just releases a Watercooled 780TI in response right? Is this what we want to be the norm now? The chip is great, but it seems it needs a die shrink to really get going (which I hope they do early next year as a refresh, then hit with new tech on 20nm in Nov). Until we see some great retail cards, this thing is going to make a lot of people unhappy with the noise. AMD should have put better cooling on it from day one. How many will be in forums asking why it's so loud? Water is not the answer. Either make a design that can run fine with a fan or go back to the drawing board. This is a publicity nightmare and that's never good right before black friday/xmas.
 

DrKlahn

Honorable
Nov 5, 2013
24
0
10,510


From the forum post over there it sounds like the owner hasn't really researched fan speed as it relates to clockspeed:

"John C (owner of the throttling card)... Only by manually raising the stock fan speed to 50% I can get a "smooth" non-throttling clock speeds. This is a screenshot with fan speed manually raised in driver's settings. Once again, image is clickable."

AMD sets the fan to 55% max in uber mode to stabilize clocks. So I don't know why it would be expected that setting it below 50% would be sufficient.
 

Hitokiri Battousai

Honorable
Jul 17, 2013
31
0
10,530
At linustechtips, they do overclocked gpu benchmarks and puts 290 ( a great overclocker as described by linustechtip's team ) above R9 290X ( not so great overclocker here ) and 1-2 fps lower than an overclocked 780 competition while maintaining better price-point before AMD's flagship R9 290X and GTX 780.
 

Oscarcharliezulu

Distinguished
Sep 30, 2010
29
0
18,530
So, they actually fuse off 4 CU's to artificially create a lower performance card to sell cheaper than their non-neutered card. I hope this is a result of yields ie they do get 1 or 2 that fail, because otherwise why don't they just sell one high end card cheaper?
 


GloFlo chips work this way a lot more than Intel Fabs, yield differs a lot for the same class chip.

It makes perfect sense. Cost effective way of chip making.
 
A couple of thoughts-reference cards are always louder and run hotter. So this launch doesn't seem much different to me. The reference cards are for early adopters and enthusiasts who will water cool their cards. Most non-reference cards will not be able to use a full cover water block and the universal solutions leave a bit to be desired.

Inevitably the reference cards dry up the quickest for the two reasons mentioned above. To me, this level of performance at $400 makes this a phenomenal card.

Look, people have been tweaking their fan profiles and whatnot for a long time to adjust for personal preference. Again for me this launch doesn't seem that unusual or unattractive. If I had $800, you can bet I'd buy two right now and put em under water and watch them blow away Titan. For reference cards, they do what they're supposed to do; satisfying early adopters and enthusiasts.
For everyone else, the partner boards will fly off the shelves this holiday season. Great for consumers, great for AMD.
 

DrKlahn

Honorable
Nov 5, 2013
24
0
10,510


The Tom's article doesn't have enough information to really to say what is happening. What we can gather from the limited information is that it is going over 40% fan speed and staying at its lowest clock. It may be that the heatsink thermal grease is not properly applied (it does happen). It may be that there is an actual defect in the silicon. It may be a software bug.

If you have been following the launch on the various forums, by and large people are seeing the same performance on retail cards the reviewers have been. So in my opinion this should have been researched more thoroughly before just throwing out numbers in the graphs and hinting this is typical. I doesn't take much of an investigation on enthusiast forums to see the results the majority of people are encountering. The responsible thing to have done would have been to made this into it's own article detailing the behavior, fan setting experimentation and possibly a teardown to see if there was any negligence in construction.

 

tinmann

Distinguished
Apr 28, 2009
1,121
0
19,660
I've heard the 290x at different speeds and at even 75% fans speed the noise level is quite unacceptable and the thermals are also. I would rather have the GTX 780 Ti which will be the fastest card there is out of the box. And don't bother telling me about how Nvidia is shafting my wallet, it's my wallet. Worry about your own.
 


You = :ouimaitre:
While nvidia : :kikou:


Sorry just HAD to, although I see your point.
 

lewist

Distinguished
May 27, 2009
5
0
18,510
Posted by watercooled - Wed 06 Nov 2013 16:36
Quote from Hexus:

kingpotnoodle
Good science is always done with many repetitions of many samples ;-)

 

somebodyspecial

Honorable
Sep 20, 2012
1,459
0
11,310


Well maybe here's some more evidence. He owns a Titan also.
http://techreport.com/forums/viewtopic.php?f=3&t=90345
Another guy getting 795 and needing to boost fan speed to 50% to get it to level out. RETAIL. With the side removed he still hit 798. Replacing the grease didn't help either. Clearly the fan/heatsink needed another $10-20, and you wouldn't be able to say this stuff I don't think. Even then though, I have a real problem with needing to run 95 while the other guy runs 81. I don't see that as a victory even without the issues here.

http://techreport.com/news/25609/updated-retail-radeon-r9-290x-cards-may-be-slower-than-press-samples
So they're not "cherry picking", but variance has landed some people with non cherry picked retail cards...ummm...OK...
http://www.bit-tech.net/news/hardware/2013/11/06/amd-denies-cherry-picking/1?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+bit-tech%2Fnews+%28bit-tech.net+News+Feed%29
"Although it denies any wrongdoing, AMD has admitted that performance variance is an issue with the R9 series. 'We’ve identified areas where [performance] variability can be minimised,' the spokesperson added, 'and are working on a driver update which will minimize this variance.' No release date for the update has been provided."

So we didn't cherry pick but it's only retail that get the hit. Ok...And it will only be minimized not killed completely? OUCH. Sad that AMD is having to admit to ANYTHING here. But they did.

Are we to believe you're incapable of making a proper driver, and it only affects retail cards? Or you DID cherry pick, and your cards are clocked too high as retail is showing now in 3 cases (at toms and Techreport forum). No press cards exhibited this, so it's really hard to believe it's just bad luck at retail right? Why didn't any press review card get affected by this "variance"? There are a LOT of 290/290x reviews out there. I'm wondering how many more we'll see as people are running retail in CLOSED cases (though that wasn't JohnC's problem as he removed the side and it didn't help anyway).

Did AMD clock these so hot, that in a CASE you end up getting clocked down over time like toms shows or raising the fan all day to stop it? I hope more people post results with stock stuff before strapping on new fans or water etc. Most people will just run out of the box stock.
 

somebodyspecial

Honorable
Sep 20, 2012
1,459
0
11,310


Probably a few weeks after 780TI hits tomorrow. But I'd guess we should see them before Black Friday, and I suggest you wait for these to get a thorough testing before buying. Now that we have AMD admitting variance problems, clearly as JohnC suggests, we need testing inside a closed CASE and run for prolonged periods to see if you drop clocks. What happens if you run loops at load for an hour or two as a gamer would be doing after hours in BF3 etc? Surely someone will test for an hour or two now (shortly) to find out. Not talking furmark loops here (I find running it dumb), but gaming loops like a player playing all day.

How many cards are experiencing this "VARIANCE" stuff? We have no idea if we are getting the truth currently either. I have issues with this "variance" not affecting a single review card, but only retail. They may get totally vindicated soon, but they are clearly admitting a problem. At the very least you wait that out.
 
Status
Not open for further replies.

TRENDING THREADS