The Cause Of And Fix For Radeon R9 290X And 290 Inconsistency

Page 12 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


It seems I was off a little and I do apologize for this inconsistency. There is no denying that SLI currently performs better than Crossfire. Unfortunately, no matter which one of the two you choose, going with multiple GPUs is going to result is unnecessary headache at the lower spectrum. For example, pairing two 650 Ti Boost would get you slightly better performance than a GTX 680 if the game happens to be compatible with SLI but even in this case, you would have been better off going with a GTX 670 and overclocking it to match the performance of the more expensive 680. The 10-15% performance drop between the SLI setup and the GTX 670 would be well worth it since performance is going to be consistent. Right now 2gb GTX 650 Ti Boosts are around $150 whereas a GTX 770 is currently $330. The $30 upgrade is worth it. Alternatively, you could go for the R9 280X for $300 or pickup a clearance HD 7970 for as low as $230-250.

I have tried GTX 660s in SLI and replaced them for7950s in quadfire (see current build in sig). Both setups have had their issues. I have been keeping an eye on the GTX 780 and R9 290s and will be replacing this quadfire with a dual card setup depending on which cards go on sale this holiday season. I have always based my decisions on price/performance and tend to challenge the rampant confirmation bias that plagues tech sites. At the end of the day, cheerleading for either company provides no value for the consumer. Competition is what drives progress and this duopoly with the fanatic polarization only hurts the consumer.
 






yeah I see now, my experience tells me always go with the best single gpu possible, crossfire is bullshit as is sli, buy the best single card in your budget, when it is completely absolete, toss it and grab yourself a new one
 


Pal, do you read my mind? lol they are all true for me specially this one '' I have always based my decisions on price/performance and tend to challenge the rampant confirmation bias that plagues tech sites'' who are you then an alter ego?! 😉
 


It is usually not practical to pickup the top tier card. The performance difference is usually within 5-15% of the 2nd in line but the price premium is usually around 25-50%. For example, the R9 290X is currently 37.5% more than the R9 290 but is only about 6% faster and the GTX 780 Ti is currently 40% more than the GTX 780 but is about 16% faster. The Titan is within 6% of the 780 but used to cost 54% more and is now 100% more. However, bare in mind that the Titan can also be used as a professional card for CUDA development.

It does make sense to pair the 780 or 290s in lieu of a single 780 Ti.

perfrel.gif
 


that was exactly why I went with r9-290 not r9-290x, to be honest I didn't have the budget for it, but even if i did, i would again go with r9-290...
 

That's funny coming from a dyed in the wool AMD fan! :lol:
 
lol this is ridiculous, crying about a difference of 100 rpm's. let me clue "author" there are manufacturing inconsistencies not just from the same plant but from manufacture to manufacture. I have owned dozens of cards from both Nvidia and AMD and have seen failures on both. Did you check the thermal grease/ thermal pad, on my Asus DCU 7970 half the grease completely missed the GPU core. Next you put the rig in an 80 degree environment, turn the fans on the quiet setting and then complain after hours of torture testing that the gpu restricts it's self, news flash, every video card will do that. In my rigs I always install many 5+ quiet 120mm case fans so that the case temps is =2-3 degrees warmer than environment and you'll hardly EVER even hear the gpu. To me these cards look amazing, 40% less $ than the Titan and much more powerful. then again I'm not trying to build a silent pc in a hot environment and write a subjective article.
 


What makes you think that reviewers are allowed to disassemble a card that has been lent to them by AMD?
 


Because reviewers do it fairly often. For example: http://www.tomshardware.com/reviews/radeon-hd-7970-benchmark-tahiti-gcn,3104-3.html
 


Without knowing the full ins and outs of the contract between AMD and THG's (or any other websites for that matter) there is no way of knowing if those pictures were taken by THG's or not and whether the same rules even applied to the 290X.
 
I think from now on, based on the results of this article, I would like to see Tom's posting ambient temperatures and fan speeds. That they haven't been up front about this, up until now, may be an oversight, but after this article, I don't trust their reviewers. The article clearly is based on the presumption that there is something wrong with the aftermarket cards, or AMD sent an unusually good sample to be reviewed, and calls into question the entire line-up altogether. None of the reviewers presumptions have turned out to be true.

The testing methodology smells very bad. Why would you hinder a cooler's ability, by putting it on it's most conservative setting, while at the same time testing it in an environment with so much inconsistency, and then not report on the environment's inconsistency, if you suspect these things may be responsible for an inconsistency in device performance?

If Tom's can give as anal a list of their test systems each time, I think ambient temperatures and running conditions of the device being tested is hardly a stretch.
 

I wonder what would happen to Tom's credibility if they pulled this stunt with an Intel CPU running under a stock cooler? Would Intel show signs of inconsistency in a similar fashion? I wonder how NVIDIA would respond (their GPUs and their PR wing!)
 

Were the retail boards, the Sapphire and ASUS, also on loan from AMD, or even their respective companies, or did somebody go to the store and purchase them? If Tom's wanted to feature a disassembled board to show the quality of it's construction, they would have no problems doing so, even if they had to outsource the pictures.
 

Article states that Chris purchased these just to investigate whether the retail cards performed identically to the reference card.

Chris thinks that many users won't tolerate the noise of high-speed mode. We can question that assertion, but he does provide the data to back up that assertion.

The problem arises when you test a card that includes two fan modes, come to the conclusion "it's incredibly noisy, but the lower speed fan mode is a reasonable compromise", and then find out that what you said about reasonable compromise isn't true when you test a retail card.
 
Thanks for the feedback.

I would agree with Chris's thoughts on users' tolerance of fan noise.

I would also agree with the reasonable compromise statement.

For the future, I really would like to see things done a bit more inclusively, with fan information and ambient temperatures included in the test system page (and it would even be useful for benchmarks to be run with a 100% fan speed where possible, to give readers a full idea of the card's potential,) where they are very easy to locate, so we the readers can draw more of our own conclusions, and the reviewers such as Chris can unburden themselves from accusations of biased testing.

I understand that reviewers can't take into consideration every minute detail, but having those pieces of information, including a 100% fan speed test, more readily in the readers face, would have changed the feel of the entire article and likely not drawn the negative attention to both Tom's and AMD.

Personally, I don't feel either outfit should have ended up with the negativity, but due to the construction of the article and testing, there it is.
 


It seems the R9 290 and 290x are necessitating that sort of information in future benchmarks. Previous to this gen I really don't think those details were particularly important - as long as the testing environments were stable and relatively 'normal'. These GPU's seem oddly over-sensitive to external factors; more so than previous generations from Radeon and current ones from nVidia. I think the reviewer's were more-or-less caught off guard by the results.
 
Hey guys i just wanted to share, i have the HIS 290x and i just ran GPU-Z and my card stayed between 1000hz - 900hz the whole time running battlefield 4. Now this is after the new beta 9.2 drivers (idk about before). But I thought i should share that my results are great and i really like the card! Note: i am using an Antec 900 case, so my card has plenty of air flow.
 


Well drag what about ambient temperatures? *poke poke*. Does the 55% fan speed still apply in climates that cultivate a naturally higher ambient temperature? Considering a higher ambient temperature will demand a higher RPM from the fan or a lower clock speed from the GPU; ultimately reducing performance. What happens if the ambient temperature is at such a point that the fan cannot compensate without turning into a leaf-blower? The GPU will be forced to reduce its capability.

I'm extremely curious to see a very specific form of testing be done:

A computer case housing an R9 290x in an open desktop environment vs. the exact same case in a desktop unit that envelopes the PC case, in the same room/environment - like one computer resting on the desktop and the other inside a compartment of a drawer-like enclosure. It's purely speculation, but a PC in an additional enclosure vs that exact same PC in a totally open environment should render - with respects to this experiment - significantly different temperatures ( ~3-5 Degrees Celsius). And with other GPU's this difference shouldn't really matter at all. But from the data that's been provided, the R9 290x might actually show a substantial decrease in performance due to an increase in ambient temperature created by an additional enclosure or a hotter environment. The mitigation to this is increasing the fan speed, which makes the GPU increasingly loud regardless of who is manufacturing the heat-sink/cooler.

 
I don't get it, if you do not get it or that's me... are you serious about your questions?! throttling? seriously? well the problem is solved in accordance with even Tomshardware, you don't need to worry about any ambient temperature, you don't need to worry about the core temperature, there was a problem with their fan speeds which is puzzled out, tell me if you have another point and I am missing that, because I am really confused, Albeit I didn't face the problem with my Sapphire R9-290 either from day 1, the temperature in my chassis is about 20c, about loudness of the fan, it is 55% it is only 7% more than it was set by default which was 40%...gtx 690 is about 55.3 dB(A) nvidia gtx 680 SLI produces 60.5 dB(A) AND R9-290 produces 57 dB(A) under full load, so it is like you have sli ed two gtx 680 at worst, I don't think it's a big deal, besides how many of us get this card to play at 1080p!? I did, so the full load scenario is sorta of void for this resolution, as well as the temperature problems, it can be the reason why I have not faced any problems mentioned here, well you might also like to take a look at this http://www.tomshardware.com/reviews/radeon-hd-4850,1957... it is an archaic article, well how many of you have had 9800gtx? did you feel anything in your cases really? well r9-290 is equal to one 9800gtx under full load...
 
I was on a research tour to a lot of websites that they are doing hardware reviews.
What did I discover?
In chronological order:

1. Tom's Hardware is all over the news about R9 290X and 290 inconsistency (and forums discussions) for the problem they revealed.
2. Nerd rage of AMD fanboys against Tom's Hardware was there too.
3. Then AMD admitted the problem, and fixed it via driver update.
4. Healthy AMD fans stopped the rage cause they saw after the update that Tom's Hardware revealed the problem and AMD responded by fixing it. Some of them are glad that Tom's Hardware found the problem early and made AMD to fix it.
But the rage continued from fanboys that were so blind by their fanboism to see even AMD's update to the drivers.

That's why we might expect to see a few of them to create new accounts in order to rage against this review.

Also for people that think that Tom's Hardware accused AMD for cherry picking, they didn't. Other sites said that and mostly people in comments.

edit: @Saga Lout:
Is there any chance that you are referring to me?
Cause I just placed the facts that happened all over the web (anyone can google them and see them) while I didn't target or attack any poster. I really dont like the word "fanboy" either but we have to describe someone who is obsessed about something while blindly avoiding the truth. And either we like it or not its all around us. We can see it everywhere in politics, in sports etc etc...
pe: I am a fanboy of Arsenal FC. I can't see things clearly when it comes to premier league. I will always support my team and blame the others even when I know that is wrong.
 

Your first question I don't understand, but if you're asking about the ambient temperature, I have no idea, as it wasn't listed, nor do I recall it listed whether or not the card was in an enclosure.

Here's the article, feel free to read for yourself:

http://www.extremetech.com/gaming/170542-amds-radeon-r9-290-has-a-problem-but-nvidias-smear-attack-is-heavy-handed

The author actually participates in the comments later as well, which is somewhat refreshing.

I suspect your second question should have had an exclamation point, since I would think we both know the answer to it, making it rather rhetorical.

Your last question seems rather pointless. Any chip with thermal throttling, run above it's thermal threshold, will throttle.
 
I'm giving serious consideration to removing any post containing the word "fanboy" or any spelling derivative on grounds of insulting behaviour.

Attack the post if you have valid comments but not the poster.

Before writing anything, try thinking of Tom's as a sort of friendly bar room then think of the consequences of facing someone with an insult and how they might react. It's far easier to do it here but there may be consequences.
 

A quote from this Tom's article, "So that left me wondering: were the press boards just lower-leakage parts able to sustain higher clocks?"

Tom's didn't have to accuse AMD of anything, they simply had to insinuate it and let others do the accusing.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom