Discussion in 'Reviews' started by W1zzard, Dec 23, 2013.
To read this review go to: http://www.techpowerup.com/reviews/ASUS/R9_290X_Direct_Cu_II_OC/
Nice review =)
Please review the Tri-x 290x and non x to =)
I know that in the page where you show the cards PCB close ups you say that the rebranded CHiL gives voltage control but have you checked firsthand if you can manually set the voltage to a desired value?
Seems like this card is for the hardcore performance/quiet setups. It's interesting that in most instances that this card is lower in power consumption versus the reference designs, which is nice. I wish that there were VRM temps recorded in the temperatures section, especially with these cards.
Nice review overall.
Excellent card and review like always.
Too bad this one doesn't overclock.
I always thought the prices listed in the comparison charts on the first page of the review were MSRP, seems I was wrong...
Yea, when did AMD raise the MSRP for a R9 290X... by $30 hadn't read that?
When you directly compare this Asus to the Asus GTX 780Ti W1zzard did a while back this isn't bad. First they appear to have the same cooler and fans, so other than the difference in what heat load they displace and the fan profiles Asus chooses for either, would be what differentiates thermal performance and noise... correct?
Power consumption at peak (actual gaming loads) the Asus 290X show it’s draws 4-5% less power, didn’t expect that! Noise the Asus GTX 780Ti at load was 39dbA, and the 290X had either 32dbA (quiet mode) or 42dbA (performance). Temperature the Asus GTX 780Ti was 80°C, the 290X (quiet mode) was 94°C while 78°C (performance). Now what I need to know is what (program/game) is used to stress and achieve those results, hopefully both the noise/temperature number are from the same stress test.
So, from that we see less power, while higher heat produced from the AMD Hawaii, while even faster fan (or at least by the noise that would be the case). Transistor density of Hawaii might point to one issue, while the abundance of that heat (some say inefficiency) uses a die area 30% smaller to dissipate it with. Just look at the two pictures W1zzarrd has of die outline of both the Asus heat-pipe coolers to realize the disparity. So is there problems with Hawaii perhaps, or did the push to much on the envelope of the 28Nm process packing it a little to much? By what little OC headroom I think a little of both is apparent.
Then look at performance, by the 2650x summary there’s something like 12% between the two Asus models, that’s noteworthy. However can one contend the Asus GTX 780Ti could logically be priced somewhere like $70 more (12%) than a Asus 290X @$600. They're basically similar BOM’s other than the chips, so what gives with $730? While then look at the perf/watts at 2650x this Asus 290X bests a reference 780Ti, although the Asus version by all account accounts would pull ahead perhaps by 4-7% using the gaming titles used currently. Surprising given what was unearth above, I thought it would’ve been much more lopsided.
So from all this we grasp while AMD went with a more densely packed part that provides a better cost, Nvidia has a large die that is less densely packed but at a higher cost per chip. Although it seems there's another 9% tack-on above the logical perf/$.
AMD obviously hit some issue’s with such a design, chiefly being able to dissipate the amount of heat from the chip surface, but after that most of the parameters aren’t all that different as perf/watt noise or temperatures. I’d like to see the 290X temperature if Asus had set the fans to provide the same 80°C or 39dbA as they did with their 780Ti.
The Asus r290x really isn't that good , I have the sapphire tri-x 290 and I don't pass the 65c while gaming , I had the chance to stress the card with furmark and it reached 75c with no audible noise at all for me , my friend has the tri-x 290x his card reached 78c with furmark .
Note : you also get an HDMI cable .
It is infantile, unintelligent and sadly short sighted comments like this that feed forums with poor discussion. If you can't be constructive with your criticism, why not just keep your fingers off the keyboard?
It's not really relevant what you are saying. It's a broad spectrum of games tested and I've said it before - more than almost any other review site. There are plenty of gaming evolved titles in that list, not just TWIMTBP. As for Tress FX being off, well, so is PhysX in Metro 2033. Physics can be done by non Nvidia GPU in Metro 2033 it's just that it's coded for NV. Tress FX is written off the back of GCN.
Also, BF4 is a massively AMD sponsored game. It favours AMD far more and you can see it in comparison differences between BF3 and BF4. In summary, the review list and conditions are fine. What matters are resolutions tested and a good mix of games. There are sites that massively cherry pick titles to suit their sponsors but that doesn't happen here.
All this being said, it does seem as if Nvidia just held back long enough to see what AMD had and the release of the 780Ti just stole the thunder. The MSI Gaming 780Ti is head and shoulders above this card and in the UK is 'only' £50 more (give or take). Now, what happens with Mantle on board will be nice to see but again, that depends on adoption, not raw hardware.
I still dearly want to see the 290X Lightning and hope to hell that MSI get it right.
I really don't understand why a video card fan doesn't just turn off in idle. I mean if 32db can keep it stable then I'm thinking it could run passive in idle.
quote="Slomo4shO, post: 3052817, member: 135417"]I always thought the prices listed in the comparison charts on the first page of the review were MSRP, seems I was wrong...[/quote]
they are the same street prices we use for the perf/$ charts, and yes they are retarded. to complain: firstname.lastname@example.org
completely off might be bad for some secondary components because they heat up more. running it extremely quiet is always possible. some fan motor designers require a minimum rpm for the fan to spin reliably though (design fail imo)
Great review as always w1zzard , can we expect tri-x reviews in the near future ?
yes, talked to sapphire today, they are getting the card back from the previous reviewer early next week.
Awesome - any chance we can have all 290x custom cards in the other custom cooler review comparison charts? (or, at least, the noise chart). Would make it much easier to see whose custom cooler is best.
Also any news on an MSI 290x gaming edition review?
I voted with my wallet . ShopBLT until prices drop and then I will go with TigerDirect over Newegg. Prices at Newegg haven't been competitive for months and TigerDirect price matches Amazon and I can circumvent the local sales taxes since Amazon now collects in Arizona
On a side note, can you also include MSRP in that initial table in addition to the Newegg pricing? The current table makes it look as if AMD raised the MSRP on the R9 290 by $100 and the R9 290X by $30 if one were to compare this review to your previous reference reviews.
First of all, thank you for your thorough review W1zz, it's greatly appreciated.
I have a question I hope you may kindly answer for me, I noticed the RAM doesn't have any sort of cooling other than the air blown by the fans, do you think there's enough room under the heatsink for some low profile copper RAMsinks (6mm)?
I also noticed you posted some new BIOS images, are these required for all cards, or only for press samples?
I noticed that two of the heat pipes don't make any contact with the GPU, greatly reducing the efficiency of the cooler, but even taking that into consideration the temps were great all around, and zero throttling to boot.
I ordered this card last night, a few hours before your review went online, I was lucky enough to find it for a little less than the MSRP ($593.00 with free shipping and no tax) by avoiding the scalpers out there (I'm looking at you Newegg...)
Once again, thank you for your review, I hope you can help me with the RAMsink question
Great review W1zz I enjoy reading these.
Is Grid 2 and Metro Last Night supposed to be on the same page?
from what I understand, all production cards will have the same BIOS or neweg. I still posted mine for full disclosure.
It would also be good to see the TPU treatment on the MSI Gaming edition, I got mine (290x) on Monday, they all seem to have Hynix so far - although my reference card (Asus) with Elpida OCs it's RAM better (5952MHz), probably needs the BIOS changed, OCUK staff apparently had the RAM on one up to 6800MHz using an Asus reference BIOS so maybe doing a bit of OC by mixing/matching BIOSes would be a fun test to add.
Regarding the other factors, it's pretty quiet and keeps the chip around 78c in games and is very quiet at idle. Very nice.
LtMatt: There's 5 heatpipes and only 3 make any contact with the core. To be honest it looks like only two of them make decent contact. That's pretty poor tbh. I think they just used the Nvidia DC cooler version for the AMD card and its clearly not an ideal fit.
I agree, it's not ideal. Heat will migrate sideways though and into the pipes on the sides.
There's a massive difference between even a small amount of airflow over the components, and none at all.
Just registered to try to discuss several points (sorry for any misspelled words as English is not my mother tongue) for which, W1zzard, your knowledge would be very helpfull (thanks in advance):
- firstly, there is one fact that has been surprising me for a long time since R9 290/290X was reviewed for the first time: 95ºC for a GPU / any part based on Si (Silicon)-technology. I am a student specialized in electronics finishing his dissertation of a five year career in Spain which is called 'telecommunications' which is a BSc+MSc in computer science + electronics + other stuff. I have studied the basics of Si-technology, properties and how more or less, make a transistor from a silicon wafer. As the temperature arises certain temperature, to form or to cut the channel (1 or 0 between the Drain and the Source of a MOSFET -remember that the tech. used in general in microchips is CMOS) in the transistor you will have to arise the voltage in more noticeable steps. There are also another studies (papers, PhD Thesis, etc) for aging related to frequency, operating temperatures and switching rate (% of times that a transistor switch between 1 to 0 or vice versa), without considering electromigration effects at the PCB nearer the chip (to prevent electromigration the general rule of thumb is to stay below 65ºC) and the quality of soldering of the BGA socket due to cold-hot cycles. So, in general, the less temperature the better for the circuit properties and features, but also in terms of aging of the circuit. Hence, has AMD traded-off lifespan of a top-tier product for pure performance? Wish I could have a not prone to failure 290 or 290X to test it around the two year guarantee to give you an answer, but I tend to think that the answer is yes. What do you think?
- secondly, most of the chips that we actually use have its components not powered at the same time. This fact is related to how to power a lot of different transistors and power density. Looking through the power consumption charts and the temperatures in über (German word by the way which means greater or higher than) and quiet, the power consumption is the same for more or less the same performance in FPS. Power consumption in general is proportional with frequency and square with voltage, and more temperature more voltage to create the channel. I really do wonder what materials and properties are used at 28nm TSMC and how AMD manage to have the same power consumption footprint; because the only answer to this is that AMD is powering certain zones and pipelines of the chip depending on the power footprint and temperature (like an automatic watchdog over power and performance)... althougth it was stated that the 290 / 290X was throttling back due to temp. issues (hence to control the power used by the chip to not enter into a positive feedback: +temp => + power => +power dissipated => +temp). What are your thoughts about it?
PS: dissipated power is related to the switching rate due to the linear portions of the MOSFETs, then to the activity of the chip commuting 0s and 1s. If you need some more links about this or something that I've told over the post, please, don't hesitate to ask for more info as I understand that maybe a too detailed technical stuff in one post could be difficult to comprehend.
Id hate to have 94'c of heat dumped inside my case... I dont think no amount of airflow will be able to keep the inside of your case fairly cool let alone the nightmare of it when it comes to crossfire unless you live in the north pole or have the AC chill your room to 8-10'c (if thats even possible) - watercooling them is a must.
for single cards though, id rather have a reference cooler so the heat gets dumped outside of the case.
Nice looking card but dam! Why cheap out on the Elpida memory chips..
Note to self: strike this one off the maybe to buy list
Separate names with a comma.