Because they suck for anything that's not a blower. Which happens to be almost everything an enthusiast purchases nowadays.
Because they suck for anything that's not a blower. Which happens to be almost everything an enthusiast purchases nowadays.
SolMiester
Diamond Member
- Dec 19, 2004
- 5,330
- 17
- 76
Those tests that say so made sure to use blowers and is even noted in the Anandtech review that that is the case.
The same problem arises in the FT02 and FT03 Silverstone cases.
The reason is that there are no exhaust fans around the graphics cards, therefore the graphics cards ARE the exhaust fans.
If the graphics cards don't exhaust most of their heat outside of the case primarily by themselves, that heat is basically trapped there slowly getting removed from the case by the case.
This is also why short benches favor large cases even if the case has 0 ventilation or fans. As long as the case starts with ambient case temperature = ambient outside of the case temperature, the thermal capacity of the larger volume of air in the case is useful for short benches.
As the most popular cards for enthusiasts are actually open air designs, the use of rotated case designs was deemed more than pointless by the case designers as seen by the Carbide Air and the FT04.
Last edited:
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
Your right that the GPU will dissipate less heat because usually lower temperatures = less leakage. In this case we won't know for sure just how effective better cooling will bring. With the GTX480, a better cooler lowered the power consumption somewhat by 15~30W but this would instantly be neglected when the clocks/volts were touched.
Anyway, when your dissipating close to 250~300W of power more efficiently (hence the lower GPU temp) your effectively transferring the heat into your case faster. This is the point I am making, and know from firsthand experience from using toasty video cards like the GTX470/480 vs an efficent card like the GTX680. Its also a reason why I sometimes like using stock blowers (with my FT02) because they don't affect nearby components/case ambient at all compared to a non-blower designs.
Last edited:
blastingcap
Diamond Member
- Sep 16, 2010
- 6,654
- 5
- 76
You guys seem to have gone off topic and turned this into a general efficiency thread.
The OP was talking about how there was good scaling for overclocking and AIBs would love it for that reason. I explained why they would not, because the oc headroom is actually pretty "meh" and one guy already posted how your performance does not scale up linearly with GPU clock and memory clock... you boost those by 10% and you might get what, 5-6% more performance? For 200W extra?
I did note that AIBs would probably love 290X for an entirely different reason: it's underpriced relative to the competition, so AIBs and retailers can gouge consumers for the difference. (Remember how 5850 launched for about $269 but retailers were charging over $330 at one point?) AIBs can offer custom 290X cards for way over $549 and still find buyers because the GTX 780 and TITAN look so bad in comparison. I would not be surprised to find MSI Lightning 290Xs with retail prices north of $649 for instance, even though it doesn't cost MSI $100 more in parts for the Lightning vs the reference board (which they don't even make themselves anyway). But this price arbitrage game has nothing to do with 290X's mediocre overclocking headroom.
The OP was talking about how there was good scaling for overclocking and AIBs would love it for that reason. I explained why they would not, because the oc headroom is actually pretty "meh" and one guy already posted how your performance does not scale up linearly with GPU clock and memory clock... you boost those by 10% and you might get what, 5-6% more performance? For 200W extra?
I did note that AIBs would probably love 290X for an entirely different reason: it's underpriced relative to the competition, so AIBs and retailers can gouge consumers for the difference. (Remember how 5850 launched for about $269 but retailers were charging over $330 at one point?) AIBs can offer custom 290X cards for way over $549 and still find buyers because the GTX 780 and TITAN look so bad in comparison. I would not be surprised to find MSI Lightning 290Xs with retail prices north of $649 for instance, even though it doesn't cost MSI $100 more in parts for the Lightning vs the reference board (which they don't even make themselves anyway). But this price arbitrage game has nothing to do with 290X's mediocre overclocking headroom.
Last edited:
Really?
It scales that poorly with clocks and yet yields such a dramatic power increase at the same time?
It scales that poorly with clocks and yet yields such a dramatic power increase at the same time?
The benefit of being late: being able to clock your cards with respect to yields to match or exceed the competition. The downside of such strategy: leaving little headroom on the table and redlining the chip. And you thought GK104 was being pushed to extreme limits!
blastingcap
Diamond Member
- Sep 16, 2010
- 6,654
- 5
- 76
http://www.techpowerup.com/reviews/AMD/R9_290X/31.html says that overclocking to 1125MHz GPU clock and 1575MHz memclock (13/26% oc, respectively) on Uber BIOS resulted in a gain of 7.2%. 7.2% is less than 13%. In fact, 7.2 divided by 13 = 5.4.
In other words: raising 290X by 10% GPU clock and 10%+ memclock will result in only 5.4% increase in framerate for that game/setting/resolution combination.
It's just one game and setting and resolution combination, but if it holds true on average....
This is with STOCK VOLTAGE, mind you, so you are still getting 7.2% more performance without raising wattage that much.
In other words: raising 290X by 10% GPU clock and 10%+ memclock will result in only 5.4% increase in framerate for that game/setting/resolution combination.
It's just one game and setting and resolution combination, but if it holds true on average....
This is with STOCK VOLTAGE, mind you, so you are still getting 7.2% more performance without raising wattage that much.
Last edited:
Cookie Monster
Diamond Member
- May 7, 2005
- 5,161
- 32
- 86
But the question is.. is the card actually at those clock speeds even in "uber" mode.
blastingcap
Diamond Member
- Sep 16, 2010
- 6,654
- 5
- 76
I haven't been following video cards much lately... I guess you are saying maybe it was throttled down or something? I don't know.
I also wouldn't draw too many conclusions from just ONE reviewer's card. It is possible he got a dud. So we'll wait and see, but the early results are not too promising and nothing to be excited about contrary to OP. I mean just look at how Quiet and Uber mode performance is almost IDENTICAL to stock GTX 780 and stock GTX TITAN performance. It seems likely AMD was targeting those two cards and asked their engineers to do whatever it took to reach those speeds, temperatures and power draw be damned. They got there, all right, but with diminished overclocking headroom and 95C temperatures and higher wattage than one would expect if you extrapolated from Bonaire (the first GCN 1.1 GPU that was more efficient than GCN 1.0, which is how 7790 managed to be the deliver the highest perf/watt among all 7xxx series cards).
To be fair, though 290X has 4GB of GDDR5 whereas most 7790s had 1GB of VRAM. So it's probably not fair to extrapolate from Bonaire like that.
Last edited:
No it doesn't seem to hold true on average. As you say, it is only one game and one resolution. See the following:
A 12% overclock (core and memory) gave a 9% boost in Crysis, a 13% boost in Unigine, and a 10% boost in BF3.
A 9% boost in the core plus a much larger boost in the memory clock (28%) results in 11% in Tomb Raider, 4% in Metro (although "considerably more performance" at 1080p), and 12% in Bioshock.
A 12.6% boost in core and 20.1% boost in memory gives you 8.5% in FireStrike, 10.6% in FireStrike Extreme, and 9.9% in FarCry 3 (see the charts at Hexus).
An 11.5% boost in core and 20% boost in memory gave an average of 12.45% in Bioshock, Tomb Raider and Hitman.
So boosting the core and memory in tandem clearly scales performance very, very well.
Last edited:
The variability in the OC results -- with some games showing near linear performance gains, and others showing more modest gains -- might be due to thermal throttling. If true, AIBs could get a lot of mileage out of good cooling solutions (which is exactly what I was asking about in my original post).
I guess we'll have to wait and see. But results so far suggest that AIBs will have a chance to distinguish themselves and offer some really premium solutions (likely for really premium prices).
I guess we'll have to wait and see. But results so far suggest that AIBs will have a chance to distinguish themselves and offer some really premium solutions (likely for really premium prices).
the R9 290X scales very well with overclocks. but the problem is the stock cooler is holding it back even with fan maxed out. we need the matrix, toxic and lightning (new 3 fan version) to run this card at 1.2 ghz without any throttling on air. custom coolers like prolimatech mk-26, arctic accelero hybrid are also capable of keeping the card running at max oc without throttling.
XiandreX
Golden Member
- Jan 14, 2011
- 1,172
- 16
- 81
The post on overclockers running the Mk-26 ran extremely cool compared to reference.
http://forums.overclockers.co.uk/showthread.php?t=18551550
stock clocks and fan speeds
85'C in heaven on Stock cooler
55'C in heaven on MK-26 cooler with random case fans we found in the office
Overclocked at 1200 Core + 1.4V
stock cooler at 85'c with 100% fan speed
MK-26 cooler at 72'C with silent fans.
I don't know about you but those Temperature delta differences are Massive.
:awe:
yeah the cooler makes a lot of difference. the custom coolers will allow the R9 290X to run at 1.2+ Ghz without throttling. :biggrin:
XiandreX
Golden Member
- Jan 14, 2011
- 1,172
- 16
- 81
I cant wait for the AIB 290x units to get reviewed. This is truly a VERY exciting time for Performance/Price ratio and at 1200 Core... well thats pretty darn impressive. :biggrin:
guskline
Diamond Member
- Apr 17, 2006
- 5,338
- 476
- 126
EK already has a water block out. I have 2 EVGA GTX670 FTWs under water using a EK block and a XSPC block (got the XSPC block used for a great price). Temp differences from a stock fan are amazing. For example running the Valley Benchmark the MAX temp on my water cooled cards are 48C/38C respectively.
Seems like the R9 290X was destined for water cooling or a serious aftermarket air cooler.
Seems like the R9 290X was destined for water cooling or a serious aftermarket air cooler.
XiandreX
Golden Member
- Jan 14, 2011
- 1,172
- 16
- 81
Again like has been mentioned, we all agree that the reference cooler is crap for the most part. However most people that run overclocks or even in general prefer AIB solutions as Temps are a lot lower, they run with far less noise often enough and overclocking in general is better.
What I don't get is that the 7970 has actually gone up in price when I checked yesterday - it is now 350-400$. What the eff, man, this is like a time warp and things have gone backwards.
But yeah, the 7970 when it was at that price was a steal. Too bad everyone jacked the price up on it, not sure what's up with that.
Are you being serious? It's significantly better. If it isn't, then that only goes to show how much of an engineering marvel GK110 is over the much smaller Hawaii chip - to be over 100mm^2 larger and consume less power while simultaneously running cooler.
Are you kidding me, dude? It's way better. It's easily the best reference cooler of all time - in the past, reference coolers were always considered duds. The 690 and Titan reference cooler changed all of that, now people WANT reference cooled NV cards. It is that good, yes. It does cost a sizable chunk more $ than the cheap plastic shroud that AMD uses, but I think it's probably worth it (in terms of higher SKU cost) if you want a ref cooler.
That said, aftermarket coolers are STILL better (and is very much my preference) due to higher OC headroom and what not, but some people need reference for small form factors and SLI. And you can't really beat the titan shroud in that instance.
A bigger chip that consumes less power is always going to run cooler that a smaller chip that consumes more power if they are cooled by the same cooler.
What do you think makes it better? AFAIK both are extremely similar. The biggest difference I see is that Titan's cooler has some fins behind the radial fan while AMD's reference cooler is fully exhausting.
BTW. Reference 7800GTX 512/7900 GTX cooler was also great for its time.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter