It's pretty clear that the card settles into an equilibrium, and also note that most reviewers perform multiple runs per benchmark which is important since CPU and GPU have boosts that are non-deterministic. Also note that there are no base clocks.
Yep, it needs additional cooling.
So its the same as the Titan, tho the TItans delta is much smaller. The boost is simply for benchmark only. What a hoax. And another example on why we need to be extra careful when the delta between boost and baseclock is high. And on any boost/turbo clock at all.
If 30% slower in prolonged gaming. Then its borderlining to a cheat.
Even the Uber mode starts to throttle badly. It cant even keep its baseclocks! This is a (short) benchmark only card. Outside that its performance dissolves.
The industry trend is towards smaller enclosures and more efficient components. It would appear that almost everyone is heading in this direction except AMD (Exhibit A: 220W FX-9590, Exhibit B: 290x). One of the real advantages of the Titan is that you can use it in small form factor PC's. Which again SFF is quite popular for LAN boxes. Who wants to lug around a full tower case? There's absolutely no way you would want a 290x in a small form factor case even if it could physically fit. It's complete malarkey to claim everyone is going to be doing this. Remember the GTX 470? Nvidia hasn't repeated that mistake in 3 years, why would they suddenly reverse course after the public beating they took over that release?
Unfortunately, TPU doesn't say what load they used. I guess Furmark, because in no review that investigates clocks while gaming have I seen such low clocks in either mode.
In short benchmarks? No, In prolonged? Oh yes... :awe:
Considering the 290X cant even keep baseclocks without throttle. Its simply a broken card that needs RMA.
Were you this concerned since boost 2.0 has the benchmark "cheat" mode as you call it which drops when hot?
It's pretty clear that the card settles into an equilibrium, and also note that most reviewers perform multiple runs per benchmark which is important since CPU and GPU have boosts that are non-deterministic. Also note that there are no base clocks.
Problem with that reference is the Titan would need to boost to 1.2GHz before dropping to 1Ghz after a few minutes for it to be a valid comparison.
Unfortunately, TPU doesn't say what load they used. I guess Furmark, because in no review that investigates clocks while gaming have I seen such low clocks in either mode.
Problem with that reference is the Titan would need to boost to 1.2GHz before dropping to 1Ghz after a few minutes for it to be a valid comparison.
THG measured full fan speed at 72.9dB. Who would want to listen to that racket? A vacuum cleaner is typically 70-80dB. So all the dust buster jokes of the past would actually be factually accurate when applied to this card.
Titan only changes 30Mhz at least. The 290X drops several 100Mhz.
Considering the 290X cant even keep baseclocks without throttle. Its simply a broken card that needs RMA.
They didnt use furmark. And they are not the only ones to observe it.
Most reviews just run short benchmarks.
Happens to 780 also.
Nothing new there.
And despite that reviewers still said that this card performs as good as the GTX 780 and Titan.
Imagine what they will do with a non crappy cooler.
First off, the 470 was pretty universally trashed for it's dreadful ergonomics. There may have been a few diehard Nvidia people that argued otherwise, but they were by far the exception to the rule. So the 290x deserves the same treatment here.
Secondly, the Titan cooler was said to have cost $80, not an additional $80 vs a standard cooler. So unless AMD paid nothing for the 290x cooling system (and based on its performance they got ripped off if they did pay more than nothing), adding a Titan style cooler would not add $80. Also, part of the huge cost of the Titan cooler was for cosmetic reasons which didn't actually improve performance. At $550, the 290x is $450 less than a Titan. That is ridiculous, in a good way. But it is also irrelevant to a number of people who don't want to have to live with the terrible ergonomics. AMD would have been much smarter to release the card at $600 with a better cooler that would appeal to a larger audience. Adding $50 to $550 would lose AMD fewer customers than the number they would gain having an even average performing cooling system.
So you could get the 780 on the 24th of April? On what planet?
Edit: Also, please list the companies that will honor their warranty after you flash your card with a bios that removes all protections from the card.
HT4U, CB and now hardwarecanucks preheat the cards for at least 15 minutes. None of these sites has observed a 30% performance difference between immediate measurements and preheated measurements.
What TPU measured there is certainly not representative.
Yes it does, in short benchmarks. Prolonged gaming and the 290X is nothing but hot air. The only way it beats a Titan in prolonged gaming is if you put it under water.
In terms of performance though, it won't change much of how we test. All this means now is that we needed to "warm up" the GPU each time we were ready to benchmark it. I tended to sit in the game for at least 5 minutes before running our normal test run and I think that is plenty of time to get the GPU up to its 95C operating temperature and push clocks to realistic levels. Be wary of benchmarks results that DO NOT take that into account as they could be 10%+ faster than real-world results would indicate.
The R9 290X doesn't run away with the performance crown, but it does it get it back, and for AMD that is something that it will be able to hang its hat on.
clocks after 5-8 minutes of gaming.
http://www.pcper.com/reviews/Graphi...-Taking-TITANs/PowerTune-and-Variable-Clock-R
You are incorrect.
The benchmarks were made after the card was warmed up.
For example, from PCPER.
As anyone can see, the R9 290X can match or beat the 780 and Titan even with clocks as low as 800MHz.
But it is, lets just take a look at Hardwarecanucks vs Anandtech.
Even while hardwarecanucks is still not perfect. It already shows the 290X losing 10-15% performance.
AMD's new R9 290X posts another solid score on the board with Metro: Last Light, matching the GTX Titan at 1920x1080 and besting it at 2560x1440. Clearly the GTX 780 is losing more of these battles than it is winning.