[hardocp] 4 Weeks with Radeon R9 290X CrossFire

[SimianR]

It seems I jump from NVIDIA to AMD and back again all the time and I'll be honest, I never seem to notice the presence of or the lack of PhysX in the games I'm playing. Maybe it's the games I'm playing or maybe it's just me.. but I'm not sure that it makes much of an impact (for me). I guess it's a nice bonus but I don't know if it ever materialized as a strong selling point.

Mantle and GSync look like they could both shake up the market though and offer compelling reasons to buy either brand.

 

[thilanliyan]

Is that the real reason he upgraded from his 2600K? :hmm:

2600K was never on x58 was it?

 

[Imouto]

I remember pretty clearly some people asking the sirens in the party to not phaselock enemies because they couldn't see anything in any narrow place. And sirens outnumber any other character in a 1:4 ratio.

It's almost always like this:

http://www.youtube.com/watch?v=IIDqCxFaQs8

 

[Final8ty]

PhysX was awful in Borderlands 2. Overdone to the point that you have to turn it off in any coop game.

Indeed.
12-01-2012

Around 8 of my friends have been playing BL2 and we all have put PhysX Low as its an over the top visual hindrance even on medium, the particles and debris flying through the air did look good but it got in the way of the gameplay.

My overall feeling comes on two fronts; co-op and single player. In co-op games that use PhysX for extra particles etc., I actually found the lack of PhysX a blessing. Up until this point, I had not realized that all the extra crap on the screen had been interfering with my gameplay. When you get four people in a map blowing stuff up there is so much debris on the screen that it makes it difficult to see very well. With that in mind, the lack of PhysX on the 290X cards made it a lot easier to play when the action was really taking place. On the flipside, in a game like Metro: Last Light, PhysX can certainly help set the mood and ambiance of a scene where a lot of the times the action is slower paced and things are getting creepy.

 

[5150Joker]

To be fair to the 290x, when I OC my Titans and they're under load, the air they pump out is pretty damn hot. And the metal brackets become very hot, enough to where I'd get a burn if I touch them.

 

[blackened23]

Lol. Kyle can be an angry man. But at least the writing doesnt sound like he is in a sort of twisted pain (so is xx) and at the same time singing omfg this is the best thing ever to happen. He is quite straight about things - brent too. Like Dustin at AT.

I remember they blasted AMD for being far less smooth than nvidia in mGPU setups. I'm sure all AMD users felt the same then. As far as the article goes, I just find it funny. Comparable performance with the 290X (although I should mention, OC'ing changes everything, the GTX 780 is significantly faster than 290X when OC'ed) So anyway, Titan and 290X have similar performance, but 290X being louder, hotter and having software issues is better. It's a logical conclusion, it makes complete sense. After all, when I do a hardware sidegrade with the same performance, I *definitely* want hardware that is significantly louder, hotter, having worse software, and potential issues. With the same performance.

Speaking of which, AMD is just now getting around to fixing CF/eyefinity microstutter on the 79xx and 280x / prior GPUs:

http://hardforum.com/showthread.php?t=1796775

Two years, 1 month later. Getting fixed.

I don't know. AMD taking two years to do a fix for microstutters on prior GPUs doesn't make me entirely confident in them right now.

 

[Erenhardt]

and potential issues

10/10

 

[SlowSpyder]

After all, when I do a hardware sidegrade with the same performance, I *definitely* want hardware that is significantly louder, hotter, having worse software, and potential issues. With the same performance.

Louder - Yes, the 290X/290 are louder. But it seems most people who actually use those cards seem to feel that the issue is way overblown on forums. They're louder, but not so much that it really impacts anything, it seems. There are wins on paper and then there is how it is in the real world.

Hotter - Other than a number on monitoring software, what impact does this have? Why is this a negative you list? Do I care if my GPU runs at 77C or 90C if it is within spec and running fine?

Having worse software, and potential issues - This is a total opinion and nothing more. Nvidia has laid some real turds of drivers in the past. Both have issues, depending on each person's setup and the software/games they use, one could have no issues or more issues with one company's GPU or the other.

Performance - Performance is probably close enough that I wouldn't bother switching one out for the other. But the author clearly states that he can play games at settings now that he couldn't before. CF scales really well on the new cards.

 

[96Firebird]

2600K was never on x58 was it?

Ah you're right, my bad.

 

[ICDP]

and even the OC 780 far outperforms the 290X. But we'll go with stock.

the GTX 780 is significantly faster than 290X when OC'ed) So anyway

This is [false], now you just seem to be making crap up. I own a GTX780 MSI Gaming and now an R9 290X (290 unlocked) and I can assure you max OC vs max OC on my two cards the GTX780 is almost always playing catchup. It matches an R9 290X at best and at worst is 20% slower.

Of course it depends on the game and while my sample list is small the GTX780 has not won one single test.

MLL is equal
Tomb Raider GTX780 is ~15% slower
Hitman GTX780 is ~20% slower
Arma 3 GTX780 ~10% slower
Crysis 3 is equal

GTX780 is 1241/6600 with a modded overvolt BIOS. 1202/6600 is the best it gets with stock BIOS.

R9 290X is 1170/5600 with +100Mv

In all honesty the R9 290X is a faster card overall and while it is fairly close in a good number of games OC vs OC, in the rest it was quite a bit slower. It most certainly is not a case of the GTX780 OC far outperforming the R9 290X as you claim. No doubt you are still stuck with the non overvolted release day R9 290X overclock results.

Profanity isn't allowed in the technical forums.
-- stahlhart

 

[Imouto]

But... but... Noise! The noise!

 

[blackened23]

This is [false], now you just seem to be making crap up. I own a GTX780 MSI Gaming and now an R9 290X (290 unlocked) and I can assure you max OC vs max OC on my two cards the GTX780 is almost always playing catchup. It matches an R9 290X at best and at worst is 20% slower.

Of course it depends on the game and while my sample list is small the GTX780 has not won one single test.

MLL is equal
Tomb Raider GTX780 is ~15% slower
Hitman GTX780 is ~20% slower
Arma 3 GTX780 ~10% slower
Crysis 3 is equal

GTX780 is 1241/6600 with a modded overvolt BIOS. 1202/6600 is the best it gets with stock BIOS.

R9 290X is 1170/5600 with +100Mv

In all honesty the R9 290X is a faster card overall but it is fairly close in most games OC vs OC. It most certainly is not a case of the GTX780 OC far outperforming the R9 290X as you claim. No doubt you are still stuck with the non overvolted release day R9 290X overclock results.

I'm going to go on a limb and state that it isn't me making crap up. This is an overclocked GTX 780:

Fact 1) This is 20% faster than the GTX Titan with a meager mid 1100s overclock.
Fact 2) 290X is even with Titan in uber mode.
Fact 3) Therefore, this puts GTX 780, when overclocked, 20% faster than the Titan and the 290X.

By all means provide proof with a reputable tech website showing such benchmarks. I have not seen the mythical 290X that is 20% faster than Titan. Based on that fact, the OC'ed 290X is slower than the OC'ed GTX 780.

Furthermore, linus at linustechtips does all reviews with maximum overclocks. In both his 290 and 290X review, the GTX 780 OC outperformed the 290X at maximum overclocks.

So please. If you're not "making crap up" (your phrase, not mine) provide proof. With a reputable tech website. Every website doing maximum overclocks has shown the 780 to outperform the Titan by 15-20%, and the 290X cannot approach this.

The ball is in your court. Show me the 290X that is 15-20% faster than Titan averaged out in 10+ games. And yeah, i'll go ahead and dismiss your "personal" testing in which you just so happen to own a 780 and 290X - like I said, show me a reputable hardware website with similar results of a 290X being 15-20% faster than Titan. TPU is showing the GTX 780 OC to do just that.

 

[AnandThenMan]

I really enjoyed this article, it's subjective stuff yes but with this type of writing style you get a better picture of what it's like to live with the hardware. I hope we see more write-ups like this.

 

[Imouto]

It's so funny when Blackened23 uses aftermarket coolers when it fits his agenda. Lets check the OC with the reference cooler:



Is that a 6%?

You currently have three active infraction points for callout.
-- stahlhart

 

[blackened23]

I really enjoyed this article, it's subjective stuff yes but with this type of writing style you get a better picture of what it's like to live with the hardware. I hope we see more write-ups like this.

oh, I see you ninja edited.

 

[blackened23]

It's so funny when Blackened23 uses aftermarket coolers when it fits his agenda. Lets check the OC with the reference cooler:

Is that a 6%?

We can happily revisit this when custom 290X cards are out. Now, I personally think that the Hawaii chip was pushed to the limit and won't overclock all that well, but I could be wrong. If I am wrong, you're free to call me out on it via PM and I will happily admit my error. If it overclocks like crazy i'll be the first to admit my error and praise aftermarket 290X cards - i'm actually quite looking forward to them, believe it or not, because they will address all of my qualms with the vanilla design. All of my reservations with 290X will be a non issue once aftermarket 290X cards arrive.

Anyway, to address your post - uber mode 290X is even with Titan which is less than the 6% faster than Titan the reference 780 is achieving here. Right? Is my math incorrect? 0% faster than Titan versus 6% faster than Titan. Or you could just get a custom card which is 15-20% faster than Titan.

 

[thilanliyan]

We can happily revisit this when custom 290X cards are out. Now, I personally think that the Hawaii chip was pushed to the limit and won't overclock all that well, but I could be wrong. If I am wrong, you're free to call me out on it via PM and I will happily admit my error.

I'm pretty sure the card clocks well, IF you can cool it. ICDP's overclock on the 290 seems to be pretty good from a stock 1000MHz doesn't it?

 

[Imouto]

We can happily revisit this when custom 290X cards are out. Now, I personally think that the Hawaii chip was pushed to the limit and won't overclock all that well, but I could be wrong. If I am wrong, you're free to call me out on it via PM and I will happily admit my error. If it overclocks like crazy i'll be the first to admit my error and praise aftermarket 290X cards - i'm actually quite looking forward to them, believe it or not, because they will address all of my qualms with the vanilla design.

I don't think so. You will find anything to blame these cards. The package, the smell or whatever. You've been quite busy doing it for the last month.

Anyway, to address your post - uber mode 290X is even with Titan which is less than the 6% faster than Titan the reference 780 is achieving here. Right? Is my math incorrect? 0% faster than Titan versus 6% faster than Titan.

Do you even read your messages before publishing them? An overclocked 290X is 9% faster than a stock Titan.

http://www.techpowerup.com/reviews/AMD/R9_290X/31.html

An overclocked 290 is 8% faster than a stock Titan.

http://www.techpowerup.com/reviews/AMD/R9_290/29.html

And yeah, your math is cool but you try really hard to paint a better picture for NV. Why suddenly the 290X can't overclock?

 

[blackened23]

I'm pretty sure the card clocks well, IF you can cool it. ICDP's overclock on the 290 seems to be pretty good from a stock 1000MHz doesn't it?

I've been wrong once or twice (heh) and if aftermarket cards have amazing OC potential, i'll be happy to state such. I mentioned this earlier but i'm very much looking forward to seeing what custom 290X cards can do - I do like the fact that aftermarket 290X cards will address all of the qualms that I currently have with Hawaii. It will turn a product that i'm not interested in into one that I could see myself getting for a secondary rig and/or recommending to others.

So, we'll see. I just want verifiable proof from tech websites, you know what I mean? In terms of actual, OC'ed gaming benchmarks. If it turns out that the aftermarket 290X cards are just flat out amazing for Oc'ing, you will never see me state anything negative about them. Promise.

 

[IEC]

I'm not sure why a certain poster here needs to rag on the R290X and R290 users. I got a sweet deal on my 290X (<$370 Radeon 290 unlocked to 290X) and it's an unbeatable value for high end gaming at that price.

 

[el etro]

This is [false], now you just seem to be making crap up. I own a GTX780 MSI Gaming and now an R9 290X (290 unlocked) and I can assure you max OC vs max OC on my two cards the GTX780 is almost always playing catchup. It matches an R9 290X at best and at worst is 20% slower.

Of course it depends on the game and while my sample list is small the GTX780 has not won one single test.

MLL is equal
Tomb Raider GTX780 is ~15% slower
Hitman GTX780 is ~20% slower
Arma 3 GTX780 ~10% slower
Crysis 3 is equal

GTX780 is 1241/6600 with a modded overvolt BIOS. 1202/6600 is the best it gets with stock BIOS.

R9 290X is 1170/5600 with +100Mv

In all honesty the R9 290X is a faster card overall and while it is fairly close in a good number of games OC vs OC, in the rest it was quite a bit slower. It most certainly is not a case of the GTX780 OC far outperforming the R9 290X as you claim. No doubt you are still stuck with the non overvolted release day R9 290X overclock results.

Profanity isn't allowed in the technical forums.
-- stahlhart

Remember HardOcp did this last "test" with a Monster chasis and a trio of monster fans blowing and spreading out air for help the 290x goes cool. But i agree with you, there's no way GTX 780 can beat a full 1Ghz locked 290x. Out-of-Box GTX 780 can even beat the Non-X variant(Nor over-vs-over nor stock)....

I'm going to go on a limb and state that it isn't me making crap up. This is an overclocked GTX 780:

Fact 1) This is 20% faster than the GTX Titan with a meager mid 1100s overclock.
Fact 2) 290X is even with Titan in uber mode.
Fact 3) Therefore, this puts GTX 780, when overclocked, 20% faster than the Titan and the 290X.

By all means provide proof with a reputable tech website showing such benchmarks. I have not seen the mythical 290X that is 20% faster than Titan. Based on that fact, the OC'ed 290X is slower than the OC'ed GTX 780.

Furthermore, linus at linustechtips does all reviews with maximum overclocks. In both his 290 and 290X review, the GTX 780 OC outperformed the 290X at maximum overclocks.

So please. If you're not "making crap up" (your phrase, not mine) provide proof. With a reputable tech website. Every website doing maximum overclocks has shown the 780 to outperform the Titan by 15-20%, and the 290X cannot approach this.

The ball is in your court. Show me the 290X that is 15-20% faster than Titan averaged out in 10+ games. And yeah, i'll go ahead and dismiss your "personal" testing in which you just so happen to own a 780 and 290X - like I said, show me a reputable hardware website with similar results of a 290X being 15-20% faster than Titan. TPU is showing the GTX 780 OC to do just that.

Linustech is a not a good source IMO. Techpowerup does the test with a much better metodology, but they select games in order to favor Nvidia in test(Battefield 4 didn't appear is test until now and Blacklist entered to tests since day one). Hardware Canucks is a better source of information IMO. Ht4u is AMD biased, Tom's IMO are parcial(last year they accused here AMD to make changes in texture quality in order to cheat in benches) The best sources IMO are anand and computerbase(I have two sources for my country if you want good Brazilian sources too, they are Hardware Secrets and Adrenaline).

 

[Fastx]

Nice deal IEC!

BK
I don't see the AIB's being amazing oc'er (290/290x) but 1225 - 1250 wouldn't be bad least on the 290 and if possible keep the price under $450.00.

 

[Pandora's Box]

But... but... Noise! The noise!

I'd like to point out that a 780 or 780ti overclocked and overvolted puts out a lot of heat and noise. Not as much noise as a 290 but it's up there.

 

[Silverforce11]

This is [false], now you just seem to be making crap up. I own a GTX780 MSI Gaming and now an R9 290X (290 unlocked) and I can assure you max OC vs max OC on my two cards the GTX780 is almost always playing catchup. It matches an R9 290X at best and at worst is 20% slower.

Of course it depends on the game and while my sample list is small the GTX780 has not won one single test.

MLL is equal
Tomb Raider GTX780 is ~15% slower
Hitman GTX780 is ~20% slower
Arma 3 GTX780 ~10% slower
Crysis 3 is equal

GTX780 is 1241/6600 with a modded overvolt BIOS. 1202/6600 is the best it gets with stock BIOS.

R9 290X is 1170/5600 with +100Mv

In all honesty the R9 290X is a faster card overall and while it is fairly close in a good number of games OC vs OC, in the rest it was quite a bit slower. It most certainly is not a case of the GTX780 OC far outperforming the R9 290X as you claim. No doubt you are still stuck with the non overvolted release day R9 290X overclock results.

He's been on a tirade ever since the R290s launched, pulling rubbish numbers out to claim 780s are faster when we all know its false. Then he repeatedly beat the dead horse that is thermal throttling... in Quiet Mode, where its designed to throttle more due to temps. TBH its very tiring having R290 threads crapped on by the same worthless argument over and over.

As I said in the other post, if he keeps pulling those numbers out, soon, he'll claim gtx770 OC is faster than R290s OC too.

Now the goal post is shifted, Hawaii can't even OC since its running at its limits. Well duh, if you can't up its voltage during launch reviews. Now that people can add 150mv to the core and crank up fanspeed or go custom cooling, we are seeing 1.2ghz+ often on OCnet and other users.

The loudest complaints are from GTX using, NON R290 owners, and that ought to ring a bell.

 

[IEC]

I'm planning on building a custom loop for water cooling sometime in the next month or so and hope to OC to 1200 MHz+ core