Titans sold out :(

[johnksss]

Yes, that is from my R1 when I had the es 2920xm CPU & 580m's.

 

[lavaheadache]

Don't forget every little detail, lavaheadache might call you out on it hahahaha. :awe:

=) that little detail is an extra $1000

 

[Lepton87]

Yes, that is from my R1 when I had the es 2920xm CPU & 580m's.

Why is there an AMD logo?

 

[chimaxi83]

Why is there an AMD logo?

Lol AMD logo on the board, explain that one away

 

[BallaTheFeared]

Lol AMD logo on the board, explain that one away

:hmm:

 

[5150Joker]

Why is there an AMD logo?

What does it matter? It's his old system, he probably forgot because he eventually upgraded it to 580M SLI. His current system is entirely different. So thanks to a new modded bios released by our buddy svl7 I was able to match my old 680M SLI setup:

1. Titan: http://www.3dmark.com/3dm11/6395487 Not the most stable at those clocks (1201 mhz core) but least it managed it.
2. Old M18x with 680M SLI: http://www.3dmark.com/3dm11/5304395

Compare link: http://www.3dmark.com/compare/3dm11/6395487/3dm11/5304395 Anyone who says 680M SLI can't reach Titan speeds should remember these numbers.

 

[Grooveriding]

ahaha nice bust.

 

[johnksss]

Why is there an AMD logo?

oh, my bad. that was my junky amd 6990m's.

ahaha nice bust.

Really? A mistake is a nice bust huh? Interesting.

ahaha nice bust.

Lol AMD logo on the board, explain that one away

Seriously?

 

[Firestorm007]

oh, my bad. that was my junky amd 6990m's.


Really? A mistake is a nice bust huh? Interesting.

 

[johnksss]

ahaha nice bust.

For the non competive


But let me know when one of you actually beat any of my scores though.

 

[5150Joker]

Forget it.

 

[AdamK47]

Stop with this thread.

 

[johnksss]

Best to delete it now

 

[AnandThenMan]

Are you guys really that dumb or just trolling? Last time I was on AT video section a few years back it had some reasonably intelligent posters (keysplayer, wreckage, rollo, bfg10k and a few others). Where'd they all go?

 

[wand3r3r]

LOL

Semi On topic, pretty impressive 680M SLI when "un" neutered. Are they cooled with LN2 or just air in a stock system?

 

[johnksss]

LOL

Semi On topic, pretty impressive 680M SLI when "un" neutered. Are they cooled with LN2 or just air in a stock system?

Cooled with just air or chilled air.

 

[RussianSensation]

LOL! But But..680M SLI could never be this fast! It's slower than 660 Ti remember! Let me go find some graphs and act like your scores don't exist..BRB!

it seems you love moving goal posts - first you were talking about 1440P Titan vs. GTX680M SLI and omitted to point out you were comparing 680M SLI at 1080P vs. Titan at 1080P since obviously you cannot reproduce benchmarks for us at 1440P on the 680M SLI setup.

The comparison was being made between a stock 680M and a stock 660Ti, not 680M OC to 1100mhz vs. a stock 660Ti. Again, moving goal posts.

I specifically linked benchmarks that show how much faster GTX670/680 are over a stock GTX680M by using 660Ti as a reference point since that card itself is faster than a stock 680M. If you cannot follow this logic, I can't help you. From that point, the Titan could be compared to 660Ti/670/680 and it can be seen in games that at 1440P, the Titan is often 2x faster than GTX670 SLI. GTX670 on its own is nearly 50% faster than a stock 680M.

All the benchmarks you and your buddy linked show an 1100mhz+ overclock on the 680M but the bandwidth still does not even come close to 192GB/sec of the stock 670. Since you aren't showing us scores at 1440P, the 680M's memory bandwidth is not being stressed, which means your comparison isn't really valid. If you wanted to talk about 1080P, you should have said so from the beginning when you contested Adam's claim.

Further, 3DMark scores are meaningless since Adam, myself and others in this thread were discussing real world games. Furthermore, you back-tracked on your own statement when you stopped comparing benches @ high resolutions. Since we were talking about 1440P as that was your entire point of discussion, you have continuously failed to acknowledge what would happen to a card with 680M's specs at that resolution. We just look at 660ti to see what happens. Even though it's very competitive at 1080P, it doesn't do so well at 1600P.

http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-660-ti/7/

The only thing you are proving is that 680M SLI OC fast at 1080P where the scores are more constrained by GPU power, rather than memory bandwidth. This is why something like an 1100mhz overclocked 680M SLI can keep up with a Titan. Again, no one was talking about 1080P resolution since the context Adam / I were coming from in response to you was your complaint that at 1440P Titan was slower than your 680M SLI.

Unless you have real world benchmarks of 680M OC SLI at 1440P or even 1600P, then you only proved that 680M SLI OC can beat a stock Titan at 1080P. Then since we cannot test 680M OC SLI at 1440P/1600P, we have to use desktop GPUs as the 2nd most reasonable measuring stick. In those cases at 1440P, we can clearly see that in certain games a stock Titan = GTX670 SLI and a 680M would need a 42% GPU overclock and a 65% memory overclock to match a stock GTX670 at 1440P. I am not seeing a 65% memory bandwidth overclock on the 680M.

Are you guys really that dumb or just trolling? Last time I was on AT video section a few years back it had some reasonably intelligent posters (keysplayer, wreckage, rollo, and a few others). Where'd they all go?

Not sure if serious.

They got rightfully banned.

You mean to tell us NV's focus group members and NV fanboys are the most reasonably intelligent members on this tech forum? At one time Rollo tried to convince the forum that buying a 7800GTX 512MB is worth it since it would appreciate in value and remain stable for 6+ months. He also infamously defended GeForce 5 series. Considering it was the worst NV GPU series ever made (atrocious AA performance/IQ, horrendous texture filtering/mip-map AF quality, non-existent DX9 performance), his credibility fell out the window. After it was found that rollo was an NV focus group member after all his denials, nothing he said after could be taken seriously.

LOL

Semi On topic, pretty impressive 680M SLI when "un" neutered. Are they cooled with LN2 or just air in a stock system?

This. Also, there is a difference between running 680M SLI uber-overclocked for 10 min to run a bench, let them cool down and re-run a new bench vs. playing a game for 3-4 hours. The Titan OC is 100% stable for 24 hours. Video would be nice of stock cooled Alienware laptop with 680Ms at 1100mhz x 2 and CPU overclocked too.

 

[Fx1]

it seems you love moving goal posts - first you were talking about 1440P Titan vs. GTX680M SLI and omitted to point out you were comparing 680M SLI at 1080P vs. Titan at 1080P since obviously you cannot reproduce benchmarks for us at 1440P on the 680M SLI setup.

The comparison was being made between a stock 680M and a stock 660Ti, not 680M OC to 1100mhz vs. a stock 660Ti. Again, moving goal posts.

I specifically linked benchmarks that show how much faster GTX670/680 are over a stock GTX680M by using 660Ti as a reference point since that card itself is faster than a stock 680M. If you cannot follow this logic, I can't help you. From that point, the Titan could be compared to 660Ti/670/680 and it can be seen in games that at 1440P, the Titan is often 2x faster than GTX670 SLI. GTX670 on its own is nearly 50% faster than a stock 680M.

All the benchmarks you and your buddy linked show an 1100mhz+ overclock on the 680M but the bandwidth still does not even come close to 192GB/sec of the stock 670. Since you aren't showing us scores at 1440P, the 680M's memory bandwidth is not being stressed, which means your comparison isn't really valid. If you wanted to talk about 1080P, you should have said so from the beginning when you contested Adam's claim.

Further, 3DMark scores are meaningless since Adam, myself and others in this thread were discussing real world games. Furthermore, you back-tracked on your own statement when you stopped comparing benches @ high resolutions. Since we were talking about 1440P as that was your entire point of discussion, you have continuously failed to acknowledge what would happen to a card with 680M's specs at that resolution. We just look at 660ti to see what happens. Even though it's very competitive at 1080P, it doesn't do so well at 1600P.

http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-660-ti/7/

The only thing you are proving is that 680M SLI OC fast at 1080P where the scores are more constrained by GPU power, rather than memory bandwidth. This is why something like an 1100mhz overclocked 680M SLI can keep up with a Titan. Again, no one was talking about 1080P resolution since the context Adam / I were coming from in response to you was your complaint that at 1440P Titan was slower than your 680M SLI.

Unless you have real world benchmarks of 680M OC SLI at 1440P or even 1600P, then you only proved that 680M SLI OC can beat a stock Titan at 1080P. Then since we cannot test 680M OC SLI at 1440P/1600P, we have to use desktop GPUs as the 2nd most reasonable measuring stick. In those cases at 1440P, we can clearly see that in certain games a stock Titan = GTX670 SLI and a 680M would need a 42% GPU overclock and a 65% memory overclock to match a stock GTX670 at 1440P. I am not seeing a 65% memory bandwidth overclock on the 680M.



Not sure if serious.

They got rightfully banned.

You mean to tell us NV's focus group members and NV fanboys are the most reasonably intelligent members on this tech forum? At one time Rollo tried to convince the forum that buying a 7800GTX 512MB is worth it since it would appreciate in value and remain stable for 6+ months. He also infamously defended GeForce 5 series. Considering it was the worst NV GPU series ever made (atrocious AA performance/IQ, horrendous texture filtering/mip-map AF quality, non-existent DX9 performance), his credibility fell out the window. After it was found that rollo was an NV focus group member after all his denials, nothing he said after could be taken seriously.



This. Also, there is a difference between running 680M SLI uber-overclocked for 10 min to run a bench, let them cool down and re-run a new bench vs. playing a game for 3-4 hours. The Titan OC is 100% stable for 24 hours. Video would be nice of stock cooled Alienware laptop with 680Ms at 1100mhz x 2 and CPU overclocked too.

1 Titan IS NOT 2x Faster than a 670 SLI!!! Where do you get this rubbish from?

670 GTX SLI is basically a 690 GTX or 95% of one. 690 OWNS Titan

Do the Math

 

[RussianSensation]

1 Titan IS NOT 2x Faster than a 670 SLI!!! Where do you get this rubbish from?

670 GTX SLI is basically a 690 GTX or 95% of one. 690 OWNS Titan

Do the Math

I didn't say on average the Titan is 2x faster than GTX670 SLI but it happens. Looking at 680M vs. 670, you can see GTX680M SLI would need a massive overclock to match GTX670 SLI. Looking at where GTX670 sits in some games, GTX670 SLI would barely match a stock Titan
http://gamegpu.ru/images/stories/Test_GPU/strategy/Company of Heroes 2/test/coh 2 2560.jpg

That's against a full-fledged GTX670 that 680M cannot touch since you cannot get 192GB/sec memory bandwidth on the 680M on air cooling.

His argument gets worse once you look at Titan OC which is only fair if comparing against 680M SLI OC. Titan OC is smashing a GTX680 in some games like Metro 2033 where memory bandwidth counts.
http://www.xbitlabs.com/images/graphics/nvidia-geforce-gtx-titan/07_metro.png

Since 680M OC cannot touch a stock GTX680 and since SLI scaling in Metro 2033 is not 90-100%, Titan OC would beat 680M SLI OC in Metro 2033 at high resolution.

So that's at least 2 games already where 680M OC SLI wouldn't be able to beat Titan OC at 1440P. 680M SLI performs really well at 1080P against the Titan, which no one disputed in the first place.

 

[Fx1]

I didn't say on average the Titan is 2x faster than GTX670 SLI but it happens. Looking at 680M vs. 670, you can see GTX680M SLI would need a massive overclock to match GTX670 SLI. Looking at where GTX670 sits in some games, GTX670 SLI would barely match a stock Titan
http://gamegpu.ru/images/stories/Test_GPU/strategy/Company of Heroes 2/test/coh 2 2560.jpg

That's against a full-fledged GTX670 that 680M cannot touch since you cannot get 192GB/sec memory bandwidth on the 680M on air cooling.

His argument gets worse once you look at Titan OC which is only fair if comparing against 680M SLI OC. Titan OC is smashing a GTX680 in some games like Metro 2033 where memory bandwidth counts.
http://www.xbitlabs.com/images/graphics/nvidia-geforce-gtx-titan/07_metro.png

Since 680M OC cannot touch a stock GTX680 and since SLI scaling in Metro 2033 is not 90-100%, Titan OC would beat 680M SLI OC in Metro 2033 at high resolution.

So that's at least 2 games already where 680M OC SLI wouldn't be able to beat Titan OC at 1440P. 680M SLI performs really well at 1080P against the Titan, which no one disputed in the first place.

You link a beta game and one that doesnt show either the 690 GTX or a 670 SLI.

Titan is NOT 2x faster than a 670 SLI.

 

[johnksss]

But it's over clocked at 1440. 24/7 clocks though...when I do over clock

 

[Will Robinson]

Quote:
Originally Posted by 5150Joker
Are you guys really that dumb or just trolling? Last time I was on AT video section a few years back it had some reasonably intelligent posters (keysplayer, wreckage, rollo, bfg10k and a few others). Where'd they all go?

Is there an atomic powered facepalm emoticon around somewhere?
Cause this one just seems inadequate...even when spammed.

 

[Mr. Fox]

Where on earth did you come up with that?

You must not be paying attention, LOL. Core clock is locked down with a stock 680M vBIOS and it throttles. It may not with a MSI 680M GPU, but Dell and Clevo 680M definitely have that problem with a stock vBIOS. As much of an NVIDIA fanboy as you are, I find it hard to believe you were not already aware of that.

Enduro crap is irrelevant on Alienware laptops based on how they are designed. Of course, the M18x does not have it (or Optimus) at all, because it has the far superior manual switching at that gives the user total control. On the M17x R3/R4 that's one of the first things that gets disabled in the BIOS. It can't be disabled on Clevo laptops... not sure about MSI. I did not mention that because it slipped my mind since it has never been relevant to me personally. If forget about the issues others have with it. I'm not a fan of Optimus either. I think both of them suck.

My 7970M CF setup was more powerful stock compared to the stock (not modded vBIOS) 680M SLI, but they sucked at overclocking, lacked the stability of 680M and the drivers were a constant pain in the tush... and of course, CF studdering, blah, blah. Glad that nightmare is over.

it seems you love moving goal posts - first you were talking about 1440P Titan vs. GTX680M SLI and omitted to point out you were comparing 680M SLI at 1080P vs. Titan at 1080P since obviously you cannot reproduce benchmarks for us at 1440P on the 680M SLI setup.

No, you are the one that keeps saying that. Over and over, in fact. If you go back and read the posts, Joker said that his Alienware M18x with 680M SLI overclocked can run as well or better than a single stock Titan. That was based on real-world performance of machines he owns and uses. I posted benchmarks that illustrate exactly that. I don't understand the logic behind tossing all kinds of graphs and theoretical mumbo jumbo into the thread to try to prove the guy wrong when the dude knows by first hand experience exactly what he is talking about. Touching and seeing is far more meaningful than what if scenarios and hypotheticals.

He never said stock 680M SLI was more powerful than a stock Titan. Please pay closer attention to what was said and stop changing things around.

I can understand how that would be difficult to believe because most people are unfamiliar with the Alienware M18x. With 680M SLI it outperforms many desktop gaming rigs with a single GPU and even some with CF and SLI, not just those with a Titan. And, there will always still be a few "doubting Thomas" that refuse to believe what they see.

 

[RussianSensation]

You link a beta game and one that doesnt show either the 690 GTX or a 670 SLI.

Titan is NOT 2x faster than a 670 SLI.

Sorry FX1, that was obviously a typo on my part. I meant Titan is often 2x faster than GTX670 or ~ GTX670 SLI. My bad. It doesn't change the main conclusion that GTX670 SLI > GTX680M OC SLI and if Titan can match a GTX670 SLI than in those games where Titan ~ GTX670 SLI, Titan OC > GTX680M OC SLI.

 

[RussianSensation]

But it's over clocked at 1440. 24/7 clocks though...when I do over clock

Is that on air cooling? If not, no one cares since the point of a gaming laptop is portability. Do you carry water cooling/ LN2 with you in your backpack? The only thing that matters for laptops is usable overclocked performance on the go. Otherwise, there is no point in getting a laptop unless you carry exotic equipment with you everywhere you travel...

I can understand how that would be difficult to believe because most people are unfamiliar with the Alienware M18x. With 680M SLI it outperforms many desktop gaming rigs with a single GPU and even some with CF and SLI, not just those with a Titan. And, there will always still be a few "doubting Thomas" that refuse to believe what they see.

Please go read posts #194 and #196 to understand the source of confusion then.

http://forums.anandtech.com/showpost.php?p=34866230&postcount=194

"So I definitely need a second, especially at 1440p with AA."

Then he says:

"No it's actually not. I'm going by real world results, not paper specs. The 680m sli was faster in every game I've tested, especially with the unlocked vbios that allowed some heavy oc with no throttling."

680M SLI is good for 1080P but 1440P is 78% more pixels. If you put up 680M SLI OC vs. Titan OC at 1440P, 680M SLI OC would lose in many games because it would run into a memory bandwidth bottleneck that's not there at 1080P. Therefore, Adam was not incorrect in post #195.

Even if we stick to 1080P, his statement can't be that universal either since 680M SLI OC would need to be at least as fast as a GTX690 to match Titan OC. GTX690 has 2 GPUs with 1536 SPs and each has 192GB/sec memory bandwidth. Since you cannot overclock GTX680M to 192GB/sec on the bandwidth side, and you can barely match it on the GPU core side, it has to be slower than a GTX690.

And like I said if you actually compared at high-rez, the Titan would smash 680M OC SLI in playability by virtue of higher/more stable minimum frames in some modern titles.

The Titan is made for high rez gaming, not 1080P. 1080P gaming on 680M SLI penalizes the Titan by virtue of many games being CPU bottlenecked and at the same time by not testing 680M SLI at 1440P or higher, it saves it from taking a dumb due to its minuscule memory bandwidth limits. It's not exactly a valid comparison to run 680M OC SLI at 1080P against Titan OC at 1080P and tell users with 30 inch monitors that 680M OC SLI setup would be faster.

On the point of synthetic benchmarks with 680M, "Initial benchmarks suggest a decisive victory for NVIDIA in three out of the four tests, but it's fair to say that this is not reflected across the run of play in our gameplay performance analysis."
http://www.eurogamer.net/articles/df-hardware-geforce-gtx-680m-review

680M OC SLI may match the Titan in 1080P in most titles, but at 1440P or above, the Titan OC would provide a smoother real world gaming experience. The canned benchmark scores of 680M vs. HD7970M show this already where 680M scores nicely in canned benches but in real world, it's entirely different.

GameGPU tests show Titan OC trading blows with GTX690. Sometimes in games like Metro 2033, GTX690 OC is only 1 fps faster than Titan OC.

Here is GTX690 with memory bandwidth of 224 GB/sec. GTX680M starts off at 115GB/sec. How can GTX680M OC SLI beat Titan OC in that game? It won't because GTX690 OC is going to beat 680M OC SLI on air.