Shootout: 780 Lightning vs 290

[blackened23]

Lol. Posting histories are quite revealing including yours but let's get this thread out of the biased gutter it's in. .

There's no unwritten rule that brand preferences aren't allowed. Are you going to tell a Cowboys fan that they're not allowed to have a preference? Oh. They're biased. Of COURSE they're biased. Everyone has their preference. I dare say based on your posting history, that yours is quite evident as well. I'm not saying that's bad, everyone has their picks and preferences. Personally i'm biased toward better products. I think the GK110 products are better products based on software features and intangible factors, and I do think it has better OC potential than Hawaii. Note you can disagree with that, and it's fine. My opinion. Last gen, I thought the 7970 was the better overclocker than GK104. Unsurprisingly, a lot of people disagreed with me on that. This gen I think the GK110 is hands down better. And of course many disagree with me. Whatever. I'm pretty particular in what makes me like a certain product. That preference is of course completely subjective and is subject to change over time.

I don't think there's any debate that GK110 scales immensely well with clockspeeds. Many websites have GTX 780 cards scaling 30% with easily achievable overclocks. (which of course makes it far faster than stock Titan). Now you can argue the 290 is better. I won't get into that again, don't care anymore, but I do know the GK110 is pretty much a beast in terms of overclock scaling. I'd say more or less that isn't "bias", but fact, as many many websites have substantiated this.

 

[ICDP]

I don't think there's any debate that GK110 scales immensely well with clockspeeds. Many websites have GTX 780 cards scaling 30% with easily achievable overclocks.

Herein lies your problem, you assume there is no debate and that the above statement cannot be refuted. Even you clearly state that many (ie not all) websites show 30% scaling. There IS a debate since I have showed that GTX780 core clocks OC does not always scale well.

Yet again you make a statement of opinion as an absolute fact. 30% OC scaling is NOT a guarantee on a GTX 780, let alone the ridiculous claim of 42% by Bella.

Here we go with Hardwarecanucks GTX780 review. ~10%-15% performance increase with Crysis 3.

http://www.hardwarecanucks.com/foru...s/61310-nvidia-geforce-gtx-780-review-14.html

GTX780 does OC better in general than R9 290X, but your claim that this is ALWAYS the case is a non-sequitor.

 

[blackened23]

NVM

 

[Imouto]

You seem to think Kepler bounces around from base to max boost during benchmark runs, and that is simply not true.

Also you seem to be projecting v-sync based variance in clocks based on a workload not at 99% usage then using that to concoct some inane scenario where a 780 which is already being throttled due to thermals is now reducing/increasing clock speed based on load which should already be at or near 99%.

Kepler will downclock with vsync on if the workload is low enough to warrant it, however it doesn't do it randomly based on nothing when vsync is off and usage is maxing out.

No Rome 2 is a rollercoaster because the load is dynamic, each area in the Rome 2 benchmark has different things going on which are harder or easier to render based on the dynamics of the image being shown. Which should be very obvious since cpu and gpu speeds are fixed.

Tomb Raider gained 40% because the core scaled with the added memory speed perfectly well, and because Tomb Raider the benchmark has very little cpu overhead, unlike Metro 2033 or Rome 2.

No they observed a max boost on their cold run of 1019MHz, when using uber in their hot box run their observed boost clock speed was 993.

If you have proof that I've lied about my results with video cards please, please share it here so everyone can see what I lair I am.

This isn't about being bias one way or another, you seem to have misunderstood the comment. Likewise I never saw you call my integrity into question when I posted the #1 fastest 7950 CF results in our forum benchmark threads.

I'm not saying that you're lying. Again, I'm saying that your results are unrealistic, made up and misleading.

The results from hardware.fr between stock and Uber for the GTX 780 is a 9.9% at 1080p while the average clocks are 876 Mhz for the stock and 993 for the Uber testing, 13.3% increase in average clockspeeds. So just removing the thermal threshold and adding a 6% power limit the card behaves way closer to your locked frequency. But hold on, you can pick the results from Sleeping Dogs for example with a 12% improvement between stock and Uber (15% faster average clock) or Splinter Cell Blacklist with a 7% (8.5% faster average clock). Just by that you can tell that GPU Boost is highly variable and you can't emulate it with a locked clockspeed.

By your own Tomb Raider figures max fps for the first bench is 27% higher than average and min fps 28% lower with a locked clockspeed. Showing that your GPU is way more or less efficient than the average in some parts of the benchmark. Depending on how GPU Boost behaves you should add or subtract an aditional X% based on other factors that your locked frequency doesn't take into account.

Your results are unrealistic because no card will behave like yours with different max and min figures. Made up because you removed the GPU Boost functionality and misleading because you're trying to pass them as the same as a stock card averaging those clockspeeds.

It's like you bought your card and rediscovered the wheel for everyone. If no one has done the same testing as you since the Titan launch is because all the reasons above.

As for your CF results I couldn't care less, I don't have the urgency to butt into every single thread.

 

[BallaTheFeared]

lol

 

[Imouto]

I won't link benchmarks again proving 30% scaling.

Elfear told you already but you're using the term "scaling" wrong like you do with "efficiency".

they're based on data of websites such as hardwarecanucks, guru3d, pcper, among others

No for HardwareCanucks at least. Going from an average of 927 Mhz with the reference card to the 1.32 Ghz from a GTX 780 Classy (40% increase) there's a 24.5% increase in perf in Tomb Raider and 22.1% in Crysis 3.

 

[BallaTheFeared]

I'm not saying that you're lying. Again, I'm saying that your results are unrealistic, made up and misleading.

They're based on the same numbers you were saying weren't valid, and are now attempting to use to prove your own point. How can it be all that? It can't, you're just reaching randomly as you dig your hole deeper because you can't admit you're wrong.

The results from hardware.fr between stock and Uber for the GTX 780 is a 9.9% at 1080p while the average clocks are 876 Mhz for the stock and 993 for the Uber testing, 13.3% increase in average clockspeeds. So just removing the thermal threshold and adding a 6% power limit the card behaves way closer to your locked frequency. But hold on, you can pick the results from Sleeping Dogs for example with a 12% improvement between stock and Uber (15% faster average clock) or Splinter Cell Blacklist with a 7% (8.5% faster average clock). Just by that you can tell that GPU Boost is highly variable and you can't emulate it with a locked clockspeed.

This is the point of the discussion, using results you were attempting to discredit saying they fluctuated.

Sleeping Dogs, 12% increase for a 15% increase in core, 0% increase in bandwidth.

Splinter Cell Blacklist, 7% faster with an 8.5% increase in core, 0% increase in bandwidth.

Boost doesn't fluctuate the clock, I can run mine all day it will stick 1137MHz at stock without adjusting anything. The only reason boost would change is if it was a) used under vsync b)throttling due to heat/power.

By your own Tomb Raider figures max fps for the first bench is 27% higher than average and min fps 28% lower with a locked clockspeed. Showing that your GPU is way more or less efficient than the average in some parts of the benchmark. Depending on how GPU Boost behaves you should add or subtract an aditional X% based on other factors that your locked frequency doesn't take into account.

Tomb Raider had a 26% increase in core speed, coupled with a +400 offset on the memory.

Min increased by 29%

Max increased by 27%

Avg increased by 25%

Way more not found. GPU boost is fixed, I'm not sure what to tell you. The clock speed does not change +/- 13MHz during the entire test.

Your results are unrealistic because no card will behave like yours with different max and min figures.

Made up because you removed the GPU Boost functionality and misleading because you're trying to pass them as the same as a stock card averaging those clockspeeds.

Not sure how they're unrealistic when they're right in front of your face? How could they be anymore real than that?

Based on what exactly? Do you have a card to test yourself the back up your claims, or are they just claims with no proof?

This is with stock GHz bios, boost is not disabled.

1019/1500

OC

1176/1750

Core OC: 15%

Min FPS increase: 16%

Max FPS increase: 12%

Avg FPS increase: 15%

Any other theories you want me to investigate?

It's like you bought your card and rediscovered the wheel for everyone. If no one has done the same testing as you since the Titan launch is because all the reasons above.

It's like I'm an enthusiast end user discussing scaling in a thread with water cooling and 1.4v OC's and 1440 MHz 780s... What did you think it would be like? Stock reference?

Edit: OPS I had adaptive V-Sync on @ 74Hz, which is why the Max didn't scale up like it should have percent wise and why there is a slight dip in usage when it it is peaking the FPS as seen in Afterburner graph.

Do I need to retest, or are we ok with these boost enabled results?

 

[Imouto]

Look how easy is to refute all that text and screens: Yours isn't a reference card and isn't throttling like every single reference at stock does.

When you modify the default fan profile, the temp target and the power limit that's not reference stock anymore. You're doing the same as hardware.fr but trying to pass it as if you were emulating reference stock performance.

 

[BallaTheFeared]

Amazing isn't it? Simply, amazing! No throttle! No clock speed variance! Boost isn't magically changing clocks for no reason, other than because of what I said before!

No hardware.fr uses default, they just warm the cards up prior to testing. Which is why your silly 1019MHz comment was so pointless, as are any other "max boost" clocks gleaned from cold start tests when compared to warm run "avg boost" performance.

None of that makes a bit of difference in the context of this thread, nor does it make any sense when discussing the actual core scaling performance of GK110.

Where will the goal posts go next?

 

[Imouto]

You just have to google images for "gtx 780 clock speed".



That's what a clock speed graph looks like in a stock reference card. Something that you can't emulate with your 900Mhz locked custom card. Before you start deflecting with your V-Sync or cold starts take a look at the duration of the run.

No site has ever posted a 40% perf increase, let alone a 100% perf scaling, from stock reference. I can refer you to every single review out there and if you don't trust them you can ask Elfear in this very thread.

That's what ICDP and me are telling you all the time and you don't seem to understand. No moving goalposts, it's simply you doing it wrong.

 

[BallaTheFeared]

Problem for your "argument" is that I've never once stated a 40% increase over stock reference. That said, 900MHz fixed would be faster than 865 avg boost with wild momentary fluctuations between 900 and 840.

If that's the strawman argument you want to stick with I'll bite, 900MHz fixed is faster than the 860 avg boost results hw.fr got. That includes the 40% performance increase I obtained over said clock speed.

That said I've only ever been discussing core scaling with clock speed and it's dependency on memory bandwidth.

Also your .01 second clock speed changes are worthless from a performance perspective, especially when trying to discount the avg boost results of hw.fr.

More strawman, why does it have to be stock reference? When you were making up these fake rules you'd argue by, were you aware this thread was about a lightning 780 and 1.4v overvolted on water 290?

"No reviewer" No kidding, these aren't your grandma's overclocking results. However there is nothing "fake" or "unrealistic" about them, anyone with a 780 can get the same results as I do.

 

[Gikaseixas]

anyone with a 780 can get the same results as I do

Not sure if serious
Cards are all different, they don't reach the same limits, you should know that by now

 

[BallaTheFeared]

Not sure if serious
Cards are all different, they don't reach the same limits, you should know that by now

Considering his complaint was with the bios (disabled boost) I'll say context lost and move on.

Basically he's trying to push the idea that stock bios with a high core oc, no memory oc and throttling reduce scaling making my results 'unrealistic'. /golfclap

 

[Imouto]

Problem for your "argument" is that I've never once stated a 40% increase over stock reference. That said, 900MHz fixed would be faster than 865 avg boost with wild momentary fluctuations between 900 and 840.

If that's the strawman argument you want to stick with I'll bite, 900MHz fixed is faster than the 860 avg boost results hw.fr got. That includes the 40% performance increase I obtained over said clock speed.

Not really, Hardware.fr posted 65 fps in the Tomb Raider bench @1080p with an average clock of 915 Mhz. You posted 46.9 fps. Hardware.fr is 38% faster with 2% faster clocks. From that rock bottom result is easy to get some nice max performance results and scaling.

Also your .01 second clock speed changes are worthless from a performance perspective, especially when trying to discount the avg boost results of hw.fr

The fluctuation is constant between 880 Mhz and 990 Mhz. That's a serious wild variance.

More strawman, why does it have to be stock reference? When you were making up these fake rules you'd argue by, were you aware this thread was about a classified 780 and 1.4v overvolted on water 290?

You were the first that answered to Elfear referring to computerbase and then hw.fr both with stock reference cards that throttle. From then on you've been referring to those sites constantly.

 

[BallaTheFeared]

Not really, Hardware.fr posted 65 fps in the Tomb Raider bench @1080p with an average clock of 915 Mhz. You posted 46.9 fps. Hardware.fr is 38% faster with 2% faster clocks. From that rock bottom result is easy to get some nice max performance results and scaling.

Well at least most everyone should see the level of knowledge with which you're arguing at now.

The fluctuation is constant between 880 Mhz and 990 Mhz. That's a serious wild variance.

And the avg is probably 965MHz, any reason to believe it would be faster than 965 fixed? Or are we to believe those instant fluctuations with no flat bar going to the right next to them grossly affected the fps results?

You were the first that answered to Elfear referring to computerbase and then hw.fr both with stock reference cards that throttle. From then on you've been referring to those sites constantly.

What does that have to do with anything I said?

You can't just make things up to create a new strawman and goalpost for everyone to strive for, that's not how this works.

 

[BallaTheFeared]

Mirrored their settings for you so you wouldn't get so lost.

1019/1500

Wait for it...

Hows my scaling bro?

 

[Silverforce11]

My undervolted 290 is at 87C mining now.
Stock clocks, 70% fan and near opened window.

2 weeks ago, same settings, same ambient temp, it was sitting at 83C

How's that for wear & tear??
Not that it surprises me much.

It's called dust, took both mine out for a quick pressure air spray after nearly a month mining 24/7.. a LOT of dust build up when you run them that fast next to an open window (hi dust, come in and play!). Also when you have time, put some good TIM on it, the stock one is rubbish.

I'm running 70% fanspeed in CF in this Australian summer and its 88C mining.

ps. You guys are getting panties in a twist, its well known scaling varies a lot based on the games tested.

 

[ICDP]

Not sure if serious
Cards are all different, they don't reach the same limits, you should know that by now

Exactly, the issue here is Bella and Blackend are using confirmation bias to generalise all GTX780s as a guaranteed monster overclockers with almost perfect OC scaling.

 

[blackened23]

And you're definitely not biased towards the 290, right? You're definitely not trying to paint it in the best light, right? Gimme a break man. Confirmation bias. Three words for you. Pot kettle black.

There's nothing wrong with having a brand preference. I think it's fine. It isn't any different than a sports fan or car enthusiast making their pick. You can like whatever you like and that's great. But when you're sitting there telling me "CONFIRMATION BIAS" I can't help but laugh. Give me a break. Pot. Kettle. Black.

 

[Grooveriding]

lol, all of you have you have ruined his thread. What was actually a good OP reduced to rubble arguing about a few % of differences between two cards that pretty much perform indistinguishably from one another.

 

[chimaxi83]

lol, all of you have you have ruined his thread. What was actually a good OP reduced to rubble arguing about a few % of differences between two cards that pretty much perform indistinguishably from one another.

Are you surprised? Once a distinguished few enter your thread, kiss it goodbye.

 

[Imouto]

@Balla You're hanging yourself here.

With that scaling:

900 Mhz 64 FPS
915 Mhz 65 FPS (100% scaling)
993 Mhz 71 FPS (8% frequency increase) (9% performance increase) (110% scaling) No wonder with GPU Boost interaction minimized.
1019 Mhz 76.6 FPS (2% frequency increase) (8% performance increase) (400% scaling) Quite a jump huh?
1293 Mhz 96.4 FPS (26% frequency increase) (26% performance increase) (100% scaling)
900 Mhz

900 Mhz -> 1293 Mhz = 43.6% Frequency increase
64 FPS -> 96.4 FPS = 50% Performance increase
114% Scaling

So no, your numbers are off and you can't simulate the GPU Boost.

That going by your own data. TFSM knows what you did to get a 400% scaling going from the 993 Mhz from hw.fr to your 1019 Mhz result stock memory with both. I really hope you're not childish enough to ridge your own numbers.

So yeah, I prefer reviewers over you any day of the week. Your data is totally opposed to everything we've read about the GTX 780 to the date. Elfear made some good points in this thread with you nowhere near him, honestly.

Even more, Elfear made a nice graph and his Lightning didn't reach 100% scaling even once. A card that that is likely to never throttle at stock.

 

[BallaTheFeared]

You're probably right, I can't simulate hotbox thrashing and reported clock speeds averages which are fallible due to human error.

I can however directly compare reported clock speeds, with the same settings using the same version of the game, while using the same settings with fixed clock speeds that do not fluctuate.

No reason to make up numbers and FPS, I'll use my own.

915 Mhz -> 1293 MHz = 41.3% increase

70.9 FPS -> 96.4 FPS = 36% increase

My numbers are fine, and I can't simulate data that is not presented correctly. Clearly the "avg" wasn't reported correctly, it must have actually been lower. Or else something else is causing the difference, which can't be the game since we are using the same version, and it can't be the settings since they're pretty easy to follow.

Yet again you accuse me of lying...

I don't really care which you prefer, I'm only presenting data. It doesn't matter to me which way you want to go with it.

Do you know what clock speeds he used for his scaling graph? Weird isn't it? I came in saying vram matters for scaling, you completely ignored that instead creating a strawman which was fairly easy to dismantle with more actual facts and a lot less meaningless opinions and words based on nothing.

 

[Imouto]

Told you earlier, I'm not accusing you of lying, it's just that you're a hopelessly stubborn folk.

Also, no need to go apeshit. I luv you both.

Thankfully you don't write reviews. You'd end with an even worse reputation than that Alien guy.

 

[BallaTheFeared]

Told you earlier, I'm not accusing you of lying, it's just that you're a hopelessly stubborn folk.

I really hope you're not childish enough to ridge your own numbers.

What exactly would you call this then?

First you say my numbers don't add up, then you say I hope you aren't "ridging" my own numbers.

What else would you possibly be attempting insinuate without actually ever saying anything?

Also, no need to go apeshit. I luv you both.

I'm especially upset you can't accept my fps numbers and good scaling that goes in the face of your beliefs :hmm:

Thankfully you don't write reviews. You'd end with an even worse reputation than that Alien guy.

Yet another personal attack.

Good thing I don't write reviews because my numbers are lies and I'm cheating like that other guy clearly is!

I won't stoop to your level, but I believe I've proven what I set out to. You've been unable to formulate any counter of substance since this started, and at this point it's just basically you insulting people while adding nothing of substance to the discussion.