Gtx 1080 benchmarked

I am very skeptical. There are no numbers on frame rate or any performance numbers for the 1080 pics. Serious Salt Stance taken.
 
Lol this card is a joke if that is really its performance. It would only be around 50% faster than a 980 based on that which is laughable for going from 28nm to 16nm, using a better architecture, and running much higher speeds.
 
misterbobby said:
Lol this card is a joke if that is really its performance. It would only be around 50% faster than a 980 based on that which is laughable for going from 28nm to 16nm, using a better architecture, and running much higher speeds.
Click to expand...

If this is true, the scary part is the 1080 is 2560CC compared to a 980ti with 2816 GTX and only being 25% faster. Problem is we don't know if the 1800mhz on the 1080 is overclocked? Or normal?

Thats why I said earlier I would like to see a maxwell 2816CC vs a pascal 2816CC. Same clocks. I want to see if pascal IPC went down a bit or not. Really has me curious.
 
Last edited:
Bowman15 said:
This thread is full of feminazi's that don't like to speculate, have opninions or want others too. Why don't you all just put this thread on ignore FFS and move on.
Click to expand...


are-you-serious-rage-face.jpg
 
Not impressed at all, this is like what AMD has been releasing. But that's a win for me because that just means that I can use my card for another year.
 
Brackle said:
If this is true, the scary part is the 1080 is 2560CC compared to a 980ti with 2816 GTX and only being 25% faster. Problem is we don't know if the 1800mhz on the 1080 is overclocked? Or normal?

Thats why I said earlier I would like to see a maxwell 2816CC vs a pascal 2816CC. Same clocks. I want to see if pascal IPC went down a bit or not. Really has me curious.
Click to expand...

My guess is that Nvidia is getting every bit of clockspeed they can out of it to try and make up for the lack of IPC. So while this may not be overclocked, the 1080 may not have a lot of room (or no room at all) for overclocking. Even if Nvidia simply shrunk Maxwell to 16nm, they should be getting a lot more performance than that at that clockspeed on a 330mm die. A lot of that silicon has to be taken up by something like DP units.
 
Last edited:
  • Like
Reactions: N4CR
like this
I thought we were expecting 20% more than the 980 Ti?
Might be disappointing OC results based on these clock speeds.
 
TaintedSquirrel said:
I thought we were expecting 20% more than the 980 Ti?
Might be disappointing OC results based on these clock speeds.
Click to expand...
Close to the performance of an overclocked 980Ti while being overclocked itself isn't what I think most people expected.
 
variant said:
Close to the performance of an overclocked 980Ti while being overclocked itself isn't what I think most people expected.
Click to expand...

Most people were idiots. They do this EVERY DAMN GENERATION.

Remember when Maxwell was going to be the second coming of black baby Jesus?
 
The worst case scenario is if they release the 1080 at $599 or so and it's only 15% - 20% faster than a 980ti. I could potentially see this happening.
 
jwcalla said:
Just out of curiosity, how much faster do you think the 1080 should be over the 980?
Click to expand...
A reference 980 has 2048 cores that run at 1250. This card has 2560 cores running at 1800. That is technically 80% faster if it scaled perfectly still using the same architecture. Instead we are only getting about 50% better performance which would be piss poor even if still using Maxwell but even more ridiculous considering this is a supposedly better architecture.
 
Brackle said:
which is why im wondering if the IPC is now slower with Pascal.

Tinfoil hat!
Click to expand...


Its not, the kingpin overclocked still has boost clocks on the 980ti......

Yeah its ALU vs ALU clock for clock Pascal is around 20% higher with that Firestrike scores.

I don't understand why you guys can't do this math????? pretty easy to do, just multiplication and division.
 
razor1 said:
Its not, the kingpin overclocked still has boost clocks on the 980ti......

Yeah it ALU vs ALU clock for clock Pascal is around 20% higher with that Firestrike scores.

I don't understand why you guys can't do this math????? pretty easy to do, just multiplication and division.
Click to expand...

And how exactly do you know its not? You got the cards in your hand doing a review for us?

Seriously people who say they know stuff about these new cards when its all speculation is just mind boggling.
 
Brackle said:
And how exactly do you know its not? You got the cards in your hand doing a review for us?

Seriously people who say they know stuff about these new cards when its all speculation is just mind boggling.
Click to expand...


Do the math. I'm not speculating anymore, and if you want to ask that question, I suggest you look at my history.

PS I haven't been speculating for quite some time now either.
 
razor1 said:
Do the math. I'm not speculating anymore, and if you want to ask that question, I suggest you look at my history.

PS I haven't been speculating for quite some time now either.
Click to expand...

Well I can tell you now your history isn't helping at all.
 
What will you do if this come true? cause I can't get any more specific without divulging things I can't?

Will you apologize to me?

PS things that are already out in the open I can talk about, anything else, well shits and giggles.

That's why I stated do the math, its pretty straight forward.

Unfortunately people want to talk about something they didn't even look into to begin with, shit thats pretty obtuse don't you think?

Also I have sent pm's to people for the past 2 or 3 weeks and told them what the performance of the 1080 will be.
 
razor1 said:
What will you do if this come true? cause I can't get any more specific without divulging things I can't?

Will you apologize to me?

PS things that are already out in the open I can talk about, anything else, well shits and giggles.
Click to expand...

No I will totally admit I was wrong. Just like I was totally wrong when the 970 GTX was as fast as a 780ti for $329. I was wrong and admitted to it.

Which is why I own a 970 GTX (I had 2 before the memory fiasco!)

The reason I ask is, the 1080 with lower CC count @ 1800mhz is matching a 980ti @ 1500mhz. It you used the came CC card and put the frequencies of the cards at say 1500mhz I am curious if pascal has faster IPC.
 
Brackle said:
No I will totally admit I was wrong. Just like I was totally wrong when the 970 GTX was as fast as a 780ti for $329. I was wrong and admitted to it.

Which is why I own a 970 GTX (I had 2 before the memory fiasco!)

The reason I ask is, the 1080 with lower CC count @ 1800mhz is matching a 980ti @ 1500mhz. It you used the came CC card and put the frequencies of the cards at say 1500mhz I am curious if pascal has faster IPC.
Click to expand...


hmm what, where did ya see that, The 1080 was at 1800 (boost clock) another words ~1600 base clock maybe even less well......, the 980 ti with LN was at 1800
 
Brackle said:
It is from another forum I was reading. Here is the comparison

I scored P18 872 in 3DMark 11 Performance

Close to the same CPU Freq. and the 980ti was boosted to 1500mhz

Lemme find that user, he was on guru3d.com

here: Guru3D.com Forums - View Single Post - First NVIDIA GeForce GTX 1080 3DMark Benchmarks
Click to expand...
oh ok my apologizes, didn't see that, yeah I wouldn't use 3dmark 11 for this kind of test doesn't work too well for these cards. You will run into bottlenecks that will skew the results. Well not skewed the results are valid, just not what would be the best in what today's games or future apps will hold up to.
 
razor1 said:
oh ok my apologizes, didn't see that, yeah I wouldn't use 3dmark 11 for this kind of test doesn't work too well for these cards. You will run into bottlenecks that will skew the results. Well not skewed the results are valid, just not what would be the best in what today's games or future apps will hold up to.
Click to expand...

I do agree, but we have nothing else to compare the benchmark too. I actually hate 3dmark, but you can have a base to go off of.

For all we know 1080 GTX is a DX12 machine, which 3dmark doesn't show.
 
Brackle said:
I do agree, but we have nothing else to compare the benchmark too. I actually hate 3dmark, but you can have a base to go off of.

For all we know 1080 GTX is a DX12 machine, which 3dmark doesn't show.
Click to expand...
And probably 99% of games will not be true DX12 games for the next year or so. The DX12 crap we have now is mostly just tacked on features to market it as a DX12 game.
 
misterbobby said:
And probably 99% of games will not be true DX12 games for the next year or so. The DX12 crap we have now is mostly just tacked on features to market it as a DX12 game.
Click to expand...

I agree. Thankfully we have a good 2-3 titles coming out that are DX12.

Let's not forget the BF5 announcement tomorrow!!!!
 
Its likely Polaris 10 and gp104 have the same shader count by the way
 
Domingo said:
Hmmmm. If that holds true, that's a big 'ol "nope" from me. At this point, my next card needs to be a 1-card 4K solution. I've already got 1080p more or less covered with my OC'd 980 GTX.
Click to expand...

I wish but I don't think that's ever going to happen. By the time we can run current games at 4k at 60+ fps, there will be newer games with much higher requirements that will need SLI/X-Fire to run acceptably. We'll never get rid of 2+ card system :/

These new cards don't seem very impressive. Waiting on next gen :)
 
Personally I'm hoping the new cards can manage 1080p@120/144Hz. I don't really care much for 4K gaming, since good 4K gaming monitors are non-existent where I live, but good 1080p/1440p gaming monitors are aplenty.
 
Razor really has been saying performance will be like this for a while

If you're disappointed you should really take a moment of introspection and try and figure out why exactly that is, because your expectations are inconsistent with history

This is an amazing card. It's extremely impressive.

20% smaller than gm204

25-30% faster than gm200

2560 ALUs should put it comfortably at 9.5tflops
 
ShuttleLuv said:
Still relevant, just not as much as the old days.
Click to expand...

3D Mark hasn't been relevant since it stopped using actual game engines. That was a very long time ago. It's scores are arbitrary numbers that do not correlate with game performance in any way. The variables that tend to raise scores in 3D Mark aren't necessarily the same variables that actually improve real world game performance.
 
Dan_D said:
3D Mark hasn't been relevant since it stopped using actual game engines. That was a very long time ago. It's scores are arbitrary numbers that do not correlate with game performance in any way. The variables that tend to raise scores in 3D Mark aren't necessarily the same variables that actually improve real world game performance.
Click to expand...

It's certainly more accurate a performance metric than any one game...

I mean, given an arbitrary number of games being benchmarked 3DM is a better indicator of overall performance than any single game of the set, which is why it exists imo
 
Ieldra said:
It's certainly more accurate a performance metric than any one game...

I mean, given an arbitrary number of games being benchmarked 3DM is a better indicator of overall performance than any single game of the set, which is why it exists imo
Click to expand...

I'm sorry but that's just not the case. 3D Mark isn't an indicator of overall performance because its numbers do not translate into anything meaningful. You can't say a score of 18,000 3D Marks equals 120FPS at 3840x2160 in Crysis 3 or anything so specific. It doesn't reflect performance relative to game engines because it isn't using any game engine. 3D Mark performance can be improved by CPU upgrades or clock speed increases that wouldn't have the same effect in games. Going from 4.4GHz to 4.6GHz for example might give you a significant jump in 3D Mark scores but that increase won't translate into substantial FPS gains.

There are countless examples of this over the years. Throwing in a PhysX card used to add a significant jump in 3D Mark scores even though it did nothing in games that didn't utilize it. Adding more CPU cores (beyond 4 cores) would impact performance even though most games can't leverage more than 4 cores. Adding more clock speed when you are primarily GPU bound improves scores, and lets not forget one of the best examples of all time: AMD's Radeon 2900XT. It was a 3D Mark queen that scored well above that of its NVIDIA counterparts only to perform worse in virtually every single game out at that time or that had ever been made up to that point.

It isn't an accurate performance metric of overall performance because it's far too general. You can say a system that gets x number of 3D Marks is a fast system, but I can look at the system specs or device manager and tell you that. Neither of those things tells you very much. At best the only thing 3D Mark is good for is measuring system performance relative to other similar system configurations and taking a diagnostic view of things. That is to say, if x system is getting x 3D Marks, then another system with the same components should achieve similar results. There are many other tests that will do the same thing, so while 3D Mark is useful in this way, it's not any more useful than any other synthetic benchmark.
 
3DMark isn't useful for determining whether a card is fast enough to play any particular game. But that's ok because that's not its purpose. It's job is to be a benchmark of "relative" performance.

Back in the day 3dmark would also provide a glimpse of what's possible with the latest rendering APIs and algorithms and that was a very useful and exciting thing to have. Now the last few versions have just been an ugly, unimpressive mess with no value whatsoever. 3dmark Vantage was a joke and the others haven't done much better since.
 
Dan_D said:
I'm sorry but that's just not the case. 3D Mark isn't an indicator of overall performance because its numbers do not translate into anything meaningful. You can't say a score of 18,000 3D Marks equals 120FPS at 3840x2160 in Crysis 3 or anything so specific. It doesn't reflect performance relative to game engines because it isn't using any game engine. 3D Mark performance can be improved by CPU upgrades or clock speed increases that wouldn't have the same effect in games. Going from 4.4GHz to 4.6GHz for example might give you a significant jump in 3D Mark scores but that increase won't translate into substantial FPS gains.

There are countless examples of this over the years. Throwing in a PhysX card used to add a significant jump in 3D Mark scores even though it did nothing in games that didn't utilize it. Adding more CPU cores (beyond 4 cores) would impact performance even though most games can't leverage more than 4 cores. Adding more clock speed when you are primarily GPU bound improves scores, and lets not forget one of the best examples of all time: AMD's Radeon 2900XT. It was a 3D Mark queen that scored well above that of its NVIDIA counterparts only to perform worse in virtually every single game out at that time or that had ever been made up to that point.

It isn't an accurate performance metric of overall performance because it's far too general. You can say a system that gets x number of 3D Marks is a fast system, but I can look at the system specs or device manager and tell you that. Neither of those things tells you very much. At best the only thing 3D Mark is good for is measuring system performance relative to other similar system configurations and taking a diagnostic view of things. That is to say, if x system is getting x 3D Marks, then another system with the same components should achieve similar results. There are many other tests that will do the same thing, so while 3D Mark is useful in this way, it's not any more useful than any other synthetic benchmark.
Click to expand...
the entire point of it is to give a relative view of what performance should be like in comparison to other configurations, same thing as passmark. how you choose to use the results is up to you. i've used results from passmark in order to extrapolate what my performance would be like going from a 2500K to a 4770K in BF4, and they were highly accurate because i understood how to use them.

i prefer actual game benchmarks like anyone else, but 3dmark does have a purpose (you also seem to be ignoring the graphics score metric).
 
Last edited:
Dan_D said:
I'm sorry but that's just not the case. 3D Mark isn't an indicator of overall performance because its numbers do not translate into anything meaningful. You can't say a score of 18,000 3D Marks equals 120FPS at 3840x2160 in Crysis 3 or anything so specific. It doesn't reflect performance relative to game engines because it isn't using any game engine. 3D Mark performance can be improved by CPU upgrades or clock speed increases that wouldn't have the same effect in games. Going from 4.4GHz to 4.6GHz for example might give you a significant jump in 3D Mark scores but that increase won't translate into substantial FPS gains.

There are countless examples of this over the years. Throwing in a PhysX card used to add a significant jump in 3D Mark scores even though it did nothing in games that didn't utilize it. Adding more CPU cores (beyond 4 cores) would impact performance even though most games can't leverage more than 4 cores. Adding more clock speed when you are primarily GPU bound improves scores, and lets not forget one of the best examples of all time: AMD's Radeon 2900XT. It was a 3D Mark queen that scored well above that of its NVIDIA counterparts only to perform worse in virtually every single game out at that time or that had ever been made up to that point.

It isn't an accurate performance metric of overall performance because it's far too general. You can say a system that gets x number of 3D Marks is a fast system, but I can look at the system specs or device manager and tell you that. Neither of those things tells you very much. At best the only thing 3D Mark is good for is measuring system performance relative to other similar system configurations and taking a diagnostic view of things. That is to say, if x system is getting x 3D Marks, then another system with the same components should achieve similar results. There are many other tests that will do the same thing, so while 3D Mark is useful in this way, it's not any more useful than any other synthetic benchmark.
Click to expand...
That's the overall score you talking about, what I'm saying is say the graphics score of card X is higher than the graphics score of card Y by 20%

This indicator of relative performance is more accurate and more applicable than comparing the two in any one game

Like say you're comparing Fury X to 980ti in 100 games.

The graphics score in 3dmark will reflect the relative performance over the whole 100 games better than any one of them will
 
You must log in or register to reply here.
Back
Top