PC Hardware Thread

[SRINI87]

They are no where near the 1.7X gain over the GTX980 that was being promised but I do not expect you to understand.

 

 

Here, this was the graph you had shown us and my quote underneath it,

 

So tell me how nVidia computed the 1.7X real-life performance gains?

 

Is 1.7X the standard frames you were getting on the GTX980 because if that were the case, it is not great anyway mathematically.

 

Is it 170% performance over the GTX980, which again is not being attained.

 

Draw of the day, nVidia marketing did a good job. People fell for it and now when the results are out, OMG! Such gains. Maybe you want to check out the performance deltas between Fermi and Maxwell just to get that this is nothing spectacular happening here.

 

At-least lie with conviction man, look at that wonderful three bar graph you put up there. :rofl:

Wtf seriously?

 

the graph mentions 1.7x times GTX 980 Not GTX 980TI?

 

1.7X means 70% faster than GTX980 not 170%

 

i think you did calculation differently

 

What actually graph says is that if GTX 980 GETs 100Fps AVG then GTX 1080 gets 170FPS.

 

is that hard to understand?

Edited May 18, 2016 by SRINI87

[ALPHA17]

i think you did calculation differently

 

Yeah! Try to parse marketing speech properly.

 

Even with the graph it shows, 66% performance increment over the GTX980. Where did I mention GTX980Ti in the above post?

 

And by your own admittance the most you are getting is ~40% increment in performance vis-á-vis a claim of ~70% that nVidia did. So is this hard to understand? Is it hard to see the reason behind why I am laughing?

 

Maybe it is because I am laughing at the collective folly of people putting up empty graphs with arbitrary, unsubstantiated performance gains and then backpedaling furiously when those claims came a cropper. :rofl:

I really want to try AMD cards but their cards run really hot as cooling is bad and driver support is also not good

 

Awwww... Such bad driver support that the last three drivers consistently have BSOD and overheating issues.

 

I have RMA'ed my GTX760 twice already and now after updating my drivers to the latest thinking that they have finally sorted their act, am again getting temperatures in excess of ~90°C. So maybe I should blame AMD for this.

Edited May 18, 2016 by ALPHA17

[SRINI87]

 

Yeah! Try to parse marketing speech properly.

 

Even with the graph it shows, 66% performance increment over the GTX980. Where did I mention GTX980Ti in the above post?

 

. So is this hard to understand? Is it hard to see the reason behind why I am laughing?

 

Maybe it is because I am laughing at the collective folly of people putting up empty graphs with arbitrary, unsubstantiated performance gains and then backpedaling furiously when those claims came a cropper. :rofl:

 

Awwww... Such bad driver support that the last three drivers consistently have BSOD and overheating issues.

 

I have RMA'ed my GTX760 twice already and now after updating my drivers to the latest thinking that they have finally sorted their act, am again getting temperatures in excess of ~90°C. So maybe I should blame AMD for this.

 

 

did you even see this Avg performance chart?

 

 

GTX 1080 ---100

GTX 980 ----59

 

(100-59)/59 = 0.6949

 

0.6949 x 100 = 69.49 = ~70%

 

i dunno from were you getting 40% from?

[ALPHA17]

GTX 1080 ---100

GTX 980 ----59

 

(100-59)/59 = 0.6949

 

0.6949 x 100 = 69.49 = ~70%

 

It is relative performance graph.

 

To the ~100% performance of at GTX1080, a GTX980 is providing 60% of the performance (relative). I dunno where you learnt to read graphs from?

[SRINI87]

Iska matlab GTX 980 gives 60% performance of GTX1080 on Average

That means GTX 1080 is 70% faster than GTX 980

 

Anandtech Quote for you: "In practice we’re looking at a 31% performance lead over GTX 980 Ti – the card the GTX 1080 essentially replaces – with a similar 32% lead over AMD’s Radeon R9 Fury X. Meanwhile against the slightly older GTX 980, that gap is 70%."

 

Tu rehne de bhai.

tere se nahi hoga.

[ALPHA17]

Iska matlab GTX 980 gives 60% performance of GTX1080 on Average

That means GTX 1080 is 70% faster than GTX 980

 

How?

 

60% of the performance equates to 60% of the performance, there is only a 40% ▲ in performance, not a 70% increase. 30fps vs. 50fps on the Anandtech benches, 100fps vs. 60fps on Witcher 3 @1080p. That is a 40% increment in performance.

 

Maybe you should go read the older reviews of GTX680 or GTX580 or so on to realise that the actual increase in performance is pretty much consistent throughout till now. No company is putting out 70% improvements in a single cycle, it is not in their own business interest.

[rdsarna]

I'm just going to be a Math arbitrator here....Just to clarify numbers

 

60fps vs 100fps is a 67% increase.

 

Here's the math-

100-60 = 40

40/60 x 100 = 66.67

 

40fps is 67% of 60fps. And 40fps is the difference between the two benches in Witcher 3.

[ALPHA17]

The math is not flawed. It is the interpretation, which I still disagree.

 

You might show it as 70% improvement in performance but that is not your gain. Your net gain is just 40fps IRL and that is 40% off 100 spin it anyway.

[quixote_1989]

 

sorry

[mohit9206]

This is precisely why i hated Maths in school so much...

[rdsarna]

The math is not flawed. It is the interpretation, which I still disagree.

 

You might show it as 70% improvement in performance but that is not your gain. Your net gain is just 40fps IRL and that is 40% off 100 spin it anyway.

Understood

 

 

This is precisely why i hated Maths in school so much...

[LaughingMan777]

The math is not flawed. It is the interpretation, which I still disagree.

 

You might show it as 70% improvement in performance but that is not your gain. Your net gain is just 40fps IRL and that is 40% off 100 spin it anyway.

 

I hate jumping into this and I shall regret it later... but 40fps additional over 60fps is not 40%. There is no way in real or imaginary life that equates to 40% "gain". And 60fps or oranges or bananas or whatever over 100 is a 40% "loss". You can't look at a horse's a*s and call it a "not much better than a mule".

[Big Boss]

I really want to try AMD cards but their cards run really hot as cooling is bad and driver support is also not good

It's actually reverse now in terms of drivers. Nvidia has been regularly releasing sh*t drivers. Mostly the drivers cause adverse effects to old GPUs.

 

Not sure about heat. My card never goes above 70-72 C on load. Idk if it's high or low compared to nvidia GPU.

[Big Boss]

The math is not flawed. It is the interpretation, which I still disagree.

 

You might show it as 70% improvement in performance but that is not your gain. Your net gain is just 40fps IRL and that is 40% off 100 spin it anyway.

Dude nothing is flawed. 100 fps vs 40 fps is 66.67% increase. So 1080 is 66.67% faster than 980 in that game because it's giving 66.67% increase in fps.

 

90 fps vs 60 fps is 50% increase.

You are getting confused with 100 number.

66 vs 60 fps is 10% increase.

 

84 fps vs 60 fps is 40% increase.

Edited May 18, 2016 by Big Boss

[ALPHA17]

I hate jumping into this and I shall regret it later... but 40fps additional over 60fps is not 40%. There is no way in real or imaginary life that equates to 40% "gain". And 60fps or oranges or bananas or whatever over 100 is a 40% "loss". You can't look at a horse's a*s and call it a "not much better than a mule".

Dude nothing is flawed. 100 fps vs 40 fps is 66.67% increase. So 1080 is 66.67% faster than 980 in that game because it's giving 66.67% increase in fps.

 

90 fps vs 60 fps is 50% increase.

You are getting confused with 100 number.

66 vs 60 fps is 10% increase.

 

84 fps vs 60 fps is 40% increase.

 

I got the mathematical logic applied to the situation but and this is a very important but, you do not count the performance increment of the GTX1080 by subtracting the gains you had from the performance of the previous generation card.

 

If you apply that logic, you can go to Fermi reviews and see the performance ▲ gained per-architecture change / die shrink process over older and connect the dots till this point.

 

Ambar the 100 number was from a legit benchmark in the Anandtech preview article (Witcher 3, All High 1080p, GTX980 -- 60fps and GTX1080 -- 100fps).

 

My point is you cannot count the performance ▲ by looking at the percentage point gained from last architecture. The extra 40fps are all you get, not 40/60 x100 ~percentage point gain.

 

There is a greater performance ▲ (percentage wise) in all 2560p resolution tests as well.

Now when it comes to DirectX 12.0 performance we see that nVidia drivers are still nowhere near their nadir and the R9 Fury cuts in close,

Does all this mean that the GTX*** series cards are bad, no. Just that people, especially the marketing guys will always want to use the number that is most amiable to their needs and it will be put up and in this case, it happens to be that 70% gain that has been trumpeted around.

[WR10]

I guess we won't be getting that 70% performance gain until Volta shows up.

[WhiteWolf]

What's Volta ?

[ALPHA17]

4. Alpha's math skills are horrible.

 

Cannot be good at everything. =P

What's Volta ?

 

Pascal's successor. Actually was to come after Maxwell but they pushed it up.

 

Either way, I do not see any company providing you a ~70% performance gain out of the box. Has not happened since the x86 transition to x86-64 days on processors and when GPU's went from 2D rendering to 3D rendering.

[LaughingMan777]

 

I got the mathematical logic applied to the situation but and this is a very important but, you do not count the performance increment of the GTX1080 by subtracting the gains you had from the performance of the previous generation card.

 

If you apply that logic, you can go to Fermi reviews and see the performance ▲ gained per-architecture change / die shrink process over older and connect the dots till this point.

 

Ambar the 100 number was from a legit benchmark in the Anandtech preview article (Witcher 3, All High 1080p, GTX980 -- 60fps and GTX1080 -- 100fps).

 

My point is you cannot count the performance ▲ by looking at the percentage point gained from last architecture. The extra 40fps are all you get, not 40/60 x100 ~percentage point gain.

 

There is a greater performance ▲ (percentage wise) in all 2560p resolution tests as well.

Now when it comes to DirectX 12.0 performance we see that nVidia drivers are still nowhere near their nadir and the R9 Fury cuts in close,

Does all this mean that the GTX*** series cards are bad, no. Just that people, especially the marketing guys will always want to use the number that is most amiable to their needs and it will be put up and in this case, it happens to be that 70% gain that has been trumpeted around.

 

The way I see it mate, either work with absolute numbers, in which case maths is maths. There's no two-ways to interpret absolute numbers. It's not qualitative.

 

Or we can stick to qualitative measurements, which essentially might mean anything beyond 60fps on a 60hz monitor is only marginal as far as game graphics are concerned to the average eye. sh*t graphics at 60fps will only look like slightly more glorious sh*t at 100fps. And sh*t graphics at 30fps will be super hideous sh*t. I for example seem to enjoy my console games more than my PC, but MP gameplay on PC is always better. Captain out :-P

[ALPHA17]

Just to cement my anti-nVidia credentials. This just happened.