• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

NXGamer - The TFlops are a lie

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
Flops is actually a mensure metric... not a lie.
It is equal for any hardware since compute hardware was created.

Saying flops is a lie is like saying km/h or mph is a lie lol

That left me with the question if the owner of the video knows what flops is lol

Maybe he could start saying some hardware uses the flops power better than others... that is actually true because the complexity of the workloads in a hardware they canuse the full power of their flops all the time and so they trend to variate giving a performance different for each hardware.

But the flops continue there...

So if two hardware had 10TFs the one that most efficiently use that power is the better hardware.... in that case actually nVidia is killing in hardware efficiency.
 
Last edited:

DonF

Member
Jul 10, 2017
3,026
2,792
440
flops aren't everything, but could be used as a summary of computational power.

Now, if you plan to run a game @720p with a good upscaler vs a native 1080p game, the flop of both machine might not tell the whole story, since, to the naked eye, both games might look similar.
 
  • Like
Reactions: CatLady

Zefah

Gold Member
Jan 7, 2007
34,751
1,903
1,475
"Don't tell the whole story" may be accurate, but that doesn't mean they are a "lie" or useless as a measurement of raw horsepower.
 
  • Thoughtful
  • Like
Reactions: donfonzie and Genx3

o'dium

Embrace the snatch.
Oct 24, 2017
2,699
4,066
685
Telford, Shropshire
www.overdose-game.com
Flops are NOT a lie, however they only tell part of a story.

For example, two cars may go 120mph max speed. So in this example your flops are your speed. However, one car may do 0 to 60 faster, or another car may have better breaks.

The point is, that whole flops tells a vague picture, or doesn’t actually mean anything unless you can see everything on the whole.

In the case of the PS5 vs the SX, the latest rumours I saw had them both about the same tf, but the PS5 SSD was over twice as quick. That means a LOT when it comes down to it.

As always, masters can make a system with 1tf look just as good as a system with 10tf in the hands of amateurs.

The only thing for sure I know is that BOTH systems will have gorgeous looking games and will advance things further.

TLDR: Flops can give a good measure to what things are, but only tell part of the story.
 
Last edited:

clem84

Member
Dec 11, 2011
3,150
99
665
They're not a lie. It's like a global number. They give you a rough idea of how much is going on under the hood. It doesn't tell you anything about how those flops are distributed throughout the architecture of the chip, but they certainly aren't a lie.
 

TriSuit666

Member
Jul 2, 2018
110
64
210
He needs to back off whatever MIC set-up he has or just pull the levels down.

Also comparing the parts of the FH4 intro that are literally 4k movies.... wat?
 

01011001

Member
Dec 4, 2018
767
932
365
the video title is super clickbaity...

and especially for the new consoles the TF numbers will most likely directly correspond to their performance in multiplatform games.
we saw that on PS4 vs Xbox One comparisons, the TF difference almost exactly matched the performance difference.

why is that the case? because both platforms get their hardware from the exact same company at the exact same timeframe. none of them will let their systems have a bottleneck if they can avoid it.

so in the end I bet that we will see games run worse on the weaker platform by almost the exact difference their TF numbers would suggest, especially in these days where these systems are so close to eachother in terms of technology and design goals.
 
  • Thoughtful
Reactions: Godzilla Emu

NXGamer

Junior Member
Jan 7, 2016
496
137
290
Flops is actually a mensure metric... not a lie.
It is equal for any hardware since compute hardware was created.

Saying flops is a lie is like saying km/h or mph is a lie lol

That left me with the question if the owner of the video knows what flops is lol

Maybe he could start saying some hardware uses the flops power better than others... that is actually true because the complexity of the workloads in a hardware they canuse the full power of their flops all the time and so they trend to variate giving a performance different for each hardware.

But the flops continue there...

So if two hardware had 10TFs the one that most efficiently use that power is the better hardware.... in that case actually nVidia is killing in hardware efficiency.
Watch the video, i literally explain it within the first minute.

Your analogy to MPH/KPH is wrong, that is a Velocity metric, i.e. an external measurement only possible AFTER an action. such as "How Much can you bench, can only be measured AFTER you lift it". If something is traveling at X MPH/KPH then this is a fact, determined from an external measurement (Speed trap, GPS).
Saying something can travel at 190MPH for example only gives you a single level of metric, what about acceleration, Fuel consumption, handling, braking.

Tflops is a predictive or potential categorised metric that is used to describe a potential of something i.e. BHP is the better analogy as you say this bike has 225BHP it enables you to group it into a category of potential based on multiple factors (Fuel consumption, CC size, Top speed possible etc).

As this test proves, 2 machines with the same Tflop 'Metric" DO NOT peform at the same level.

2 Bikes traveling at 200MPH/KPH are equal, your example is completely out of place, you should have said cake.


EDIT: Reading through the thread it is clear that many posts did not even watch the video, come on guys. I love a discussion but at least listen to what I am saying. YES the title is a headliner grabber, of course I do not make a video for no-one to watch it (I have many already that tick that box). but it is all around the discussion.
 
Last edited:

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
Watch the video, i literally explain it within the first minute.

Your analogy to MPH/KPH is wrong, that is a Velocity metric, i.e. an external measurement only possible AFTER an action. such as "How Much can you bench, can only be measured AFTER you lift it". If something is traveling at X MPH/KPH then this is a fact, determined from an external measurement (Speed trap, GPS).
Saying something can travel at 190MPH for example only gives you a single level of metric, what about acceleration, Fuel consumption, handling, braking.

Tflops is a predictive or potential categorised metric that is used to describe a potential of something i.e. BHP is the better analogy as you say this bike has 225BHP it enables you to group it into a category of potential based on multiple factors (Fuel consumption, CC size, Top speed possible etc).

As this test proves, 2 machines with the same Tflop 'Metric" DO NOT peform at the same level.

2 Bikes traveling at 200MPH/KPH are equal, your example is completely out of place, you should have said cake.
You say my analogy is wrong just shows you don't understand the subject...

Two cars have max speeds at 180mph.
The two reach these max speeds the same way? Nope.
That makes these max speeds of 180mph a lie? Nope.

You can use Horse Power (HP) if you wish... it will be the same as analogy.

Your video titles is extreme out of place.
 
Last edited:

StormCell

Member
Dec 11, 2018
692
589
380
Pfff. Remember when millions of polygons were the big number being publicized?

And it used to be all about the GHz with processors.

FLOPS are a much better means of comparison, though, just like how in baseball WAR is a better metric for comparison than raw averages and counting numbers. In terms of real world value, how else are you to compare two computers designed to do basically the same thing if not by FLOPS?

Of course, there are other things that matter just like how cars are not always/only measured by their engines. For instance, I really like the Switch even though it's kinda gutless. That little gutless titan needs so little power that power outages have no impact on it--wish I could say that about my PS4! The PS4 also has a lot of features, though, that the Switch doesn't, such a the ability to play Spotify while gaming. I can jump in and out of video streaming without closing my game. The Switch is like a weird urban electric clown vehicle while the PS4 is a BMW, but I'll be damned if I don't appreciate both as marvels in gaming devices.
 
Last edited:

NXGamer

Junior Member
Jan 7, 2016
496
137
290
You say my analogy is wrong just shows you don't understand the subject...

Two cars have max speeds at 180mph.
The two reach these max speeds the same way? Nope.
That makes these max speeds of 180mph a lie? Nope.

You can use Horse Power (HP) if you wish... it will be the same as analogy.

Your video titles is extreme out of place.
read what I said, you are not getting it at all, I just explained it in detail....above.
 
Jan 29, 2019
1,676
1,131
355
2 Bikes traveling at 200MPH/KPH are equal, your example is completely out of place, you should have said cake.
I don't think he meant it in the way he should have, you can use max KM/h of two vehicles at a proxy for performance, then add weight (let's say a motorcycle V.S. a moving truck that both top at 200KM/h).
- The motorcycle will reach the speed faster
- The truck can move more stuff around and still maintain a decent speed

So, if you want to move only your GF you use the motorcycle to get home fast. but if you ever move in with her then you need the truck to move all those gadgets of yours.

As you said, speed doesn't work as an analogy to flops, it works as one for clock speeds, even that is stretching it.
 

StormCell

Member
Dec 11, 2018
692
589
380
Watch the video, i literally explain it within the first minute.

Your analogy to MPH/KPH is wrong, that is a Velocity metric, i.e. an external measurement only possible AFTER an action. such as "How Much can you bench, can only be measured AFTER you lift it". If something is traveling at X MPH/KPH then this is a fact, determined from an external measurement (Speed trap, GPS).
Saying something can travel at 190MPH for example only gives you a single level of metric, what about acceleration, Fuel consumption, handling, braking.

Tflops is a predictive or potential categorised metric that is used to describe a potential of something i.e. BHP is the better analogy as you say this bike has 225BHP it enables you to group it into a category of potential based on multiple factors (Fuel consumption, CC size, Top speed possible etc).

As this test proves, 2 machines with the same Tflop 'Metric" DO NOT peform at the same level.

2 Bikes traveling at 200MPH/KPH are equal, your example is completely out of place, you should have said cake.


EDIT: Reading through the thread it is clear that many posts did not even watch the video, come on guys. I love a discussion but at least listen to what I am saying. YES the title is a headliner grabber, of course I do not make a video for no-one to watch it (I have many already that tick that box). but it is all around the discussion.
I'll be sure and watch. I'm sorry the majority of people on these forums (GAF included) are numbskulls.
 
  • Love
  • Like
Reactions: Md Ray and NXGamer

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
read what I said, you are not getting it at all, I just explained it in detail....above.
I will explain again...

Flops is a metric and it is not a lie... it is a useful metric.
Now if you want to say two different hardware use it flops in different ways, or better for the analogy, two different hardwares reach that flops in different ways.
 
Last edited:

NXGamer

Junior Member
Jan 7, 2016
496
137
290
I don't think he meant it in the way he should have, you can use max KM/h of two vehicles at a proxy for performance, then add weight (let's say a motorcycle V.S. a moving truck that both top at 200KM/h).
- The motorcycle will reach the speed faster
- The truck can move more stuff around and still maintain a decent speed

So, if you want to move only your GF you use the motorcycle to get home fast. but if you ever move in with her then you need the truck to move all those gadgets of yours.

As you said, speed doesn't work as an analogy to flops, it works as one for clock speeds, even that is stretching it.
Absolutley bob on!
 
  • Like
Reactions: MiyazakiHatesKojima

NXGamer

Junior Member
Jan 7, 2016
496
137
290
I will explain again...

Flops is a metric and it is not a lie... it is a useful metric.
Now if you want to say two different hardware use it flops in different ways, or better for the analogy, two different hardwares reach that flops in different ways.

One hardware can be more efficient to reach these x TFs use than other but both can reach that TFs use.
If you had watched the video, you would have seen this is covered already, in detail and this discussion would have not been needed, but thank you as this is the point of the videos.
 

o'dium

Embrace the snatch.
Oct 24, 2017
2,699
4,066
685
Telford, Shropshire
www.overdose-game.com
I don’t usually agree with ethomaz, but... I do.

The thing is, you can’t lock down a consoles performance based on a single figure. You just can’t. It’s the same way a GPU can’t be locked down either. You may have a faster clock, but it may have a worse overall fillrate. Yet two gpus could both reach let’s say 10tf. They would both perform differently under different loads, and different stress tests, but would both be on paper “10tf”.

This is as simple an example I can give but the example can get much more complex when you start adding more and more different areas.
 
  • Like
Reactions: Bl@de

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
The video title just try to spread the bullshit claim that "AMD flops are different from nVidia flops".
Flops are the same for AMD and nVidia... it is a metric, it a fixed metric, it is not 1 for AMD and 2 for nVidia... it is equal for both.

How the hardware efficiently uses these flops, and that includes all the parts of the hardware, is what delivery the final performance.
 
  • Like
Reactions: Genx3 and CatLady

StormCell

Member
Dec 11, 2018
692
589
380
So why the false title?
Why not?

The cake is a lie...

Everyone is talking about the TFLOPS between PS5 and SX as though it could be the decisive detail between the two when that has almost never been the case. It's possible to build a better gaming machine with less TFLOPS, isn't it? Isn't it?
 

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
Why not?

The cake is a lie...

Everyone is talking about the TFLOPS between PS5 and SX as though it could be the decisive detail between the two when that has almost never been the case. It's possible to build a better gaming machine with less TFLOPS, isn't it? Isn't it?
First both are using the same hardware so the efficiency in both machines at least in GPU level will be pretty close.
Second the comparison is being made due the unconfirmed 9TFs PS5 vs 12TFs Xtower... that different is power is big to the point that even if the Xtower is a hell of inefficient to use its flops it will still have the advantage.

A 12TFs Xtower machine will delivery better graphical games than 9TFs PS5... PS5 can have the edge in art design in some games but in graphic render it will be below what MS have.

So why spread a lie in a video? Makes more uneducated gamers become more uneducated?
 
Last edited:

Jigsaah

Member
Jan 31, 2018
3,025
2,197
475
Tflops aren't a lie...they're just not 1:1 when it comes to comparing this gen to next because of all the other factors involved.

I said LONG TIME ago that Tflops wouldn't be used this time as a measurement because of this very reason.

I WAS SO WRONG cuz that's all people seem to be talking about. Hell...I even made a thread about it. Go figure that one out.
 
  • Like
Reactions: Godzilla Emu

StormCell

Member
Dec 11, 2018
692
589
380
First both are using the same hardware so the efficiency in both machines at least in GPU level will be pretty close.
Second the comparison is being made due the unconfirmed 9TFs PS5 vs 12TFs Xtower... that different is power is big to the point that even if the Xtower is a hell of inefficient to use its flops it will still have the advantage.

A 12TFs Xtower machine will delivery better graphical games than 9TFs PS5... PS5 can have the edge in art design in some games but in graphic render it will be below what MS have.

So why spread a lie in a video? Makes more uneducated gamers become more uneducated?
A 12TFS Xtower can deliver better graphical games. It's possible it never does, especially if multi-platform games reserve that power for 8K TV support or some such dumb thing. Or it'll simply have the better looking multi-plats with better textures and smoother frames.

But there are still areas in gaming that exist outside of pure raw graphical performance that could definitely use a big boost that only CPU can give such as a fully simulated game world that lives and breaths along with you as you play. We may be a very long way from something like this, but if Sony unveiled a processing unit that made it very easy to implement on PS5, you'd have to think that counts as a major chip.

Totally not saying they will, and I'm no Sony fan boy. I just believe there should be more to gaming than the damn graphics!
 

Mista

Dragonslayer
Nov 21, 2014
13,202
18,682
1,260
The title is dogshit but the discussion he made during the video is good. I like the lad but those titles needs to go away especially when you’re saying something else in the video😊
 

Grinchy

Member
Aug 3, 2010
23,238
7,250
1,115
a cave outside of Whoville.
I know jack shit about this stuff, so I'm wondering why the PS3 was billed as having 2 Tflops and neither the base PS4 or Xbone was that high, when clearly both systems shit on the PS3's ability to run games. Was it just that the CELL was that theoretically powerful?
 
Last edited:

Neofirezero

Member
Mar 25, 2019
140
97
200
You say my analogy is wrong just shows you don't understand the subject...

Two cars have max speeds at 180mph.
The two reach these max speeds the same way? Nope.
That makes these max speeds of 180mph a lie? Nope.

You can use Horse Power (HP) if you wish... it will be the same as analogy.

Your video titles is extreme out of place.
I agree with NXgsmers analogy but I also agree he shouldn't of put "put TPs are a lie" in the header. Something more accurate to the argument should of been used.
 

Kazekage1981

Member
Apr 7, 2019
197
224
230
Lets say PS5 had 13 TFLOPS and Xsex had 12.2 TFLOPS. Someone explained to me in the other thread that it would NOT be as substantial difference as Xbone and PS4. I guess it would be like comparing $12 million to $13 million dollars as oppose to $12 to $13? If that is the case, then other performance metrics should matter am I correct? Are we reaching the threshold of TFLOPS not mattering any more in the same way as polys/sec, #bits, and Ghz speed on the CPU?

I have also noticed the GPU speed mattering a lot. Does 1.7 Ghz GPU speed vs 2.0 Ghz speed make a difference in the same way as Xbone and PS4 TFLOPS mattered back in 2013?

I dont know if you explained it in the video, I didnt fully watch it.
 

ethomaz

Member
Mar 19, 2013
27,001
6,675
835
37
Brazil
Lets say PS5 had 13 TFLOPS and Xsex had 12.2 TFLOPS. Someone explained to me in the other thread that it would NOT be as substantial difference as Xbone and PS4. I guess it would be like comparing $12 million to $13 million dollars as oppose to $12 to $13? If that is the case, then other performance metrics should matter am I correct? Are we reaching the threshold of TFLOPS not mattering any more in the same way as polys/sec, #bits, and Ghz speed on the CPU?

I have also noticed the GPU speed mattering a lot. Does 1.7 Ghz GPU speed vs 2.0 Ghz speed make a difference in the same way as Xbone and PS4 TFLOPS mattered back in 2013?

I dont know if you explained it in the video, I didnt fully watch it.
GPU speeds alone not matter that much.
10 CUs at 2Ghz will delivery almost the same as 20 CUs at 1Ghz (the difference will be more due the rounding... there is no 1Ghz exactly neither 2Ghz exactly... it is always something like 999Mhz or 1001Mhz or so... the number is rounded to 1Ghz or 2Ghz).

The combination of numbers of units (SPs / CUs) and the clock speed of the GPU that matters.

It is different from CPUs because in GPUs there are non-parallel tasks that runs better in a core with 4Ghz than a core with 3.2Ghz (so some tasks will run better in a 2 cores @ 4Ghz than 4 cores @ 3Ghz)... on GPU all tasks are parallelized.

TFs still makes difference... you will give more flops to developers works with... so having 12TFs to develop a game is better than 9TFs... in the case of 12TFs and 13TFs it is just that the difference is not big enough to say it is big like PS4 to XB1... it is only 8% difference but that means that devs has 8% more power to delivery their games... so maybe 13TFs give you a flawless 60fps but 12TFs give you 58-60fps in the same GPU bound games.

Actually with ray-tracking we are reaching the threshold that GPUs needs to have more power (TFs).... maybe when we reach GPUs with 100TFs ray-tracking will be used 100% in the render.... I don't know when that will happen.

If 10% more TFs didn't matter MS should never have boosted the clock of XB1 before launch due Activision asking more power to make the CoD for their console... ~10% more TFs (1.19TFs to 1.31TFs) was enough to make Activision delivery the game.
 
Last edited:
  • Like
Reactions: Kazekage1981
Jan 29, 2019
1,676
1,131
355
Absolutley bob on!
One thing you should have done when overclocking the 750 ti would have been to only overclock the GPU, so that the TF value is better isolated, it kind of muddies the water.

- So overclock the GPU only
- Then put the GPU back to regular speed and overclock only the RAM
- Overclock both GPU and memory

That would be a lot of work, but it would show how each element affect the final result.

That is a lie because there are equal.... it is a fixed metric.
People should say something like: They don't manifest in the same manner in all architecture....

My assumption is that this is way too subtle for most people to understand the nuance.