Evo 4g/Evo 3D

[gTen]

Ok I found the video I think you spoke of:

http://www.youtube.com/watch?v=DM6sPTcFpAc

Its Quadrant..and here is what happened..there are 3 things to notice in this..

1) Galaxy S2 started out slower because it got jammed on the I/O

2) Now the main reason the score was higher on the SGS 2 was because of the opposite..for some reason the Sensation was running at 17-50fps averaging 30fps..while SGS2 was running at its cap of 60fps..hence why the score was higher..

3) Quadrant is a dead benchmark..its not dual core friendly..all you saw was signle core vs single core..this is probably why the sensation had such bad fps as well...

Edit:

they both have an sd card expandable to 32gb..hence I left it out...

As far as ram and performance..yes more ram heps..but do you honestly think these benchmarks that they use actually use over even 100mb ram? as far as "current benchmarks" its insignificant

[BlackDynamite]

Originally Posted by gTen

Ok I found the video I think you spoke of:

http://www.youtube.com/watch?v=DM6sPTcFpAc

Its Quadrant..and here is what happened..there are 3 things to notice in this..

1) Galaxy S2 started out slower because it got jammed on the I/O

2) Now the main reason the score was higher on the SGS 2 was because of the opposite..for some reason the Sensation was running at 17-50fps averaging 30fps..while SGS2 was running at its cap of 60fps..hence why the score was higher..

3) Quadrant is a dead benchmark..its not dual core friendly..all you saw was signle core vs single core..this is probably why the sensation had such bad fps as well...

Edit:

they both have an sd card expandable to 32gb..hence I left it out...

As far as ram and performance..yes more ram heps..but do you honestly think these benchmarks that they use actually use over even 100mb ram? as far as "current benchmarks" its insignificant

The benchmark itself may not be using all the RAM. But certainly all the stuff that is running in the background is using RAM.

The fps it is running at probably has about as much influence on these tests as the higher resolution the Evo 3D is running.

[gTen]

Originally Posted by BlackDynamite

The benchmark itself may not be using all the RAM. But certainly all the stuff that is running in the background is using RAM.

The fps it is running at probably has about as much influence on these tests as the higher resolution the Evo 3D is running.

I assure you those phones he was testing in had minimum running in the background..keep in mind that right now we are somehow living with half the ram in our phones..overall 1gb ram is obviously better then 768mb ram..but I'm sure we can both agree they had 0 effect on these benchmarks in particular..and both the SGS2 and Evo3d have 1gb ram..so its a moot point either way we slice the pie...

Yes, higher resolution would effect the gaming score..but productivity would remain unaffected...(sensation has same resolution as Evo3d of qHD)..

Either way the benchmarks on those videos are not very accurate due to Quadrant and even linpack they used both being flawed benchmarks..so far Smartbench and GL bench seem to be the best benchmarks I have seen.

I know GL bench does not account for resolution in calculations..not sure about smartbench..will have to research this..

[BlackDynamite]

Originally Posted by gTen

I assure you those phones he was testing in had minimum running in the background..keep in mind that right now we are somehow living with half the ram in our phones..overall 1gb ram is obviously better then 768mb ram..but I'm sure we can both agree they had 0 effect on these benchmarks in particular..and both the SGS2 and Evo3d have 1gb ram..so its a moot point either way we slice the pie...

Yes, higher resolution would effect the gaming score..but productivity would remain unaffected...(sensation has same resolution as Evo3d of qHD)..

Either way the benchmarks on those videos are not very accurate due to Quadrant and even linpack they used both being flawed benchmarks..so far Smartbench and GL bench seem to be the best benchmarks I have seen.

I know GL bench does not account for resolution in calculations..not sure about smartbench..will have to research this..

So wait, you can tell me how much RAM the new Sense utilizes on the Sensation?

I assure you, no matter how "minimal" the apps running in the background are, it is still better to have 25% more RAM. This is not some small, insignificant difference. Having 25% more RAM is a pretty major upgrade, and you absolutely WILL notice it in performance. I can't believe you are even disputing this.

It's a pretty reasonable expectation that the Evo 3D will score higher benchmarks than the Sensation.

[gTen]

Originally Posted by BlackDynamite

So wait, you can tell me how much RAM the new Sense utilizes on the Sensation?

wait..I thought you were implying sense was a good thing lol

I assure you, no matter how "minimal" the apps running in the background are, it is still better to have 25% more RAM. This is not some small, insignificant difference. Having 25% more RAM is a pretty major upgrade, and you absolutely WILL notice it in performance. I can't believe you are even disputing this.

It's a pretty reasonable expectation that the Evo 3D will score higher benchmarks than the Sensation.

It has 0 effect the ram usage..this is common sense stuff we are talking about..have you ever benchmarked a 512mb card vs a 1gb card? unless the game took more then 512mb ram the numbers were the same withing margin of error... I am not denying 25% is a major upgrade..but it has 0 effect on benchmarks in this case..and it has 0 effect as long as your not using more then 768mb ram...


Also..I have been looking around smartbench results on qHD Tegra 2 phones vs wvga Tegra 2 phones...and I dont see enough distinction of the gaming benchmark to imply resolution having an effect..so either smart bench accounts for resolution..or Tegra 2 renders at a higher resolution then rescales back to phone resolution.

[BlackDynamite]

Originally Posted by gTen

wait..I thought you were implying sense was a good thing lol

You thought wrong, lol. I run an AOSP ROM with SPB Shell 3D as my UI.

Originally Posted by gTen

It has 0 effect the ram usage..this is common sense stuff we are talking about..have you ever benchmarked a 512mb card vs a 1gb card? unless the game took more then 512mb ram the numbers were the same withing margin of error... I am not denying 25% is a major upgrade..but it has 0 effect on benchmarks in this case..and it has 0 effect as long as your not using more then 768mb ram...

Wait, are you talking about a video card with 512 MB of RAM versus a video card with 1GB of RAM? Because that is a totally different scenario. We are talking about a netbook with 512 MB of RAM total (in your example, but actually 768 MB in real life), versus a netbook with 1 GB of RAM total. And you can be sure you WILL see a pretty significant performance increase in the netbook that has 25% more RAM. It's not a matter of "the game" taking more than 512 MB of RAM. It's a matter of how much RAM is left for "the game" after Sense, Android, Sprint bloatware, HTC bloatware, and whatever else is running in the background. And even if there is still RAM left, that doesn't mean Sense won't run better with even more RAM. Maybe Sense sees there is not enough RAM to run everything so certain parts are automatically disabled, but it would leave them running (resulting in faster performance) if there was more RAM. Seriously, to say you won't see a performance increase on a stock ROM with 25% more RAM is not true at all.

Originally Posted by gTen

Also..I have been looking around smartbench results on qHD Tegra 2 phones vs wvga Tegra 2 phones...and I dont see enough distinction of the gaming benchmark to imply resolution having an effect..so either smart bench accounts for resolution..or Tegra 2 renders at a higher resolution then rescales back to phone resolution.

You can't say fps has a major affect on benchmarks and then later say resolution does not. That doesn't even make sense. Resolution is every bit as big of an impact on benchmarks as fps.

[gTen]

Originally Posted by BlackDynamite

Wait, are you talking about a video card with 512 MB of RAM versus a video card with 1GB of RAM? Because that is a totally different scenario. We are talking about a netbook with 512 MB of RAM total (in your example, but actually 768 MB in real life), versus a netbook with 1 GB of RAM total. And you can be sure you WILL see a pretty significant performance increase in the netbook that has 25% more RAM. It's not a matter of "the game" taking more than 512 MB of RAM. It's a matter of how much RAM is left for "the game" after Sense, Android, Sprint bloatware, HTC bloatware, and whatever else is running in the background. And even if there is still RAM left, that doesn't mean Sense won't run better with even more RAM. Maybe Sense sees there is not enough RAM to run everything so certain parts are automatically disabled, but it would leave them running (resulting in faster performance) if there was more RAM. Seriously, to say you won't see a performance increase on a stock ROM with 25% more RAM is not true at all.

The slowdown happens on a netbook when you go over your ram usage and start using pagepool, also pc ram is a bad example as if you have dual channel memory your better off having 2x512mb chips then a 512mb+256mb chip(this does not apply to cellphones yet though)..the sensation in the review could only have been an international model so no bloat-ware from manufacturer at the very least..how much ram free is irrelevant..this isn't SSDs..as long as there is enough ram to run the process thats more then enough..

Anyways on android when you run an app it allocates a certain amount of memory for that app..once the memory is allocated it belongs to that app and that app alone..so no matter how much memory is used outside the allocation is irrelevant..this isn't winmo..the only thing that has an effect is access to the ram and cpu cycles...both are unaffected by the amount of ram..

So yes..sense definitely effects the benchmark..but more or less ram is irrelevant unless the ram is faster or in dual channel...which I doubt is the case..

You can't say fps has a major affect on benchmarks and then later say resolution does not. That doesn't even make sense. Resolution is every bit as big of an impact on benchmarks as fps.

we were comparing 2 different benchmarks..I am nto changing my statement..I just said I looked at Smartbench results of gHD and wvga Tegra 2 phones as atm they are the only ones that come with qHD and Wvga variants..and I see no indication of it making a difference..hence either tegra 2 is rescaling or smartbench calculates ACCOUNTING to the resolution...

What you showed was Quadrant..different benchmark...each benchmark is unique in its calculation formula..I know 100% quadrant is effected by resolution..we are talking about smartbench here now..

[BlackDynamite]

Originally Posted by gTen

The slowdown happens on a netbook when you go over your ram usage and start using pagepool, also pc ram is a bad example as if you have dual channel memory your better off having 2x512mb chips then a 512mb+256mb chip(this does not apply to cellphones yet though)..the sensation in the review could only have been an international model so no bloat-ware from manufacturer at the very least..how much ram free is irrelevant..this isn't SSDs..as long as there is enough ram to run the process thats more then enough..

The Sensation "could have" been an international version with no bloatware? lol. Try again man. The correct way to say it is, "The Galaxy S 2 absolutely WAS an international version with no bloatware, and the Sensation may or may not have been." And your assesment of RAM is not correct. If there is enough RAM to run the process, sure, it will run. If the level of RAM drops below a certain level, other processes will not run. This is how HTC Sense works. Heck, that's how Android works in general.

Originally Posted by gTen

Anyways on android when you run an app it allocates a certain amount of memory for that app..once the memory is allocated it belongs to that app and that app alone..so no matter how much memory is used outside the allocation is irrelevant..this isn't winmo..the only thing that has an effect is access to the ram and cpu cycles...both are unaffected by the amount of ram..

That is not entirely accurate. If you run something else that takes up more RAM, the first app is terminated. HTC Sense is no different. HTC Sense could easily take up all the RAM on a phone if it did everything it is capable of doing at the same time. But it doesn't work like that. It uses only a percentage of what is available and dynamically reallocates more or less memory for certain functions. There is absolutely no debating that more RAM would lead to better bench scores on any HTC Sense device, especially the latest Sense version running on the Sensation, and even more so on the Sensation running the latest (and most bloated) version of sense, and also a bloated Stock ROM.

Originally Posted by gTen

So yes..sense definitely effects the benchmark..but more or less ram is irrelevant unless the ram is faster or in dual channel...which I doubt is the case..

Again, I disagree. If you have an Evo, you have probably ran ROMs with the new version of sense ported over. You saw how it affected performance before the cooks cleaned it up. Even though we never totally ran out of RAM on our Evo's (which are just as close to the sensation in RAM as the Sensation is to the Evo 3D). There is no doubt that any device running Sense will perform better with more RAM. And it's absolutely ridiculous to even try to debate this. Put a Sense 3 ROM on your Evo. Now put one on your Hero. Tell me there is no difference.

Originally Posted by gTen

we were comparing 2 different benchmarks..I am nto changing my statement..I just said I looked at Smartbench results of gHD and wvga Tegra 2 phones as atm they are the only ones that come with qHD and Wvga variants..and I see no indication of it making a difference..hence either tegra 2 is rescaling or smartbench calculates ACCOUNTING to the resolution...

No idea what any of that has to do with an Evo 3D. You are speculating that the Evo 3D's resolution may or may not be factored into a benchmark result based on a test ran on 2 seperate and totally unrelated phones, that have different specs from each other and neither has the same specs as the Evo 3D.

Originally Posted by gTen

What you showed was Quadrant..different benchmark...each benchmark is unique in its calculation formula..I know 100% quadrant is effected by resolution..we are talking about smartbench here now..

Any test you run, anything you do, will run faster at a lower resolution. That's why you turn down the resolution on games on older computers. It's common sense- can I render 10 pixels faster or 1 pixel faster?

[gTen]

Originally Posted by BlackDynamite

The Sensation "could have" been an international version with no bloatware? lol. Try again man. The correct way to say it is, "The Galaxy S 2 absolutely WAS an international version with no bloatware, and the Sensation may or may not have been." And your assesment of RAM is not correct. If there is enough RAM to run the process, sure, it will run. If the level of RAM drops below a certain level, other processes will not run. This is how HTC Sense works. Heck, that's how Android works in general.

Typo on my part..I meant to say no bloatware from the carrier..not manufacturer..hence my alluding to it not being t-mobile..

That is not entirely accurate. If you run something else that takes up more RAM, the first app is terminated. HTC Sense is no different. HTC Sense could easily take up all the RAM on a phone if it did everything it is capable of doing at the same time. But it doesn't work like that. It uses only a percentage of what is available and dynamically reallocates more or less memory for certain functions. There is absolutely no debating that more RAM would lead to better bench scores on any HTC Sense device, especially the latest Sense version running on the Sensation, and even more so on the Sensation running the latest (and most bloated) version of sense, and also a bloated Stock ROM.

It wont terminate a process thats running front end...in the background yes..not on your screen..that gets full allocation attention..both the evo 3d and sensation have new sense...there will be no difference between benchmarks of evo 3d and sensation..this has been proven by HTC Incredible and the Evo having virtual the same benchmarks...

Again, I disagree. If you have an Evo, you have probably ran ROMs with the new version of sense ported over. You saw how it affected performance before the cooks cleaned it up. Even though we never totally ran out of RAM on our Evo's (which are just as close to the sensation in RAM as the Sensation is to the Evo 3D). There is no doubt that any device running Sense will perform better with more RAM. And it's absolutely ridiculous to even try to debate this. Put a Sense 3 ROM on your Evo. Now put one on your Hero. Tell me there is no difference.

Um...hero is a REALLY bad example..its not even a cortex A8 phone...again sense does effect performance..but the effect is equal throughout both phones..on a benchmark that only takes up 1.3mb it has no effect...when comparing sensation to the Evo 3d..

No idea what any of that has to do with an Evo 3D. You are speculating that the Evo 3D's resolution may or may not be factored into a benchmark result based on a test ran on 2 seperate and totally unrelated phones, that have different specs from each other and neither has the same specs as the Evo 3D.

I am basing this on available data..I am not putting this up as a FACT..mearly an educated guess...

Any test you run, anything you do, will run faster at a lower resolution. That's why you turn down the resolution on games on older computers. It's common sense- can I render 10 pixels faster or 1 pixel faster?

This is normally the case yes..but as I said there are 2 senarios that are also possible:

1) Smartbench takes into account resolution when doing its calculation

or

2) Tegra 2 really renders at a higher resolution then qHD then rescales it down to a lower resolution.

There are more to this then even resolution..there is also dithering for example...Tegra 2 and previous snapdragon dithered at 16bit, while hummingbird dithered at 24bit..

Also keep in mind that the Gaming score was calculated on the SGS2 while being limited to 60fps...

And just so we are clear, your not disputing the Productivity score? correct?

[BlackDynamite]

[QUOTE=gTen;2112395]

Originally Posted by gTen

It wont terminate a process thats running front end...in the background yes..not on your screen..that gets full allocation attention..both the evo 3d and sensation have new sense...there will be no difference between benchmarks of evo 3d and sensation..this has been proven by HTC Incredible and the Evo having virtual the same benchmarks...

The Thunderbolt benches higher than the Evo. And there will absolutely be a difference in the Sensation and Evo 3D. It's unbelievable that you would try to dispute this.

Originally Posted by gTen

Um...hero is a REALLY bad example..its not even a cortex A8 phone...again sense does effect performance..but the effect is equal throughout both phones..on a benchmark that only takes up 1.3mb it has no effect...when comparing sensation to the Evo 3d..

You keep missing the point. That 1.3 MB test is NOT the only process running. Run that test on a Sense phone that only has 5 MB of RAM. Now run that same 1.3 MB test on a Sense phone that has 512 MB of RAM. Try to explain to me why the results would be even remotely similar. They wouldn't be. And a phone with 25% more RAM will run it faster too.

Originally Posted by gTen

I am basing this on available data..I am not putting this up as a FACT..mearly an educated guess...

And my educated guess is none of those tests mean anything yet. For one thing, like I posted, I saw a test that actually scored the Galaxy S2 higher even though it took longer to run the test. And for another thing, the Evo 3D will absolutely bench better than the Sensation anyway.

Originally Posted by gTen

This is normally the case yes..but as I said there are 2 senarios that are also possible:

1) Smartbench takes into account resolution when doing its calculation

or

2) Tegra 2 really renders at a higher resolution then qHD then rescales it down to a lower resolution.

There is also a 3rd option- the phones had other differing factors that influenced the tests. You act like just because they both had a Tegra 2 that everything else was identical with the only difference being the resolution.

Originally Posted by gTen

There are more to this then even resolution..there is also dithering for example...Tegra 2 and previous snapdragon dithered at 16bit, while hummingbird dithered at 24bit..

Also keep in mind that the Gaming score was calculated on the SGS2 while being limited to 60fps...

Again- resolution is every bit as big a factor as fps, and dithering. We kNOW the Sensation is running at a significantly higher resolution. This is not an educated guess, but a documented fact. So common sense tells us that you will have less fps at qhd than you will at wvga.

Originally Posted by gTen

And just so we are clear, your not disputing the Productivity score? correct?

Yes I am disputing all of those tests at this point. Unless you have 2 identical phones, then the tests lose value for each difference. Like I said, common sense tells us that qhd rendering will have a lower fps than wvga rendering. And if the OS and test aren't utilizing the dual cores, then it loses even more value. And even then, the Evo 3D was not even the phone in the test. It was a Sensation that has lower specs, and likely a worse bench than the Evo 3D. When these phones are released I'm sure we'll start to see some tests that more accurately put them against each other. But when I see a phone finish the test significantly faster, yet score significantly worse, I am not going to trust that test.