Evo blowing past the normal 30 FPS!

[jmxp69]

YouTube - Evo Godmode

This is a hacked kernel, Fresh 0.3, and SetCPU running @ 1.19ghz. The cameras don't work, but everything else is stupid fast and stable.

This is what the EVO WILL BE when XDA Devs are done. It's really too bad HTC released such a gimped device; fortunately, the opensource developers are all over it and they're making a measurable difference.

Imagine what these folks will do when the Froyo source is released.

 

[cabbie]

i hate to be a debbie downer because i love this phone...but the more 'real world' testing they did only showed around 14 and then 28 FPS on the last two demos running those graphics intensive tests. Have those been run on the DroidX and what does it score? 50 FPS on a moving ball is...just not quite what I had hoped.

 

[reemas]

how did you install this kernel?

 

[cabbie]

powerdrill, like everyone else does, sheesh

 

[MinimalistChris]

Sorry, but a bit off topic...

What kind of case is that in the video? Awesome vid though!

 

[droidSPARX]

on the vid its 14.2fps on nena.
the benchmark of droidx,nexus (froyo), and evo4: 22.9fps, 11.3, 10.3

 

[habitual.line.stepper]

What kind of case is that in the video?

It's a Chinese case of some sort, i found them at one of those cell phone kiosks they have in the mall today for 14.99. It's a nice case aesthetically, but it doesnt really offer up protection if you were to drop the phone

 

[jmxp69]

i hate to be a debbie downer because i love this phone...but the more 'real world' testing they did only showed around 14 and then 28 FPS on the last two demos running those graphics intensive tests. Have those been run on the DroidX and what does it score? 50 FPS on a moving ball is...just not quite what I had hoped.

Keep in mind, the kernel enabling these speeds is hacked. It's really more a proof of concept at this point to show the EVO can exceed the 30fps limit imposed by HTC.

With the exception of Linpack running on the Nexus one with Froyo, the EVO and N1 are turning in very similar speeds. The Droid-X does really well with its GPU, but not as good on Linpack.

The developers are hard at work trying to find the root of the FPS cap. When the kernel source comes out, and when they figure out what's really causing the cap, stand-by.

J

PS For those interested in the case: Search "htc evo red rubberized" on eBay and you should find it for ~ $6 shipped.

 

[jmxp69]

on the vid its 14.2fps on nena.
the benchmark of droidx,nexus (froyo), and evo4: 22.9fps, 11.3, 10.3

Edit: Yup. That's right. My EVO outperformed the N1 running Froyo on that test.

 

[jmxp69]

how did you install this kernel?

[WIP] Higher framerate on Novatek panels | Kernel inside - xda-developers

Use it at your own risk. You can *severely* mess up your phone if you're not comfortable with the tools required to do this.

 

[White Sharingan]

Damn....damn HTC :\

 

[jmxp69]

FWIW: This hack has been a little tough on the battery. I dropped 30% overnight with radios off. I'll do a little tweaking with cpu freqs to see if I can get it more manageable.

 

[HuDuVuDu]

It's a Chinese case of some sort, i found them at one of those cell phone kiosks they have in the mall today for 14.99. It's a nice case aesthetically, but it doesnt really offer up protection if you were to drop the phone

I have the same case from 3gaccessory.com (via ebay) ordered two one for me one for the wife, and it cost ~$10 after shipping for both.

 

[KAuss]

What would the FPS be if you ran the thing at stock clock? I OC my computer religiously but I am not on the same page when it comes to my phone... Nor would the phone have the battery capacity to make it worth while...

 

[RiverOfIce]

Shame on htc?

I was swiping my evo when he was doing his, guess what, no difference.

There is zero stall in the changing screens. People just love to complain.

Oh the last two test where priceless.

28fps, oh my god htc is horrible for capping it at 30.

 

[jmxp69]

What would the FPS be if you ran the thing at stock clock? I OC my computer religiously but I am not on the same page when it comes to my phone... Nor would the phone have the battery capacity to make it worth while...

Linpack drops to 6.7MFLOPS at stock clock.
No change in gears (expected).
Nena and Neocore will experience the larger differences due to 3d.
Neocore: 26.6
Nena: 10

Shame on htc?

I was swiping my evo when he was doing his, guess what, no difference.
There is zero stall in the changing screens. People just love to complain.

There is no "speed" difference in swiping the screen, it's how the content on the screen refreshes. At 50fps, there is a very visible difference in the smoothness of motion. I have 5 different ROMs stored on SD including Froyo v8 which gets 39MFLOPS and 30fps, and it's still not as smooth as this kernel.

Oh the last two test where priceless. 28fps, oh my god htc is horrible for capping it at 30.

The last two were 3D tests. Those are the fps you can expect in an OpenGL game and are far different than what you experience in 2D in the "desktop". fps2d shows a 70% improvement in 2d frame rates. If you think that's a petty gripe, fine. I do not. This phone is gimped, and it's non-trivial. I am flashing a stock kernel ATM and will post the 2d and 3d results with nothing changed but the kernel.

J

 

[reemas]

Shame on htc?

I was swiping my evo when he was doing his, guess what, no difference.

There is zero stall in the changing screens. People just love to complain.

Oh the last two test where priceless.

28fps, oh my god htc is horrible for capping it at 30.

You sound like you have no idea what you are talking about.

 

[jmxp69]

Ok, I installed a standard Fresh 0.3 ROM and here are the results:

          Hacked       Stock
[B]2D[/B]
Gears     50fps        30fps/3.64 70% improvement
fps2d     50fps        30fps/7.76 70% improvement

3D
Nena      14.2         10.2       40% improvement over stock
Neo       27.3         24.9       10% improvement over stock

CPU
Linpack   8.2M/F       6.5M/F     26% improvement over stock** 

**due to overclocking, not FPS hack.

Average improvement across all tests: 43.2%
Average improvement ignoring Gears or fps2d: 36.5%

Yes, HTC should be ashamed. Why? Because a phone they made which is already a year old beats the EVO running virtually the same hardware. They have intentionally mitigated the end user experience, and regardless if fps is perceptible to you, it is very real to those who know what they're looking for.

 

[Droidone]

If htc lifts the 30fps, why would you buy the next htc phone. They hold back to make sure you keep coming back for more IMO. I know they said it's because of battery saving they decided to do that, but I think that's a bunch of crap. There was a behind the scenes reason why they stopped at 30fps. Then they said it was a hardware limitation but yet this kernel allows this to work. Come one HTC, stop playing us for dumb.

 

[jmxp69]

After re-flashing the non-hacked kernel, I can really feel the difference now. If you're not sure what all the fps thing is about, the best observation I can make is this:

With the stock kernel, when I swipe or scroll, I can see vertical distortion as I look at the line of pixels on a vertical edge. As the desktop image moves, the re-paint makes the image blur slightly. With the hacked kernel, the screen is fluid, like liquid. There is absolutely no vertical distortion and the image remains crystal clear during all movements.

Stock        Hacked
|               |
|               | 
|               |
 |              |
 |              |
 |              |

I know this seems trivial, but the thing I think so many are upset about is that the hardware clearly supports the higher fps rate and the developers at XDA were able to make a fairly minor change to the board files to prove this out. It's hard to comprehend why HTC would intentionally inhibit performance. And they certainly didn't advertise on the box: "30fps video performance cap"

BTW: If you haven't seen the androidcentral.com comparison between the DroidX, N1, and EVO, go watch. Our phone performs the worst out of all of them in a big way. YouTube - Droid X benchmark tests

 

[White Sharingan]

If htc lifts the 30fps, why would you buy the next htc phone. They hold back to make sure you keep coming back for more IMO. I know they said it's because of battery saving they decided to do that, but I think that's a bunch of crap. There was a behind the scenes reason why they stopped at 30fps. Then they said it was a hardware limitation but yet this kernel allows this to work. Come one HTC, stop playing us for dumb.

I don't know, I would think that running at 60fps the battery would drain much quicker, and I THINK they weren't lying in that statement (correct me if im wrong, I would love to know for sure, maybe jmxp69 can clear this up). But I wish HTC would at least allow us to choose if we want to stay at 30fps or if we want to go up to "60fps", you know like have a "Standard Performance" and a "Higher Performance" as options.

 

[Rob_A]

I wonder if stock was 50fps, someone would create a rom that capped the fps to 30 to increase battery life, lol

 

[jmxp69]

During the kernel hack testing, the devs found that with the screen off, the fps rate skyrocketed which is why they're focusing on the panel boardfiles. That said, it's unlikely a battery related issue because if it were, they would have prohibited the GPU from generating anything above 30fps.

An XDA member even has an e-mail from a supposed HTC Exec by the name of Keith Nowak who dispelled the battery myth. Source: Reply from Keith Nowak about the 30FPS issue - xda-developers;however, he maintains there is a link between fps capabilities of the panels and HDMI:

"The HDMI output that delivers HD content is, as you would expect, very resource intense, and as such, puts a limit on the output that can be used for the main display."

WRONG

Further, he goes on to say,

"We have seen reports of a hack that purports to “fix” this, but after investigation, it turns out that this hack is simply allowing data to be pushed through the display interface at a higher speed than it can reasonably handle – resulting in, as you would expect, a choppy, unacceptable image of very poor quality."

WRONG

Obviously, he hasn't seen my video....

At the moment, I have no reason to believe anything HTC says on the matter until somebody comes clean, offers the kernel source, and lets the developers sort it out.

WRT my personal experience with the battery, it's no worse than stock so far. I had some troubles the first night dropping 30%, but I solved that by lowering the min/max to stock with SetCPU. I got much better battery performance with flipz' ROM because he really goes out of his way to optimize battery.

So, short answer, I don't believe battery has anything to do with it.

 

[RiverOfIce]

I give up, time and again this has been proven a non issue. There is not a thing that 30 fps would do better then 50. Nothing.

Complaining about this is like saying, car x max speed is 190 mph car y 200 mph.

The average user is not going to go past 190mph, a lot of video games on xbox and ps3 do not go pass 30fps.

Now 200mph is really nice, but for most people 100mph is ok.

I do not expect to very play video games on a phone. Just dont. The evo is insane fast when it comes to actually using the phone.

But of course I do not know what I am talking about. Of course.

So lets see what the experts in phones say. Let us go to one of the best reviewers of phone out there.

Phone scoop.

YouTube - HTC EVO 4G Review

You can start the software at 2:08.

Were he start the software section by saying that it is fast.

So if you want to go 200mph, when the 190 is ok by most people.

I swear you can not try around here.

 

[drksilenc]

roi this is prbly the only thing i will not agree with you on. most games dont go beyond 30 fps please tell me your joking the video game industry standard for all games is 60 fps. you cant use motion blur in video games that constantly change. im a high end pc gamer and there is a massive difference between 30 and 60 because on is actually pushing 60 the other is a scaled 30