Database Info

HTC Performance cab??

I searched for it and cant find it. I also looked at duttys cabs and saw nothing in there. Anyone know where this cab is??

i havent tried this yey but here it is

Thanks!! Where did you get it?!

iceman4357 said:
Thanks!! Where did you get it?!
Click to expand...
Click to collapse
i found it a few months ago in one of the kaiser overclock threads

What does it do?
Can you post the link to the original thread?
Thanks.

it is from another forum site... i cant remember the name. LCS just implented this into his new rom. I have it installed on the new sleuth wm6.1 rom and i personally cant see a difference. not only that, but... I ran a benchmark today and also saw no difference. this cab is about as helpful as the htc_d3dm.dll
its designed for a different processor... all in all, its a placebo. unless im somehow not activating it right. but idk, try it out for yourself... it doesnt harm the device and is pretty self explanatory how to activate it as far i know...
hope that helps

Thanks
I searched for this cab in the forums but couldn't find it (even using the google search option).
If this is some overclocking function, then I won't try it.

laurentius has the benchmarks with this activated at 520 mhz. it does make a bit of difference. here is the link and the benchmarks are in the 2nd post http://forum.xda-developers.com/showthread.php?t=366014

lildogg said:
laurentius has the benchmarks with this activated at 520 mhz. it does make a bit of difference. here is the link and the benchmarks are in the 2nd post http://forum.xda-developers.com/showthread.php?t=366014
Click to expand...
Click to collapse
His benchmarks prove there is nothing special about this, as there is less than 2% difference on all benchmarks which can be attributed to statistical margin of error.
If you don't believe in margin of error, then his benchmarks tell you that HTC performance makes things WORSE! The numbers below were extracted from his thread, with the before on the left, after on the right, and the bolded stuff being the loser.
242.1398 vs 241.7512
5.479 vs 5.477
494 vs 490
736 vs 728
11614.37 vs 11770.11
12880.50 vs 13044.59
603.65 vs 601.82
9337.39 vs 9266.97
1505 vs 806
2573 vs 2596
2561 vs 5884 (lower is better)
Out of 11 tests, 8 of those tests scored higher WITHOUT HTC Performance enabled.

_Alex_ said:
His benchmarks prove there is nothing special about this, as there is less than 2% difference on all benchmarks which can be attributed to statistical margin of error.
If you don't believe in margin of error, then his benchmarks tell you that HTC performance makes things WORSE! The numbers below were extracted from his thread, with the before on the left, after on the right, and the bolded stuff being the loser.
242.1398 vs 241.7512
5.479 vs 5.477
494 vs 490
736 vs 728
11614.37 vs 11770.11
12880.50 vs 13044.59
603.65 vs 601.82
9337.39 vs 9266.97
1505 vs 806
2573 vs 2596
2561 vs 5884 (lower is better)
Out of 11 tests, 8 of those tests scored higher WITHOUT HTC Performance enabled.
Click to expand...
Click to collapse
glad to hear that im not crazy, i can take my flame proof jacket off now lol

I guess this has not been run on a Stock rom?
Any chance it could perfrom better? Was it only CPU set at 520 vs 400 mhZ?

ceevee369 said:
I guess this has not been run on a Stock rom?
Any chance it could perfrom better? Was it only CPU set at 520 vs 400 mhZ?
Click to expand...
Click to collapse
If you have nothing better to do, you are welcome to try. It won't make a bit of difference if its run on WM6, WM6.1 or WM7.0, shipped or cooked. HTC performance doesn't work on Kaisers (or Apache, or a bunch of other devices too)...

_Alex_ said:
If you have nothing better to do, you are welcome to try. It won't make a bit of difference if its run on WM6, WM6.1 or WM7.0, shipped or cooked. HTC performance doesn't work on Kaisers (or Apache, or a bunch of other devices too)...
Click to expand...
Click to collapse
Thanks for the advise. I have other more important things to do. getting us all better video drivers for example
So, high performance Cab goes down the drain right now

one thing u could do is install custel's mobile crack file.. it really speeds slueth's 6.1 build up...like faster then dutty stuff

jakeypoo said:
one thing u could do is install custel's mobile crack file.. it really speeds slueth's 6.1 build up...like faster then dutty stuff
Click to expand...
Click to collapse
Custel's mobile crack is the same as using KaiserTweak and changing the three cache settings.
The name 'mobile crack' is easier to remember than 'file system cache', 'file system filter cache' and 'gdi/font cache'.

Hi _Alex_,
Maybe you're right but I do notice a nice speed improvement when activated!
Cheers,
Leo
_Alex_ said:
His benchmarks prove there is nothing special about this, as there is less than 2% difference on all benchmarks which can be attributed to statistical margin of error.
If you don't believe in margin of error, then his benchmarks tell you that HTC performance makes things WORSE! The numbers below were extracted from his thread, with the before on the left, after on the right, and the bolded stuff being the loser.
242.1398 vs 241.7512
5.479 vs 5.477
494 vs 490
736 vs 728
11614.37 vs 11770.11
12880.50 vs 13044.59
603.65 vs 601.82
9337.39 vs 9266.97
1505 vs 806
2573 vs 2596
2561 vs 5884 (lower is better)
Out of 11 tests, 8 of those tests scored higher WITHOUT HTC Performance enabled.
Click to expand...
Click to collapse

Laurentius26 said:
Hi _Alex_,
Maybe you're right but I do notice a nice speed improvement when activated!
Cheers,
Leo
Click to expand...
Click to collapse
And you can read another 191 posts about how the Apache folks at PPCGeeks argued over how it improved the speed of their devices too. My favorite quote is:
not mesurable via cpu tests, or wall clocks. Just oozes "feeling" mojo into your head. Works for me. 624 is too much mojo for me though, I really only am comfortable feeling like its 520...
Click to expand...
Click to collapse

2%
Hi _Alex_
I did both Benchmarks after a softreset and same situation on both times!
For me every improvement is an improvement, even if it's for you 'just a 2%'.
Regards,
Leo

Laurentius26 said:
Hi _Alex_
I did both Benchmarks after a softreset and same situation on both times!
For me every improvement is an improvement, even if it's for you 'just a 2%'.
Regards,
Leo
Click to expand...
Click to collapse
It's not an improvement though.
Like I said, if you don't believe in the margin of error of 2% and look at the raw numbers, 8 out of 11 tests show that it was -2%, which means only the remaining 3 out of 11 was +2%.
So you're willing to sacrifice 16% loss in other areas for 6% gain in others? That's a -10% result.

Ok... that's a good point!
In my next release I will remove it.
I'm only human haha...
Thank you
_Alex_ said:
It's not an improvement though.
Like I said, if you don't believe in the margin of error of 2% and look at the raw numbers, 8 out of 11 tests show that it was -2%, which means only the remaining 3 out of 11 was +2%.
So you're willing to sacrifice 16% loss in other areas for 6% gain in others? That's a -10% result.
Click to expand...
Click to collapse

WinMo modded like OSX Mobile??

Well this is my question. Windows Mobile can vary from 26mb to 80mb in a rom. But the iPhone operating system is 700mb.
http://www.roughlydrafted.com/2007/0...ernel-and-ram/
Could we develop a version of Windows Mobile, or even a small linux that could be installed on a SD card and utilize the phones memory as RAM? This would alot us more room software render better graphics and give us a much larger kernel that would also increase stability and speed.
With the current specs of a Kaiser, you might be able to use even something like Windows 98 in the background with a java GUI.
I'm just throwing out an idea, let me know what you think.

LWMcClam said:
the iPhone operating system is 700mb.
Click to expand...
Click to collapse
It's less than 500MB but we won't quibble.
a much larger kernel that would also increase stability and speed.
Click to expand...
Click to collapse
Go away, read what you wrote and come back when you realise that's just drivel.

[dev] kingdom lockscreen port

thanks to seo we have kingdom nice lockscreen
http://forum.xda-developers.com/showthread.php?t=1062962
problem is that if it's used with default bravo adreno libs we have no icons and no caller picture.
Logcat says
18:39:47.211 Debug mode10 1887 [mode10] Maximum number of texture units exceeded!
If you replace adreno libs with the ones from ace the is somehow working (after alot of errors and other problems - expected since is for ace not for bravo)
So best solution would be for us to patch lockscreen not to use other gpu libs
now the libs responsible for that error message are libmode10GLES20_ml.so libmode10Graphics.so
The reason for ace adreno libs *working* is because of the more OPEN GL extension supported ( i don't know if is only in driver or also in gpu )

Call 911 (snq)
sry for the OT.

What with overclocking the gpu, i men we can overclock cpu, maybe that would help make things faster, i mean so we can't raise the available texture channels, but it would be faster.

JeckyllHavok said:
What with overclocking the gpu, i men we can overclock cpu, maybe that would help make things faster, i mean so we can't raise the available texture channels, but it would be faster.
Click to expand...
Click to collapse
Useless imho to overclock gpu... Will result in more battery drain only without significant boost... I think the easier solution is to port or create the right drivers that can handle allctextures.or, if possible, to reduce them maybe with lower resolution...

That probably means that the driver can't handle that much textures, but if the Ace driver works, then it's not a hardware limitation, but rather a software one. So IMO something should be done not with the lockscreen but with the drivers to remove that software cap.
UPD: I wonder if the same limit is present in all drivers... More texture units unlocked => more general graphics performance...
Sent from my HTC Desire using XDA Premium App

well its impsossible to add more texture units to the desire so that wont be possible so only way is to create some hacked libs but snq can help but sadly he aint been working on these ports atm

JeckyllHavok said:
What with overclocking the gpu, i men we can overclock cpu, maybe that would help make things faster, i mean so we can't raise the available texture channels, but it would be faster.
Click to expand...
Click to collapse
Oc'ing the GPU will not give you the texture power that is required.

K900 said:
That probably means that the driver can't handle that much textures, but if the Ace driver works, then it's not a hardware limitation, but rather a software one. So IMO something should be done not with the lockscreen but with the drivers to remove that software cap...
Click to expand...
Click to collapse
This has to be the way forward, if this works with the ACE drivers then it is the case that the Adreno 200 driver software is saying 'no way' to the requests when the actual hardware can handle it.
The other idea I had is whether we can bundle the ACE drivers and edit/hack the lockscreen to use them only whereas everything else can use the standard Desire Adreno 200 drivers.

whats with reducing the animation to a desire handable level?
sry for bad english

revthanki said:
The other idea I had is whether we can bundle the ACE drivers and edit/hack the lockscreen to use them only whereas everything else can use the standard Desire Adreno 200 drivers.
Click to expand...
Click to collapse
I'm not a Java guru, but it seems to me that the libs are actually called by the VM itself and applications interact with the public API through Dalvik. So it's not like lockscreen - hack - libs, but more of lockscreen - hack? - dalvik - hack? - libs
Sent from my HTC Desire using XDA Premium App

K900 said:
I'm not a Java guru, but it seems to me that the libs are actually called by the VM itself and applications interact with the public API through Dalvik. So it's not like lockscreen - hack - libs, but more of lockscreen - hack? - dalvik - hack? - libs
Click to expand...
Click to collapse
Got it. Therefore, to run both drivers concurrently you'd need to hack the VM to provide a different API just for the lockscreen to call. That sounds intuitively messy and complex.
I think your idea, to hack the Adreno 200 drivers and raise the
limit on the shaders sounds like a good way. Especially because of two facts:
1) That the ACE (Adreno 205) drivers do work in this regard, there's no graphical corruption to suggest that the Adreno 200 cannot handle the called number of shader units.
2) Backed up by the fact that v26 of Baad's InsertCoin Rom, in which he bundled the ACE drivers did seem to have incredible general graphical performance (apart from the maps/music/games force closes) ---> so maybe the Desire hardware is capable of more than the software drivers limits it to...

Edit: lol I didn't read !!!!!!!!!!!

NeoPhyTe.x360 said:
Try this. It works, not too fine, but works.
Click to expand...
Click to collapse
We already know about that, this thread is about finding a better solution...
HTC Swyped from my Sense 3 Desire using XDA Premium

K900 said:
That probably means that the driver can't handle that much textures, but if the Ace driver works, then it's not a hardware limitation, but rather a software one. So IMO something should be done not with the lockscreen but with the drivers to remove that software cap.
Click to expand...
Click to collapse
Some more evidence that the drivers on the Adreno 200 might be the limiting facor, from http://www.mobiletechworld.com/2011/03/31/first-adreno-220-benchmarks-show-that-it-is-a-beast/
"Qualcomm’s Adreno GPU’s have always been seen as being a sub-par graphics compared to competing offerings from Imagination (PowerVR) and others like NVIDIA’s Tegra. This was essentially true with the Adreno 200 on Windows Mobile and Android mainly because of poor driver quality. Things are a bit different on Windows Phone 7 where the Adreno 200 is the only currently supported GPU, thanks to the tight integration with the OS and relatively good support provided by Microsoft. You can check out my recent Fable Coin Golf video to see that some pretty nice things can be done with it."

Hey baadnewz, the line "Maximum number of texture units exceeded" your searching is stored inside of libmode10GLES20_ml.so.
Code:
StartFrameControllereglCreateImageKHReglDestroyImageKHRexternal/MagicLab-mode10/mode10_trunk_2821_20110218/trunk/graphics/mode10GLES20/src/GraphicsDeviceGL2.cppm_eglDisplay != ((EGLDisplay)0)[mode10] Maximum number of texture units exceeded!
Problem is that contains spaces in original file, so you cant find string, try opening file and try searching it with spaces like below:
Code:
M a x i m u m n u m b e r o f t e x t u r e u n i t s e x c e e d e d !
Hope it helps.

I was thinking about a thing. The problem is that if screen is locked it cant handle textures to show caller image and icons right? But thinking about the significate of texture, we know that they are images that "wear" tridimensional or 2d object, right?
So if it is right, they must be stored in some files that the lockscreen try to recognize but is blocked by the amount of these and give error... If is all corrent, can't be easier to find these texture and remove some or resize? Is a textures number or space (mb)probem?
I dont know if is all correct or not, sorry if not..

ivicask said:
Hey baadnewz, the line "Maximum number of texture units exceeded" your searching is stored inside of libmode10GLES20_ml.so.
Code:
StartFrameControllereglCreateImageKHReglDestroyImageKHRexternal/MagicLab-mode10/mode10_trunk_2821_20110218/trunk/graphics/mode10GLES20/src/GraphicsDeviceGL2.cppm_eglDisplay != ((EGLDisplay)0)[mode10] Maximum number of texture units exceeded!
Problem is that contains spaces in original file, so you cant find string, try opening file and try searching it with spaces like below:
Code:
M a x i m u m n u m b e r o f t e x t u r e u n i t s e x c e e d e d !
Hope it helps.
Click to expand...
Click to collapse
thanks alot mate ,now i feel like an idiot

revthanki said:
Some more evidence that the drivers on the Adreno 200 might be the limiting facor, from http://www.mobiletechworld.com/2011/03/31/first-adreno-220-benchmarks-show-that-it-is-a-beast/
"Qualcomm’s Adreno GPU’s have always been seen as being a sub-par graphics compared to competing offerings from Imagination (PowerVR) and others like NVIDIA’s Tegra. This was essentially true with the Adreno 200 on Windows Mobile and Android mainly because of poor driver quality. Things are a bit different on Windows Phone 7 where the Adreno 200 is the only currently supported GPU, thanks to the tight integration with the OS and relatively good support provided by Microsoft. You can check out my recent Fable Coin Golf video to see that some pretty nice things can be done with it."
Click to expand...
Click to collapse
read the line after your highlighted line. i watched the video of the fabile golf running on adreno 200 on WP7 phone and it looks pretty good! so the adreno 200 should be able to run Sense 3 as sense 3 does not contain as much 3D calculations as a game. Our issue is poor drives for adreno on android.

Nekromantik said:
read the line after your highlighted line. i watched the video of the fabile golf running on adreno 200 on WP7 phone and it looks pretty good! so the adreno 200 should be able to run Sense 3 as sense 3 does not contain as much 3D calculations as a game. Our issue is poor drives for adreno on android.
Click to expand...
Click to collapse
Hopefully... Though the worst case scenario (which I don't believe - as the ACE drivers do work) is that HTC have built the lockscreen and other elements of Sense 3.0 using some kind of GPU trickery that just cannot be supported by the Adreno 200.
Question: might it be worth looking at how the Adreno drivers are different for WP7? Of course there will be much OS related difference but could some binary hacking reveal anything of use to us?

baadnewz said:
thanks alot mate ,now i feel like an idiot
Click to expand...
Click to collapse
And i forgot also inside of file libmode10Graphics.so, same line.
BTW i use FileSeek, you can search entire rom and all the libs and stuff he can search for text, and it doesnt have prob's with spaces in names.
I dont know what you use, but if you wana try it, its free.
http://www.fileseek.ca/

Interesting bug....involving the RAM

I didn't notice this in my first 16GB Tab, but on the 32GB tab, if I open the Task Manager through the minibar, it shows 775 MB total ram I know we are supposed to have 1GB ram, so I figure the 225MB missing is due to god knows what, that's not an issue.
If you go to Settings, Applications, Running Applications, it shows you have only 512 MB ram......the real question is, which one is it?

I also noticed that in the 16gb, but ignored it since the usage of ram in the ram manager exceeds 512mb. What's bothering me is why do the usage of ram differ between the ram manager and the one in running applications.

This was the first thing I noticed when I turned on my 7.0 plus for the first time when I got it on Christmas day. I just disregarded it but I am happy I am not the only one who is curious as to why it is saying 775 MB of ram.

If you take a look at /proc/meminfo you can see the kernel only recognizes 775 MB of memory. Guess that explains why it felt like apps were closing more frequently than on my 1 GB phone.
Question is, why are we getting less than memory than was advertised?

On mine:
Code:
# cat /proc/meminfo
MemTotal: 794900 kB
We are getting all the memory "adverstised", but you have to remember that things like the kernel itself and certain other things take memory before hand. The same thing happens on android phones.

Well wouldn't that still show it as being "used" so to speak out of the total 1GB or 1024 MB? Or does the kernal just allocate itself and then let everything else take over from there?

Zadeis said:
Well wouldn't that still show it as being "used" so to speak out of the total 1GB or 1024 MB? Or does the kernal just allocate itself and then let everything else take over from there?
Click to expand...
Click to collapse
I honestly don't remember. I wouldn't be concerned, however, about the lower number. It's perfectly normal on the few android devices I have.

garyd9 said:
I honestly don't remember. I wouldn't be concerned, however, about the lower number. It's perfectly normal on the few android devices I have.
Click to expand...
Click to collapse
could it be that some of the ram preserved for the GPU?

dewadewi said:
could it be that some of the ram preserved for the GPU?
Click to expand...
Click to collapse
Ah... Never thought of that, but at the same time (IMHO) I think it's unlikely it would use THAT much.
Though I guess if other android devices do it according to garyd it's nothing we should worry about.

dewadewi said:
could it be that some of the ram preserved for the GPU?
Click to expand...
Click to collapse
I believe it's RAM reserved for GPU and a few other things, like the Samsung MFC and camera interface.
On Captivate, some of the "speedmod" kernels changed the reserved memory sizes to free up user RAM - but often reducing the reserved memory can break stuff. (For example, trying to port those changes to Infuse caused the camera to stop working, along with video decoding acceleration.)

Zadeis said:
Well wouldn't that still show it as being "used" so to speak out of the total 1GB or 1024 MB? Or does the kernal just allocate itself and then let everything else take over from there?
Click to expand...
Click to collapse
Keep in mind this is the Linux kernel plus android.
In this case the system shows how much memory is available after the Linux kernel and android overlay are loaded. the galaxy 7 plus has 1GB of ram.
Now if this was straight Linux without android you would see the full amount of ram listed.

[Q] 64 Bit Processor?

so i saw that the apple a7 chip is actually just a snapdragon 800 processor just like our nexus 5, does that mean the nexus 5 kernel does support 64bit?

Toxina said:
so i saw that the apple a7 chip is actually just a snapdragon 800 processor just like our nexus 5, does that mean the nexus 5 kernel does support 64bit?
Click to expand...
Click to collapse
No, it doesnt. The entire system has to be optimised for 64-bit and not only the chip.

gee2012 said:
No, it doesnt. The entire system has to be optimised for 64-bit and not only the chip.
Click to expand...
Click to collapse
Android wouldn't really need to be optimized for 64-bit like the iPhone 5s because Android uses a virtual machine (Dalvik) to run apps, whereas iOS runs applications natively so those applications would need to be optimized for 64-bit.
Sent from my HTC Sensation using Tapatalk

android1234567 said:
Android wouldn't really need to be optimized for 64-bit like the iPhone 5s because Android uses a virtual machine (Dalvik) to run apps, whereas iOS runs applications natively so those applications would need to be optimized for 64-bit.
Sent from my HTC Sensation using Tapatalk
Click to expand...
Click to collapse
Thanks. learned something.

Toxina said:
so i saw that the apple a7 chip is actually just a snapdragon 800 processor just like our nexus 5, does that mean the nexus 5 kernel does support 64bit?
Click to expand...
Click to collapse
just FYI , right now 64bit on iOS is a gimmick , notice i said " right now" ... dont you find it odd Apple never mention specs ? didnt tell us what the CPU speed was , how much ram... but all of a sudden , they tell us its 64bit architecture? hmm

android1234567 said:
Android wouldn't really need to be optimized for 64-bit like the iPhone 5s because Android uses a virtual machine (Dalvik) to run apps, whereas iOS runs applications natively so those applications would need to be optimized for 64-bit.
Sent from my HTC Sensation using Tapatalk
Click to expand...
Click to collapse
Incorrect. Much,much, MUCH, of android is native code (even the dalvik interpreter), which is dominated by ARMv7 architecture at the moment. With the transition to 64 bit ARM (AArch64 mode on the ARMv8 processor), the kernel, drivers, bionic libc library, libgralloc (graphics), and countless other libraries need to be ported to 64 bit. This involves 64 bit compilers becoming release quality, and code reviews to ensure that pointer casts are handled appropriately for the transition to a larger address map.
However, there is nothing inherently better about '64 bit' and the Snapdragon 800 chip is a monster.

adma84 said:
Incorrect. Much,much, MUCH, of android is native code, which is currently ARMv7. With the transition to 64 bit ARM (AArch64 mode on the ARMv8 processor), the kernel, drivers, bionic libc library, libgralloc (graphics), and countless other libraries need to be ported to 64 bit. This involves 64 bit compilers becoming release quality, and code reviews to ensure that pointer casts are handled appropriately for the transition to a larger address map.
However, there is nothing inherently better about '64 bit' and the Snapdragon 800 chip is a monster.
Click to expand...
Click to collapse
Learn something new every day But the apps themselves on Android wouldn't need to be re-written for 64-bit like they do on the iPhone 5S, right?
Back to Apple's A7 chip, I think Apple did this to get a head start on 64-bit development; I doubt the iPhone 5S has 3.5GB+ of RAM so 64-bit doesn't seem practical for the 5S.

android1234567 said:
Learn something new every day But the apps themselves on Android wouldn't need to be re-written for 64-bit like they do on the iPhone 5S, right?
Back to Apple's A7 chip, I think Apple did this to get a head start on 64-bit development; I doubt the iPhone 5S has 3.5GB+ of RAM so 64-bit doesn't seem practical for the 5S.
Click to expand...
Click to collapse
Yes and No. Some Apps (most Games) run native code the would have to be rewitten in 64bit.
I think it has been confirmed that the 5S has 1GB of Ram.

64-bit is meh.. its going to take time for it to mature. Its still in its infant stages and will take time... but eventually down the road, it'll become the standard. For now, I don't think its that much of a thing to look at when buying a phone.

zephiK said:
64-bit is meh.. its going to take time for it to mature. Its still in its infant stages and will take time... but eventually down the road, it'll become the standard. For now, I don't think its that much of a thing to look at when buying a phone.
Click to expand...
Click to collapse
Yep. In fact, the best thing about 64 bit is the ability to see a 4GB+ memory space. ARMv8 does a nice job of cleaning up the instruction set (I spend my days writing ARMv8 right now), but I expect power to be an issue even at the cost of possible speed improvements due to doubling neon/VFP registers and other such improvements

adma84 said:
Yep. In fact, the best thing about 64 bit is the ability to see a 4GB+ memory space. ARMv8 does a nice job of cleaning up the instruction set (I spend my days writing ARMv8 right now), but I expect power to be an issue even at the cost of possible speed improvements due to doubling neon/VFP registers and other such improvements
Click to expand...
Click to collapse
This...
If the addressable mem space (RAM) goes unutilized, the cleaner instruction set remains the only pro. For now, gimmick.... Down the line, standard.
booooom

A7 is not by any means close to a Snapdragon, completely different designs. But similar performance though.
Sent from my GT-I9100 using Tapatalk 4