[TUT] How to include 1GHz-Hack in stock-ROM - HD2 Windows Mobile 6.5 ROM Development

Is it possible for one of our chefs to include only the 1GHz-Hack to an unchanged Stock ROM? Would love to have the German-Stock-ROM speeded up ;o)

K-Hole said:
Is it possible for one of our chefs to include only the 1GHz-Hack to an unchanged Stock ROM? Would love to have the German-Stock-ROM speeded up ;o)
Click to expand...
Click to collapse
What I would like to know if this hack is really working. I'm monitoring the 1 Ghz rom threads but I haven't read a single post from a user about this hack.

COPY PAST FROM ANOTHER THREAD
-------------------------------------------
Hello
I read a lot about that, and even there is a cooked rom with title 1 GHz hack
but
I don't read any thread about "how to calculate Qualcom speed", or a an app prove that this rom has 1 GHz hack
With 576 ram we get answer
with 3D Driver Patch we have a proof by simple app
I appreciate any one can guide us to that 1 GHz hack
but until that, I satisfied there is no 1 GHz hack
dusty_nz said:
I tried the 1ghz rom and it def is a lot faster with some things.
Benching the CPU gets approx 5 - 10% on the pure CPU marks.
Disk was only average. The biggest diff I noticed was launching TomTom took a flat 2 seconds where most roms are around the 3.5 seconds.
Very interested to see this in other roms.
Click to expand...
Click to collapse
ok, great things, but this is tweaked things, we are talking about hacking from 700 Mhz to 1 Ghz about 50% increasing in speed, can you calculate the maximum CPU clock speed before and after this rom ??
that is what I talk about
---------------------------------
I think that before How to include 1GHz, must know how to calculate 1 GHz

@appelflap - the HD2 is so quick to flash, if you've got the time, it's worth giving it a go. I recently got a HD2 after a year's Blackstone, and it's a revelation.
Obviously first stop is XDA Devs for a rom, and I've currently settled on LynxEX.
Yesterday I got saucy and tried a few others, including the Eugenia rom.
The Eugenia rom is fast, but as my betters have already noted, it's difficult to make a slow HD2 rom. However, Eugenia is fast. But the difference is probably what one might imagine: 700 something hz vs 1000hz: it's the difference between stock fast, and pretty much instant (is it a third faster? Feels like it!) Switching tabs with Eugenia is pretty much instant with everything (on Sense, didn't test many other apps).
However, this difference may well be subjective, or down to other tweaks in the rom, but Eugenia does feel faster.
Unfortunately: despite the technical tour de force, I couldn't get it to sync via activesync (using google sync), it had problems changing wallpapers, had some random lock-ups and various other problems.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Small portions - so many courses!
I'd suggest that if the difference can be bench marked and proven, kholk's patch should be cooked in as an option for all roms.
I tried a few other roms, and as much as I'd love to have the Documents tab, for me, Footprints seems more useful, and most cooked roms seem to have either a working Footprint OR a documents tab - for me, LynxEX is my current sweetspot - 6.5.x, working footprints and 576mb, with little garish eye poisoning eye candy. Bravo LynxEX from me.
V

vijay555 said:
@appelflap - the HD2 is so quick to flash, if you've got the time, it's worth giving it a go. I recently got a HD2 after a year's Blackstone, and it's a revelation.
Obviously first stop is XDA Devs for a rom, and I've currently settled on LynxEX.
Yesterday I got saucy and tried a few others, including the Eugenia rom.
The Eugenia rom is fast, but as my betters have already noted, it's difficult to make a slow HD2 rom. However, Eugenia is fast. But the difference is probably what one might imagine: 700 something hz vs 1000hz: it's the difference between stock fast, and pretty much instant (is it a third faster? Feels like it!) Switching tabs with Eugenia is pretty much instant with everything (on Sense, didn't test many other apps).
However, this difference may well be subjective, or down to other tweaks in the rom, but Eugenia does feel faster.
Unfortunately: despite the technical tour de force, I couldn't get it to sync via activesync (using google sync), it had problems changing wallpapers, had some random lock-ups and various other problems.
Small portions - so many courses!
I'd suggest that if the difference can be bench marked and proven, kholk's patch should be cooked in as an option for all roms.
I tried a few other roms, and as much as I'd love to have the Documents tab, for me, Footprints seems more useful, and most cooked roms seem to have either a working Footprint OR a documents tab - for me, LynxEX is my current sweetspot - 6.5.x, working footprints and 576mb, with little garish eye poisoning eye candy. Bravo LynxEX from me.
V
Click to expand...
Click to collapse
wahey glad you've got a proper device again
are we likely to see some more coding from you to make our funny little rocketships even more powerful?
no probs with documents and footprints here.. using my own rom based on 202..
and it's very fast with out the need to swap in other devices msm_clk... which is what kholks rom appears to do.

@fards:
Yeah, I'm gonna be coding again.
I got my first HD2 last week, and was loving it for a couple of days. Then I decided to try a custom Graphics engine I wrote on it, and was royally p*ssed off when it showed up 2 dead pixels that were only apparent on a particular shade of magenta
Had to send it back to O2 and get another one, another delay with couriers, so finally got the new one back.
I need to check what the HD2's framebuffer layout is like, as I was doing some interesting graphical interface work back in May, but then got a bit burnt out by the coding.. also got made redundant, had to find a new job, interviews, family stuff - all Real Life blah blah blah. Finally I'm back on the rails and moving forward, so will get back into coding again soon.
I'd say one thing tho - the HD2 is almost a bit too big - as the wife often despairs of an evening, I can barely get my thumb from one side to the other.
An interesting tweak is to run it with a virtual mouse cursor - my old VJVirtualMouseWinFX app runs beautifully well on the glass touchscreen, and an emulated dpad and mouse cursor are damn useful with old non-touch Windows Mobile apps.
V

vijay555 said:
@fards:
Yeah, I'm gonna be coding again.
I got my first HD2 last week, and was loving it for a couple of days. Then I decided to try a custom Graphics engine I wrote on it, and was royally p*ssed off when it showed up 2 dead pixels that were only apparent on a particular shade of magenta
Had to send it back to O2 and get another one, another delay with couriers, so finally got the new one back.
I need to check what the HD2's framebuffer layout is like, as I was doing some interesting graphical interface work back in May, but then got a bit burnt out by the coding.. also got made redundant, had to find a new job, interviews, family stuff - all Real Life blah blah blah. Finally I'm back on the rails and moving forward, so will get back into coding again soon.
I'd say one thing tho - the HD2 is almost a bit too big - as the wife often despairs of an evening, I can barely get my thumb from one side to the other.
An interesting tweak is to run it with a virtual mouse cursor - my old VJVirtualMouseWinFX app runs beautifully well on the glass touchscreen, and an emulated dpad and mouse cursor are damn useful with old non-touch Windows Mobile apps.
V
Click to expand...
Click to collapse
helios may be interesting in discussing the framebuffer with you he's been trying to write some gl-es testing apps but doesn't have a HD2 to access, so I've been trying what he sends me.

I'm sorry for your broken pixels.
You might want to try to fix them yourself! there is an app called SmartUDPixel that will spawn a blinking square on your screen. Try to run it in the night, when your HTC HD2 is in charge and set it to NOT shut down the screen after N minutes.
This app has fixed dead pixels on my other devices (and on my old HD2, too!).
Talking of Eugenia ROM Project for HTC HD2, I can say that all the lockups maybe aren't because of the 1GHz Hack. Anyway, I'm on the way for fixing these lockup problems: it seems that I have to fix some settings and use some other files.
The 1GHz Hack is real and you can see its full benefits on my Eugenia ROM for HTC HD2: its browsing experience with Opera and its speed in many other things is really high, and it's doing all the things almost instantly even after a week of usage (tested on my HD2).
Unfortunately, in the WindowsMobile world, there isn't an app that really needs all of this processing power, so the CPU scaling is really fast and you won't notice the 1024MHz speed with your eyes on a CPU clock control program, but you'll see it in the everyday use of your device.
This is because of our NAND chip: we haven't got an HDD that is capable of reading/writing at 20MB/s but only a fast NAND (that is really slow compared to an HDD).
I'm saying this because every program here is processing on CPU and saving to NAND/RAM (even the RAM is slower than DDR2 memory of a normal PC), so it'll process faster going to 1GHz but reverting almost instantly to ~720MHz because it have to save the "changes" to RAM/NAND.
Anyway, it's a proud to read all these compliments on my project.
Make a good use of my work!
The Eugenia Project Developer
-kholk.

Hi kholk
congratulation for your hack
i have tested in my rom
and work well
i have tested with navifgation software iGO8 with gfx and graphic effect
after hack,this result is very very good
better graphical fluidity and speed
thank's for sharing this hack
Cheer's

alesscam said:
Hi kholk
congratulation for your hack
i have tested in my rom
and work well
i have tested with navifgation software iGO8 with gfx and graphic effect
after hack,this result is very very good
better graphical fluidity and speed
thank's for sharing this hack
Cheer's
Click to expand...
Click to collapse
Ale don't tell him or he will start saying he's the better
Awesome work my friend

Fab985 said:
Ale don't tell him or he will start saying he's the better
Awesome work my friend
Click to expand...
Click to collapse
.....
it deserves
still the user they are not convinced of the improvements
I now yes
Good work kholk

Here are my feeling about the 1 ghz hack. It doesn't cause any problems that I'm aware of at this point so there is no stability issue. If it raises the ceiling a little bit all the better. From what I have read when trying to push the HD2 to it's limit the CPU seemed to not get over the 700 + mark I forget the exact number. With the new driver it may allow the cpu to go all the way to 100 %
I don't think you can put any measurable difference on it with the tools we have right now. I'm sure someone will create something to test with in the near future. As already mentioned there is not really any programs intensive enough to keep the processor up for longer than a few seconds so you might see something, feel smoother or faster but it is not a night and day difference. Whatever the difference as long as it does not cause adverse effects it is worth using.

why the spbbenchmark should not show it?
Why the cpu measurements of SPBBenchMark should not show it?
If the same rom is available on both mode, I can test it with these bench to see if I can measure the diffrences.
Is there possible to provide a cab to switch mode? or it's too low level to be managed by this way?

bobsbbq said:
Here are my feeling about the 1 ghz hack. It doesn't cause any problems that I'm aware of at this point so there is no stability issue. If it raises the ceiling a little bit all the better. From what I have read when trying to push the HD2 to it's limit the CPU seemed to not get over the 700 + mark I forget the exact number. With the new driver it may allow the cpu to go all the way to 100 %
I don't think you can put any measurable difference on it with the tools we have right now. I'm sure someone will create something to test with in the near future. As already mentioned there is not really any programs intensive enough to keep the processor up for longer than a few seconds so you might see something, feel smoother or faster but it is not a night and day difference. Whatever the difference as long as it does not cause adverse effects it is worth using.
Click to expand...
Click to collapse
Cofirm all Bob
but with software navigation iGO8
the difference he is truly remarkable
the driver with it comes taken advantage a lot

@ vijay555
Hey old friend...happy to see you over on HD2...cant wait to see what you will develop for this device.

Hi,
i have also tested.
test with Chainfire 3D Driver Patch:
SPB Benchmark (without 1GHz Hack) said:
Graphics test: DDB BitBlt 6 ms
Graphics test: DIB BitBlt 12 ms
Graphics test: GAPI BitBlt 5 ms
Compress 1 MB file using Zip 778 ms
Decompress 1024x768 JPEG file 173 ms
Arkaball frames per second 15 ms
CPU test: Whetstones MFLOPS 1455 ms
CPU test: Whetstone MOPS 440 ms
CPU test: Whetstone MWIPS 2870 ms
Copy 1 MB using memcpy 4 ms
Click to expand...
Click to collapse
SPB Benchmark (with 1GHz Hack) said:
Graphics test: DDB BitBlt 0 ms
Graphics test: DIB BitBlt 14 ms
Graphics test: GAPI BitBlt 6 ms
Compress 1 MB file using Zip 755 ms
Decompress 1024x768 JPEG file 146 ms
Arkaball frames per second 12 ms
CPU test: Whetstones MFLOPS 1473 ms
CPU test: Whetstone MOPS 388 ms
CPU test: Whetstone MWIPS 3008 ms
Copy 1 MB using memcpy 4 ms
Click to expand...
Click to collapse
new test without Chainfire 3D Driver Patch:
SPB Benchmark (without 1GHz Hack) said:
Graphics test: DDB BitBlt 6 ms
Graphics test: DIB BitBlt 12 ms
Graphics test: GAPI BitBlt 5 ms
Compress 1 MB file using Zip 753 ms
Decompress 1024x768 JPEG file 173 ms
Arkaball frames per second 9 ms
CPU test: Whetstones MFLOPS 1451 ms
CPU test: Whetstone MOPS 550 ms
CPU test: Whetstone MWIPS 2871 ms
Copy 1 MB using memcpy 3 ms
Click to expand...
Click to collapse
SPB Benchmark (with 1GHz Hack) said:
Graphics test: DDB BitBlt 6 ms
Graphics test: DIB BitBlt 12 ms
Graphics test: GAPI BitBlt 5 ms
Compress 1 MB file using Zip 761 ms
Decompress 1024x768 JPEG file 135 ms
Arkaball frames per second 6 ms
CPU test: Whetstones MFLOPS 1452 ms
CPU test: Whetstone MOPS 551 ms
CPU test: Whetstone MWIPS 2869 ms
Copy 1 MB using memcpy 4 ms
Click to expand...
Click to collapse
with friendly greet
starbase64

starbase64 said:
Hi,
i have also tested.
test with Chainfire 3D Driver Patch:
new test without Chainfire 3D Driver Patch:
with friendly greet
starbase64
Click to expand...
Click to collapse
What ROM is this test on? The compress values are really slow. Mine are 630 and 131
It doesn't show much difference does it. With or without

bobsbbq said:
What ROM is this test on?
Click to expand...
Click to collapse
Hi,
test based on Stock Rom (recooked with the Hack).
with friendly greet
starbase64

a little test:
i'd convert from modules to file modded msm_clk then i'd compare msm_clk.dll with the msm_clk.dll put in oemdriverhigh of the shipped italy rom 1.66 and ... surprise.... ARE THE SAME.... point to point (also in hex)... very strange.....
what day is today? april 1th

I don't know why I didn't think of this? I just did a compare of the 2.02 msm_clk.dll and the one for the supposed hack and they are Binary the same with Beyond Compare. When means Identical. There is something fishy going on.

Related

XCPUScalar 2004 2.6

Hello,
Has anyone tried using XCPUScalar? There is a new version available released today. I'd just like to know your comments if it's worth it, and if this type of application is safe to use.
Poch
I use v2.53 which seems very stable.
I'm not sure how good it is at over-clocking and the real difference this makes yo the XDA's useability.
However, the fact that it can reduce CPU speed depending on need, and therefore extend battery life is quite useful.
From the release notes for 2.6, there are a couple of improvements I'd welcome, such as the ability to set the speed on boot rather than revert to default.
Is it stable...? Did it hardreset anyone?
Last time I tried to overclock a PDA was with one of my Ipaqs... it hard reset... and I never did it again...
Can someone give me enough reasurance to try this new piece of SW?
Thanks,
Chris
started using xcpu scalar last night, and i gotta say, its incredible!
got my xda2 running at 530mhz at the moment
before that i used to use pocket hack master, and was good, but complicated,* plus my xda2 would only run stable at 528mhz*
so i got 2mhz extra lol! ;-)
i highly recomend!!
Thank you!
But... how do you know up till what point is stable... and at what point it hardresets? How do you know you could push it to 530 rather than 500 or 700? You push it until it crashes? You get a warning message?
Sorry if the question is a bit naive :roll:
Thanks,
Chris
The maximum stable frequency for xda2 is 472-530 Mhz.
Where 472 has more benefits in stability and fatest memory clock.
Mine runs good on 530 in Os menu,but freezes instantly in hard processing soft..like Quake,WmPlayer.(it is also like to anaylze if some programs are running in background and don't support these frequencies)
Note: some application like Camera are more sensitive than others to overclocked fr. but personally I did only encontered benefits in Camera app with more fps.
A benchmark reported that in same cases better performance is on 400mhz and 472 max. (on xda2)
Faster overclock will only degrade your performance,but you can try it.. at least you will have to Soft-Reset or Hard-Reset your device,so backu-up first all your data.
aikon359 said:
started using xcpu scalar last night, and i gotta say, its incredible!
got my xda2 running at 530mhz at the moment
before that i used to use pocket hack master, and was good, but complicated,* plus my xda2 would only run stable at 528mhz*
so i got 2mhz extra lol! ;-)
i highly recomend!!
Click to expand...
Click to collapse
There is no 2mhz extra,the real freq. is 528 but approx. to 530
Also Hack Master is recommende for more professional use,because it offer many options like no other overclock application.
I am sure you are interested to read the overclock results:
http://www.mtekk.com.au/browse/page729.html
Regards,
Ser

Speedbooster and your Diamond will fly!!

Hi to all, I would share with you my experience with this software:
http://www.teksoftco.com/index.php?section=speedbooster
just installed on my Diamond, and now everything runs smoother and faster!!
I don't know exactly how it works but I must say that it's amazing!
I didn't notice any difference ... :/
Thanks Diego,
I will play around with it this to compare the before and after.
give us a screenshot with the settings that you made so we can try them, I didn't notice any difference either...
diego1810 said:
Hi to all, I would share with you my experience with this software:
http://www.teksoftco.com/index.php?section=speedbooster
just installed on my Diamond, and now everything runs smoother and faster!!
I don't know exactly how it works but I must say that it's amazing!
Click to expand...
Click to collapse
Yes, it makes a difference for my diamond too.
I'm using the settings shown on their forum:
Code:
increase gwes.exe for overall drawing speed
increase targetapp.exe to make it run smoother
change target app to the process you use most, like browser or game.
Does this app have any impact on Battery?
xavierdemon said:
Does this app have any impact on Battery?
Click to expand...
Click to collapse
overclocking usually kills the battery but diamond microprocessor cant be overcloced... i dont understand very well what does this app do to the device....
i am gonna try it... the posted one is trial 5 days.
It does not overclock the diamond CPU as so far it is not possible, i have CPU benchmarked before and after to prove this and it is no different.
I also do not notice any performance difference from this software, but that may be due to running duttys rom which already includes many performance tweaks.
I installed the trial and maybe I was being paranoid but my battery levels were just dropping more than normal and I wasn't even using the phone . Didn't notice any improvement on the phone either while I had a quick play.
Speedbooster uninstalled.
Speedbooster - Does it work??
Well... I downloaded the 5 day trial but the trial is limited to only partially tweaking/enhancing aspects of the Diamond.
But in saying that I noticed some apps refreshing faster. At almost 15 Euro's I am not game enough to buy this program to see if it works. Our poor Australian dollar would make this program as expensive as 9 BigMacs (without fries!) or half a tank of petrol.
I have tried so many tweaking programs and my conclusion is that if you want your Diamond to run faster wait for future models with a faster processor than the now old & slow Qualcomm 528Mhz. I read that 600 Mhz to 1 Ghz processors are coming.
I would certainly give up my 9 BigMacs for a faster CPU. Only with that will we see a real speed boost in our HTC's!!!
Aussie
correct me if I am wrong but doesnt changing the Pagepool do the same?
anyway all we are talking about here is a milliseconds in time difference.
Are we that time poor that we cant wait 0.5 second for an app to load?
The diamond has its limitations & i would say after a year or so of tweaking the ROMs & all the apps that we have just about reached that limitation.
sounds like a scam to me
No much good reactions, thats sad, i hoped this program would be a nice add-on to our Diamond.
Yeah, just lets wait on new phones with a decent processor, 1 Ghz would be awesome, i'm sure they are will coming very soon
aussie1234 said:
Well... I downloaded the 5 day trial but the trial is limited to only partially tweaking/enhancing aspects of the Diamond.
But in saying that I noticed some apps refreshing faster. At almost 15 Euro's I am not game enough to buy this program to see if it works. Our poor Australian dollar would make this program as expensive as 9 BigMacs (without fries!) or half a tank of petrol.
Click to expand...
Click to collapse
it tried a full cracked version and didnt like it. after a lot of benchmarks i didnt notice any difference...
Ι also used ful verison, to my custom rom, and turn to full gwes.exe and manila2.exe . The diference is very big. I didnt expected it. I dont know if the same will happen to original rom version . (I dont think there is manila2.exe to the original rom)
sosimple said:
Ι also used ful verison, to my custom rom, and turn to full gwes.exe and manila2.exe . The diference is very big. I didnt expected it. I dont know if the same will happen to original rom version . (I dont think there is manila2.exe to the original rom)
Click to expand...
Click to collapse
Yeah its the same for me. Speedbooster makes some apps run much faster.
Strange thing is that the benchmark result is usually the same. This might be a good thing - meaning that the CPU speed remains the same, but it just gives more power to some specific app instead of another. Actually this is exactly what I want:
I don't care about the stupid OS doing stupid things, but I DO CARE about my GPS or webbrowser - and here this proggie seems to give a useful hand.
I agree about the price, is a bit expensive, but at least it works
About the price...
You can get it for free on the developer's website, it's some kind of promotion, you only need to install it, run a benchmark with the built in tool, and post your score:
http://teksoftco.com/forum/viewtopic.php?t=2011
hurry up, I don't know how much it will last!
How it works.
It modifies the process priority similarly to the task manager in windows.
It does other stuff too but that's basically what it does.
and what's the verdict ? does it speed up our Diamonds ? (i was testing it... and didn't notice any difference :/ )
It DOES have impact on performance.
Its very easy to test. Try to lower the system apps and services and you will se that the device run slower.
On TF3D its very clear the it makes it run faster/slower if you change the priority.
My opinion; It works...somehow.

Looking for a 3D rematch, results after NEON

I'm sure you're all familiar with this benchmark test:
http://distinctivegame.wordpress.com/2010/02/09/dd-tech-talk-1-nexus-versus-iphone-3gs/
http://www.youtube.com/watch?v=GYqZK25KU3U
Its not a fair fight because:
1)The stock Android 2.1 OS does not utilize the new NEON instruction set yet.
2) the iphone is running at 480x320 where as the N1 is at 800x480. (i.e. the Nexus one's resolution is 2.5 times that of the 3GS...meaning atleast 2.5x the amount of processing). He did try running the game at 480x320 but got similar results with more fighters
Can someone with a new Custom ROM (compiled with NEON included) like the latest Cyanogen ROM, run this test again and let us know what you find.
update: Since nobody knows which game this is, lets try the free app "NeoCore" from the market so we see what kind of a difference NEON makes.
NeoCore. Dec 30/2009 Nexus One. 26.7 FPS
http://www.slashgear.com/google-nexus-one-3d-neocore-benchmark-video-3067327//
This is a weapon iphone users keep using against us and we need to set the record straight! (like the screen colors fiasco)
ps: I Do not know which game this is.
update 3/12/10, NeoCore gives me 27.1 FPS at 800x480 with my CPU at 1113Mhz
I'll be happy to run some tests if you can supply the correct link. I saw nothing but a long review(?).
Your link points at an HD2 vs. N1 shootout from Phonedog. Is this the video you are talking about?
http://www.youtube.com/watch?v=GYqZK25KU3U
flarbear said:
Your link points at an HD2 vs. N1 shootout from Phonedog. Is this the video you are talking about?
http://www.youtube.com/watch?v=GYqZK25KU3U
Click to expand...
Click to collapse
Yes! thanks
I'd be happy to run it if someone can give me a link to the benchmark program used.
It is very odd that the app/tool they used is mentioned nowhere.
A little googling for iPhone benchmark apps showed me that "GLBenchmark" has no android version, same with "Mad Monkey Tech"
There is a "NeoCore" free utility on the market though. Maybe we can at-least get some Neon/NonNeon FPS values.
Dec/2009 Nexus One. 26.7 FPS
http://www.slashgear.com/google-nexus-one-3d-neocore-benchmark-video-3067327/
I would very much be interested in the results as well..
Well I just ran neocore on cm 5.0.4.1 and it gets the exact same score. I'm pretty sure the app has to be compiled with neon support.
I got 26.0 FPS, stock ROM.
skozsert said:
Well I just ran neocore on cm 5.0.4.1 and it gets the exact same score. I'm pretty sure the app has to be compiled with neon support.
Click to expand...
Click to collapse
I'm not sure if the app needs to be compiled specially in order to explicitely use the hardware FP instructions, or if the OS uses it automatically when enabled
NeoCore: 26.5 fps with sound, CM 5.0.4.1 with matching .33 UV kernel.
Well in the review linked in the original post it sounds like you need a compiler that uses the neon instructions as well as os support in order to take advantage of them. Isn't that how floating point extensions usually work i.e. mmx sse etc?
NeoCore gives me 27.1 FPS at 800x480 with my CPU at 1113Mhz. (UV/OC .32 kernel from 3/10, with NEON)

Got PSX games running on my apache!

Every time I think there is absolutely no life left in this poor little phone, it always surprises me. I tried it on a whim after I got it working on my Tilt 2, with no expectation it would work. Ten minutes later I'm playing Final Fantasy Tactics on my phone, and I don't mean it boots up and you have to deal with crippling lag or slowness, it's playable! On 5 frame skip it runs at about 45 FPS with and sound on!
Here is my set-up.
FPSEce 0.10
Omnia D3D Driver
HTC Performance to overclock to 624Mhz
PDAJ'S Clean WM 6.5 Build 21501 ROM (You can find it at PPCgeeks)
Only S2U2 and iContact Avian Edition are running in the background
I've only tested Final Fantasy Tactics due to having limited space on my memory card, but I plan on trying out a few more games. I also heard that disabling the sound can get you quite a few more frames per second, plan on trying that out. If anyone needs either the Omnia D3D Driver or HTC Performance cabs, I'll be more then happy to post them for you. Try it yourself and post the results.
Happy Gaming!
Niceee
That's pretty awesome dude, i would love to get PSX games running on my HTC Tilt 2.
you should see what gran turismo looks like on my expo running at 60fps.
spenceizzle said:
If anyone needs either the Omnia D3D Driver
Click to expand...
Click to collapse
Please do post it here. Is it needed to run FPSce? I Remember my old Ipaq 1940 could run them full speed without any drivers...
I hate how I spent so much money on the HTC Apache back in the day only so it could become a useless pile. I saw that post up there on a successful PSX game on the Apache and I got really excited.
I tried installing all that same stuff (the D3D driver, the overclocking utility, FPSEce) only as for the Apache's ROM, I used a custom ROM I built with PPCKitchen (it couldn't differ much from the 'clean' ROM, because you can hardly add anything to it because of the limited space).
I tried running Castlevania: Symphony of the Night and it didn't run past the "Press Start" screen. I was hoping that since the game is a fairly simple sidescroller, it would run on the Apache, but no luck.
So far (for me) the only emulator I've tried that runs well and plays smooth on the Apache is that PocketNester. The PocketSNES runs about like the DreamSNES for Dreamcast: only runs at an acceptable framerate without sound. I thought it would be better, since the Dreamcast runs at 200mhz and the Apache can be overclocked to 624mhz... Granted I've only tried these three emulators, but those were the three I knew I'd probably get the most use out of.
With this in mind, I figure a gameboy emulator would run perfectly also.

Gaming on the Find 5

Hello fellow Find 5 friends! ( say that 5 times fast :laugh: )
As we all know the Find 5 packs a:
Qualcomm S4 Pro APQ8064
Krait 200 CPU: 4 cores @ 1.5GHz
Adreno 320: 400MHz
2GB RAM @ 533 MHz dual channel
Click to expand...
Click to collapse
in its belly, and shares this with quite a list of phones:
Asus PadFone 2
HTC Droid DNA
HTC J Butterfly
LG Optimus G
Nexus 4
Pantech Vega No.6
Pantech Vega R3
Sharp Aquos Phone Zeta
Sony Xperia UL
Sony Xperia Z
Sony Xperia ZL
Sony Xperia ZR
Xiaomi MI-2
Panasonic P-02E
Note that some new devices (the Nexus 7 2013 is a good example) has a "S4 Pro" SoC in the spec sheet, but is actually a lower-binned / lower-clocked 600 chip [APQ8064–1AA to be exact] and uses Krait 300 CPU cores and therefore outperforms its Krait 200 cousins in the Nexus 4 / Find 5 / etc.
The silicon is quite formidable- and led pack until the release of Qualcomm's 600 and 800 chips. Although some argue that the Adreno 320 is slightly underpowered for 1080p, the point is the Find 5 packs alot of power.
However despite this custom ROM gaming performance was poor initially:
Entropy512 said:
Some games apparently hang the device. Others, such as Ingress, work perfectly.
Problem is our display subsystem is a really nasty hackjob that is a mix of early 4.1 source code (Oppo's original baseline) and 4.2 stuff. Ideally, it will eventually be a pure CAF jb2.5 branch baseline - this is what most other CM devices are on or going to. Some stuff is easy to rebase, some is harder. I have a feeling I'm probably going to fix the camera the same way as the Sony and Samsung guys did it - copypasta the entire camera subsystem drivers from the original baseline and use those. (The camera drivers in the kernel seem isolated enough that it works fine on Samsung and Sony qcoms.) Biggest thing is the power management, especially charging - Oppo made major changes here to Qualcomm's early-4.1 baseline, AND Qualcomm made major changes to this code. As a result, getting that stuff right and properly merged is gonna be a PITA. It's sort-of-barely working in the codebase on my personal github, but it's doing all sorts of stuff it shouldn't, and also not doing stuff it should (like charging from USB). The standby power drain is also insane, which is strange since the newer CAF codebase should be MORE power efficient, not less.
Click to expand...
Click to collapse
However since the 4.2 kernel rebase gaming performance has dramatically improved in my testing. This no doubt is related to all the kernel display changes work from our wonderful developers made possible by Oppo who released the sources for 4.2..
So pick up a ROM / kernel combination that has this 4.2 goodness and feel improvements across the board! Here is a list of the ROMs:
Color ROM [formerly known as Project Firefly] v1.0.0_33xx and later
Maxwen's AOKP 20130810 and later
CM10.1 Official Nightly 20130810 and later
CodefireX
IMPORTANT NOTE- CUSTOM ROMS USE THE 4.1 MODEM FIRMWARE
This means if you've flashed a 4.2 Color ROM and want to move to a custom ROM, you MUST flash the modem firmware from a 4.1 Color ROM before flashing a custom ROM or else WIFI WILL NOT WORK. Simply flash the 4.1 Color ROM (or the firmware package) and then flash the custom ROM immdediately after, you do not need to boot the 4.1 Color ROM.
Now back to the actual thread, I thought it would be helpful to compile a master list of games that work and don't work (for whatever reason) and use this as a future reference and for troubleshooting. Feel free to contribute: just the game name, your experience on it and if possible your ROM / kernel configuration.
My device settings:
Latest Maxwen AOKP / kernel
MP-Decision: Off
Multicore Power Saving: 0
CPU: 162MHz / 1.134 GHz
Governor: smartmax_eps
CPUQuiet: load_stats
GPU: 400 MHz
Governor: simple
Click to expand...
Click to collapse
PLAYS FLAWLESSLY
7x7
Asphalt 7: Heat
Cut the Rope / HD
Fieldrunners 2
Formerly crashed after logo load, reinstall fixed it. Perhaps file was corrupted first time downloading?
Flow Free
GTA: Vice City
Infinite Flight
Make sure to update to the latest to fix crashing.
Ingress
PewPew
PewPew 2
Plants vs Zombies
Plasma Sky
Rymdkapsel
SpeedX 3D
Super Hexagon
The Conduit HD
The Room
Tiny Thief
True Skate
PLAYABLE, BUT WITH ISSUES
Eufloria HD
Can play on max settings as well as the best device but will still stutter. Game not as optimized as it could be. Heats up fast!
Need For Speed: Most Wanted
Not perfect frame rate, experiment with governors / OC.
Osmos HD
Slight fps drops from time to time, heats up the phone pretty fast. Otherwise no problems
Subway Surfer
Not fully optimized, occasional fps drops / stuttering. Heats up phone pretty fast. Otherwise no problems.
Trial Xtreme 3
Occasional choppy sound.
BROKEN
NBA JAM
A server error has occured. Please visit www.eamobile.com/countrygate to choose your country and click the customer support link. (5002)
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
EMULATORS
With the release of DraStic I felt it was time to add a section for emulators. I will list them by console rather than app.
DS
DraStic is my emulator of choice. I'll leave the feature list to the developer, but the biggest draw is that it is very fast. Since version 2.1.1a (released 8/1x/2013) the DS BIOS file is no longer needed for unencrypted ROMs. If my word (and the word of many across the internet) is not enough to convince you to drop $8, there is a demo version of the app (full functionality, 20 minute limit)
Some select quotes from the FAQ:
Q: Does DraStic support 3DS games? Will it in the future?
A: No, it does not and never will. 3DS is an entirely different system, a full generation ahead, like N64 to Gamecube.
Click to expand...
Click to collapse
Q: How fast should I expect DraStic to be?
A: It varies greatly depending on the game you're trying to run. A device with 1GHz Cortex-A8 will run many games with a fair amount of frameskip. A 1.4+GHz dual-core Cortex-A9 will run most games with little or no frameskip most of the time.
Click to expand...
Click to collapse
Q: Does it help to have multiple CPU cores?
A: Yes. If you have multiple cores DraStic is sped up by being able to offload screen update tasks to another thread. If you select the multi-threaded renderer option you will get even more benefit, but this option can cause graphical glitches and instability.
Click to expand...
Click to collapse
Q: Are MOGA controllers supported?
A: It is possible to make it work by installing Moga Universal Driver from Google Play. We are currently working on including native support.
Click to expand...
Click to collapse
Q: Does DraStic use my GPU to render graphics? Does it help to have a faster GPU
A: The GPU is only used to scale the image and apply filters so having a better one won't make much difference. OpenGL ES is not used to render the Nintendo DS graphics. This is because of idiosynchracies in the DS graphics hardware that would make this style of emulation incomplete and/or slow.
Click to expand...
Click to collapse
Q: Does DraStic support homebrew games?
A: Currently many homebrew games are not supported because DraStic does not yet support DLDI filesystem integration which many homebrew games rely on. This is a planned feature for a later date.
Click to expand...
Click to collapse
I have only played a few games (namely ones that have simple controls) and they have been flawless.
My settings:
Frameskip: Auto [4]
Filter: Scale2X
Multi-threaded 3D rendering: Off [haven't played a game that needs it]
Sound Latency: Medium [Low is the best quality but introduces some clicking in my testing]
Click to expand...
Click to collapse
Ace Attorney Investigations: Miles Edgeworth
Apollo Justice: Ace Attorney
Ghost Trick: Phantom Detective
Phoenix Wright: Ace Attorney
Phoenix Wright: Ace Attorney: Justice for All
Phoenix Wright: Ace Attorney: Trials and Tribulations
Well I don't game much but even the lightest game I play, Rage of Bahamut, lags worse than on any other Handset I tried it.
Also I have to admit that on AOSP Roms even the interaction with the UI feels like there is something wrong with the phone. Compared to a Nexus 4 or a S3 or S4 the UI lags or slows down everywhere from time to time...
I don't play many games, in fact I only play Tekken Card Tournament. But since I have got my Find 5 the game always crash my phone, sometimes already after seconds.
Hope the devs will fix this soon and get the phone stable. Currently using cm10.1 and maxwen kernel.
Sent from my Find 5 using xda app-developers app
Updated OP with some new findings and reports from other users and developers. Found a fix for the performance issues (ended up being an extension of a weird interaction I found)
Basically, if the game / application lags, turn the screen off and on again without exiting the application- upon resuming the lag should be gone!
House of the Dead
+ Playable, seem to run smooth at high fps.
+ Responsive controls.
- Random crashes (general game issue, not device or rom).
Street Fighter 4
+ Playable, high fps
- Audio feedback delayed
Gaming performance on color rom.is much better than previous roms
Sent from my X909 using xda app-developers app
Any free game that I can us to reproduce the issues?
It should happen as fast and as often as possible
Carmageddon free reboots. I can't. Logcat, I have the temp issue and cm is broken for me.
Sent from my X909 using xda app-developers app
Please give latest test kernel build a try
http://forum.xda-developers.com/showpost.php?p=44033660&postcount=646
I have included a few CAF fixes in the display driver and would be
very interessted if they change anything in games that have issues
I have one positive report on oppoforum that with latest test
kernel games issues are gone
please give it a try and report back
that's great news, unfortunately for me I still get low frame rate, although it seems better than before.
screen off / on trick still works fine
Sent from my Find 5 using xda app-developers app
DarK_MischieF said:
that's great news, unfortunately for me I still get low frame rate, although it seems better than before.
screen off / on trick still works fine
Click to expand...
Click to collapse
can you provide another kernel log taken when it happens
maxwen said:
can you provide another kernel log taken when it happens
Click to expand...
Click to collapse
sure can, can I get specific instructions? I did it once and I don't think the results were what you were looking for
Sent from my Find 5 using xda app-developers app
DarK_MischieF said:
sure can, can I get specific instructions? I did it once and I don't think the results were what you were looking for
Sent from my Find 5 using xda app-developers app
Click to expand...
Click to collapse
I have to admit that I lost your last log somewhere
I would need both a logcat and a kernel log
simply attach your device via adb and run
Code:
adb logcat
adb shell dmesg
copy the output of both commands into a file and upload it
maxwen said:
I have to admit that I lost your last log somewhere
I would need both a logcat and a kernel log
simply attach your device via adb and run
Code:
adb logcat
adb shell dmesg
copy the output of both commands into a file and upload it
Click to expand...
Click to collapse
I think I'm doing something wrong- I have the device connected via adb, then I launch the game and start the logcat / desg (adb logcat < log.txt). The device hard reboots after about 10 seconds. After it finished I run (adb shell cat /proc/last_msg > kmsg.txt) to get the last_kmsg
Oddly the frame rate is actually better while connected to ADB and logging then normal (of course nowhere close to post screen on/off).
logcat
https://docs.google.com/file/d/0B77jyXOXVTlRZEFOYW1iQ1ItOEk/edit?usp=sharing
dmesg
https://docs.google.com/file/d/0B77jyXOXVTlRejFCaGVQbW5YLXM/edit?usp=sharing
kmsg
https://docs.google.com/file/d/0B77jyXOXVTlRVWEwYWJNZExZMjg/edit?usp=sharing
Well - the low framerate is because device is spilling to logs
and that slows down everything
Right now I am unable to identify a change in CAF that "could" fix this issue
Problem is also that I still dont have an easy way to reproduce this here
so I can not even do some trial-and-error things.
maxwen said:
Well - the low framerate is because device is spilling to logs
and that slows down everything
Right now I am unable to identify a change in CAF that "could" fix this issue
Problem is also that I still dont have an easy way to reproduce this here
so I can not even do some trial-and-error things.
Click to expand...
Click to collapse
Of course I/O and CPU would be spent logging and frame rate would suffer, but when I was testing the FPS was actually higher than it is usually even with the logging in process- maybe because it was plugged in through USB?
Did I do the logs correctly? I don't know why my logs are so damn big!
DarK_MischieF said:
Of course I/O and CPU would be spent logging and frame rate would suffer, but when I was testing the FPS was actually higher than it is usually even with the logging in process- maybe because it was plugged in through USB?
Did I do the logs correctly? I don't know why my logs are so damn big!
Click to expand...
Click to collapse
yes they are pretty fine
still I dont see any free game where this happens - right?
I need a way to reproduce it
If you tell me that it happens with "cut the rope HD" I would be willing
to spent the bucks but I need to be sure cause with the "normal" free one
it hasnt happend so far
maxwen said:
yes they are pretty fine
still I dont see any free game where this happens - right?
I need a way to reproduce it
If you tell me that it happens with "cut the rope HD" I would be willing
to spent the bucks but I need to be sure cause with the "normal" free one
it hasnt happend so far
Click to expand...
Click to collapse
Okay excellent, I'll install the free version, and report back in 10 minutes with a log if I see frame rate issues after a clean reboot, would that help?
I also suspect I have the Chinese version (seller wasn't clear)- according to Oppo there is only a band difference (1700)- but maybe thats a hidden confounding variable?
DarK_MischieF said:
Okay excellent, I'll install the free version, and report back in 10 minutes with a log if I see frame rate issues after a clean reboot, would that help?
Click to expand...
Click to collapse
sure
DarK_MischieF said:
I also suspect I have the Chinese version (seller wasn't clear)- according to Oppo there is only a band difference (1700)- but maybe thats a hidden confounding variable?
Click to expand...
Click to collapse
dont think so

Categories

Resources