Guest Stephen Davidson Posted November 23, 2010 Report Posted November 23, 2010 I've just made a few changes to the stock 2.2 rom @ 1190 from Dell and am getting amazing results in Quadrant. Anyone else out there have some results for comparison? Quadrant: 2224 (Tested 3 times and all near that) Linpack: 38 Neocore: 32.4 (Sound) Neocore: 32.7 (NoSound)
Guest synaptic.flaws Posted November 24, 2010 Report Posted November 24, 2010 I've just made a few changes to the stock 2.2 rom @ 1190 from Dell and am getting amazing results in Quadrant. Anyone else out there have some results for comparison? Quadrant: 2224 (Tested 3 times and all near that) Linpack: 38 Neocore: 32.4 (Sound) Neocore: 32.7 (NoSound) how did you overclock the stock rom?
Guest Stephen Davidson Posted November 24, 2010 Report Posted November 24, 2010 how did you overclock the stock rom? Used DJSteve's Kernel, and stock system. Works really, really well.
Guest 3feng Posted November 24, 2010 Report Posted November 24, 2010 Used DJSteve's Kernel, and stock system. Works really, really well. how did you do it? thanks.
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 Pretty impressive stevenD! What version of steves kernel did you use? With the initial bfs oc I could get 40.3 in linpack, 1800 quadrant. With the current version I get 39linpack, 1950 quadrant What other changes have you made? I/o?
Guest synaptic.flaws Posted November 24, 2010 Report Posted November 24, 2010 Just so I am clear 1) install stock 2.2 315bb 2) flash clockwork 3) update the 1.4.5 from Steve?
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 (edited) Just so I am clear 1) install stock 2.2 315bb 2) flash clockwork 3) update the 1.4.5 from Steve? thats not what StephenD has done.. he's installed the stock rom and then a boot.img that's based on DJ_steves kernel. which is what smoku has done http://forum.xda-developers.com/showthread.php?p=9394243 but not with the same results, as quadrant isnt scoring that high Edited November 24, 2010 by fards
Guest synaptic.flaws Posted November 24, 2010 Report Posted November 24, 2010 thats not what StephenD has done.. he's installed the stock rom and then a boot.img that's based on DJ_steves kernel. which is what smoku has done http://forum.xda-developers.com/showthread.php?p=9394243 but not with the same results, as quadrant isnt scoring that high Oh ok, so I fashion my current 1.4.5 pull the system partition and flash it onto of the 2.2 official I was happy with my 1600+ quadrant until I read that
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 Oh ok, so I fashion my current 1.4.5 pull the system partition and flash it onto of the 2.2 official I was happy with my 1600+ quadrant until I read that thats why I want to find out what he's done... I'm on 145 and can't get quadrant past 1900 with stock system and the same boot.img smoku used I can't get past 1800
Guest Stephen Davidson Posted November 24, 2010 Report Posted November 24, 2010 Ok so the primary changes are as follows. In build.prop I commented out the last line: dalvik.vm.stack-trace-file=/data/anr/traces.txt Changed dalvik heapsize to 32m, And set the LCD density to 200. No other changes that I can think of although my streak has a 16GB internal with ext2 vs ext3. I also had to copy the dhd.ko from DJSteve's rom to get wireless working. Everything else seem to be the same. Actually, I just realized that I'm using DJSteve's build.prop (modified as above). Try that, I'd be interested to see if that does it. thats why I want to find out what he's done... I'm on 145 and can't get quadrant past 1900 with stock system and the same boot.img smoku used I can't get past 1800
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 (edited) Ok so the primary changes are as follows. In build.prop I commented out the last line: dalvik.vm.stack-trace-file=/data/anr/traces.txt Changed dalvik heapsize to 32m, And set the LCD density to 200. No other changes that I can think of although my streak has a 16GB internal with ext2 vs ext3. I also had to copy the dhd.ko from DJSteve's rom to get wireless working. Everything else seem to be the same. Actually, I just realized that I'm using DJSteve's build.prop (modified as above). Try that, I'd be interested to see if that does it. cheers. testing on steves 145 (with some more of my tweaks) in quadrant 3 runs each change disabling traces gives small amount of difference, an extra 50-70 points vm 32 drops it back 50-80 points lcd density is the real change 200 gives over 250 points difference, av 5fps extra in the corridor test best score I've got is 2210 Edited November 24, 2010 by fards
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 I got 1826 on stock speeds with no overclock B) using 240 lcd http://yfrog.com/5y5kd0j
Guest Stephen Davidson Posted November 24, 2010 Report Posted November 24, 2010 Right now all of my runs, score between 2240 to 2550. I assume its background processes making the difference. Aside from the benchmarking, my dell is running the best I've ever seen it. Everything is smooth and quick. The larger icons and objects actually seem to make it easier to use. Oh and I gotta say, them angry birds seem to be seriously flying... hehe ok, bad joke. What are the chances that running a class 10 card with ext2 making a difference?
Guest fards Posted November 24, 2010 Report Posted November 24, 2010 Right now all of my runs, score between 2240 to 2550. I assume its background processes making the difference. Aside from the benchmarking, my dell is running the best I've ever seen it. Everything is smooth and quick. The larger icons and objects actually seem to make it easier to use. Oh and I gotta say, them angry birds seem to be seriously flying... hehe ok, bad joke. What are the chances that running a class 10 card with ext2 making a difference? quite a bit to the quadrant.. it scores on IO after all. great scores B)
Guest synaptic.flaws Posted November 24, 2010 Report Posted November 24, 2010 (edited) quite a bit to the quadrant.. it scores on IO after all. great scores B) Fards? Is it ok. To hate him? **auto text is killing me ** Edited November 24, 2010 by synaptic.flaws
Guest Stephen Davidson Posted November 24, 2010 Report Posted November 24, 2010 Well I'm certainly not complaining. I might even just leave my dell alone for a while. It's finally hit a point where I'm more then happy with it. Of course with gingerbread right around the corner, Dec 6th, it won't be long before I'm ripping it apart again :-)
Guest synaptic.flaws Posted November 24, 2010 Report Posted November 24, 2010 Well I'm certainly not complaining. I might even just leave my dell alone for a while. It's finally hit a point where I'm more then happy with it. Of course with gingerbread right around the corner, Dec 6th, it won't be long before I'm ripping it apart again :-) so did you swap your internal SD card? I could not function with a dpi that high.. it would remind me of when I used EGA instead of VGA.. ohh I hated those days..
Guest BigBlue007 Posted November 25, 2010 Report Posted November 25, 2010 Can someone explain me something about that dpi thingy please? B) On a PC, rising the dpi value leads to an increased amount of graphics information that is being displayed ( = more dots per inch). Stuff like icons get smaller by doing that. On the other hand, whenever I read something about the density setting on Android, rising the value seems to actually LOWER the density, resulting in larger icons. The fact that a higher density value leads to better Quadrant scores would also mean that there's less graphics to compute the higher the density value is. So it would seem to me that the density value is not really a value for density, at least not in terms of "dots per inch". So the question is: What is it then?
Guest jmhalder Posted November 25, 2010 Report Posted November 25, 2010 at least not in terms of "dots per inch". So the question is: What is it then? think of it this way... lets just say the PHYSICAL size of an icon is 1 square inch, things in android should PHYSICALLY be the same size, so an icon on my TV running android should take an inch, it should also take an inch on my G1, streak, etc... now lets just hypothetically say that my TV is also 800x480... it would have less "dots per inch" than my phone right? so if it has 100dpi (just makin numbers up) it would make icons incredibly small (as it should, its a TV!) so that they are equal to one inch... now if we ran that same 100dpi setting on the streak, it would be too small to be usable... If you still don't understand, then you're doing it wrong.
Guest fards Posted November 25, 2010 Report Posted November 25, 2010 think of it this way... lets just say the PHYSICAL size of an icon is 1 square inch, things in android should PHYSICALLY be the same size, so an icon on my TV running android should take an inch, it should also take an inch on my G1, streak, etc... now lets just hypothetically say that my TV is also 800x480... it would have less "dots per inch" than my phone right? so if it has 100dpi (just makin numbers up) it would make icons incredibly small (as it should, its a TV!) so that they are equal to one inch... now if we ran that same 100dpi setting on the streak, it would be too small to be usable... If you still don't understand, then you're doing it wrong. I didn't understand a word of that B) But then the density settings make no sense to me either.
Guest fards Posted November 25, 2010 Report Posted November 25, 2010 Fards? Is it ok. To hate him? **auto text is killing me ** LoL, hate him? Nah we're all part of #teamstreak anything that furthers development is cool. And he's getting close to cyanogens speeds there! Now if we can up scores in linpack then everyone will feel the benefit..at least dj_steves hit 42 that in linpack, with a seriously overclocked kernel though...
Guest alert_bri Posted November 25, 2010 Report Posted November 25, 2010 Right now all of my runs, score between 2240 to 2550. I assume its background processes making the difference. Aside from the benchmarking, my dell is running the best I've ever seen it. Everything is smooth and quick. The larger icons and objects actually seem to make it easier to use. Oh and I gotta say, them angry birds seem to be seriously flying... hehe ok, bad joke. What are the chances that running a class 10 card with ext2 making a difference? Stephen, what make and model card are you using? may be worth it for speed alone to replace the internal 2Gb card with a faster type. Hopefully standard support of a larger capacity card could be done via a custom rom? Cheers Brian
Guest Stephen Davidson Posted November 25, 2010 Report Posted November 25, 2010 Stephen, what make and model card are you using? may be worth it for speed alone to replace the internal 2Gb card with a faster type. Hopefully standard support of a larger capacity card could be done via a custom rom? Cheers Brian I'm using a 16GB Kingston Class 10. Bought it a while ago for my hero :-)
Guest jnovello Posted November 25, 2010 Report Posted November 25, 2010 I can't stand to leave the density at anything other than 160. 200 and all the program labels are cut off and I feel like a geriatric. I can't seem to even hit 1700, but I guess it really is fine blowing all the other devices out of the water.
Guest Stephen Hyde Posted November 25, 2010 Report Posted November 25, 2010 the faster card will certainly be helping, problem is android has an annoying bug where it cant use a data partiton bigger than 2gb,
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now