Jump to content

Welcome to CyanogenMod

Welcome to our forum. Like most online communities you must register to post, but don't worry this is a simple free process that requires minimal information for you to signup. Be a part of the CyanogenMod Forum by signing in or creating an account. You can even sign in with your Facebook or Twitter account.
  • Start new topics and reply to others
  • Subscribe to topics and forums to get automatic updates
  • Get your own profile and make new friends
  • Download files attached to the forum.
  • Customize your experience here
  • Share your CyanogenMod experience!
Guest Message by DevFuse
 

Linpack scores


  • Please log in to reply
19 replies to this topic

#1 YottaByte

YottaByte
  • Donators
  • 155 posts
  • Country:
    us - United States
  • Device Model:
    G2, N1, and G1
  • CM Version:
    CM7 nightly
  • Recovery Image:
    ClockworkMod

Posted 30 May 2010 - 02:26 AM

Linpack, which is a benchmark app for Android shows list of top MFLOP scores by device:
http://www.greenecom...pack-by-device/
If you score down to the G1 section, people are getting 10-15 MFLOPs!
I only got 3.09 =\

Now I'm wondering, if people are getting these scores, are their phones simply running faster?
I mean, I'm running a 32mb linux swap, compcache is enabled, JIT enabled (brought my score from 1.5 to 3.09, very easy to install), apps2sd, and I have less than 50 apps on my class 6 micro sd. My internal memory is also more than 50% free.

So how are these guys making there G1's get better scores than stock HTC Incredible's, which is advertised as the fastest Android device?
Anyone have any tips to increase my G1's speed, like increase the size of my swap maybe?

edit: I am also overclocking my phone to 528mhz, and I have no themes installed.

#2 Authority

Authority
  • Donators
  • 33 posts
  • Country:
    us - United States
  • Location:
    Salt Lake City, Utah USA
  • Device Model:
    Samsung Galaxy S III
  • CM Version:
    Current
  • Recovery Image:
    ClockworkMod

Posted 30 May 2010 - 03:18 AM

I only got a 2.332  But my phone seems fast enough, I got Black Bar theme... and Lots of apps.... so Yea... 3.0 seems good to me?

I dont know how people are making their G1's get 15 or anything like that but I think there was an old post on the forums before reset, and the Average was 2.0-2.6.

#3 crossed.out

crossed.out
  • Members
  • 5 posts
  • Device Model:
    |2|
  • CM Version:
    5.07
  • Recovery Image:
    Amon_RA

Posted 30 May 2010 - 03:24 AM

Just installed JIT on my Magic and I'm getting around 3.8.

64 mb Swap
Apps2SD enabled
CPU overclocked to 576 mhz

#4 Transmition

Transmition
  • Members
  • 41 posts
  • Country:
    us - United States
  • Device Model:
    HTC Dream
  • CM Version:
    2.2 RC2 DS
  • Recovery Image:
    Amon_RA

Posted 30 May 2010 - 05:23 AM

2.18
32 MB Swap
Overclocked to 576 MHz

HTC Dream.

I'll be checking out JIT though. Hopefully it works with the EBI1 kernel.

#5 j-riv

j-riv
  • Members
  • 10 posts
  • Country:
    us - United States
  • Location:
    Bremerton,wa
  • Device Model:
    |7|
  • CM Version:
    none
  • Recovery Image:
    TWRP

Posted 30 May 2010 - 12:50 PM

the highest i have ever gotten its 4.8 thats with jit and overclocked to 653mhz

#6 YottaByte

YottaByte
  • Donators
  • 155 posts
  • Country:
    us - United States
  • Device Model:
    G2, N1, and G1
  • CM Version:
    CM7 nightly
  • Recovery Image:
    ClockworkMod

Posted 30 May 2010 - 02:48 PM

So if I made my swap 100mb, would it be beneficial for the speed of my phone?
And how do you overclock your phone past 528mhz?

#7 Transmition

Transmition
  • Members
  • 41 posts
  • Country:
    us - United States
  • Device Model:
    HTC Dream
  • CM Version:
    2.2 RC2 DS
  • Recovery Image:
    Amon_RA

Posted 30 May 2010 - 04:41 PM

View PostYottaByte, on 30 May 2010 - 02:48 PM, said:

So if I made my swap 100mb, would it be beneficial for the speed of my phone?
And how do you overclock your phone past 528mhz?

It probably would not. Swap tends to be more useful in terms of multitasking. As for overclocking, use setcpu, and instead of choosing dream/magic as your phone model, have it autodetect based on your kernel.

I enabled JIT and my score went up to 3.8

#8 j-riv

j-riv
  • Members
  • 10 posts
  • Country:
    us - United States
  • Location:
    Bremerton,wa
  • Device Model:
    |7|
  • CM Version:
    none
  • Recovery Image:
    TWRP

Posted 30 May 2010 - 04:59 PM

View PostTransmition, on 30 May 2010 - 04:41 PM, said:

It probably would not. Swap tends to be more useful in terms of multitasking. As for overclocking, use setcpu, and instead of choosing dream/magic as your phone model, have it autodetect based on your kernel.



if you have the latest set cpu there is only two options there is no more chosing by phone option.chose the first which is auto detect after you select auto detect you have to adjust the freqeuncy to what you want dont go above 652mhz though otherwise you will get random reboots

#9 Skrki

Skrki
  • Moderators
  • 623 posts
  • Country:
    rs - Serbia
  • Device Model:
    MT3G 32B / Desire Z
  • CM Version:
    CM6.1 2708 port / CM7 nightly 263
  • Recovery Image:
    Stock

Posted 30 May 2010 - 05:08 PM

Seeing the direction this thread is going, the subtitle should be "...or, how to kill your Dream"
lol just kidding guys (:

Seriously, I've just enabled JIT because I haven't tried it on CM5, but my phone got really slow (MT3G), Market started failing at downloading, battery drain ...don't even get me started. I am back to stock.. stock CM5, that is.

Overclocking passed the max frequency will, however, kill your Dream...eventually. No jokes there.

Swapping on CM5 = boring topic. Seen this? I'm sure you have:

Cyanogen@GitHub[change log] said:

* D/S: Change oom-killer behavior to be more "swap friendly" if you want to go against everyones advice and use swap anyway


Linpack scores are irrelevant - they're just that, benchmark scores.

Ditch the JIT for now.
Have a stable device.

Just my 2c.

#10 3mp0w3r

3mp0w3r
  • Members
  • 1120 posts
  • Country:
    ie - Ireland
  • Location:
    Ireland
  • Device Model:
    HTC Dream
  • CM Version:
    CM D-S nightly
  • Recovery Image:
    Amon_RA

Posted 12 June 2010 - 11:45 PM

+1

#11 nwordjim

nwordjim
  • Members
  • 267 posts
  • Country:
    us - United States
  • Location:
    Oregon
  • Device Model:
    Tmobile G1 & G2
  • CM Version:
    CM 7 Nightly 58
  • Recovery Image:
    ClockworkMod

Posted 12 June 2010 - 11:50 PM

G1, 5.0.8 test4, No Overclocking, JIT enabled, no swap or comp cache.
Best was 3.709

#12 nwordjim

nwordjim
  • Members
  • 267 posts
  • Country:
    us - United States
  • Location:
    Oregon
  • Device Model:
    Tmobile G1 & G2
  • CM Version:
    CM 7 Nightly 58
  • Recovery Image:
    ClockworkMod

Posted 12 June 2010 - 11:55 PM

View Postskrki, on 30 May 2010 - 05:08 PM, said:


Seriously, I've just enabled JIT because I haven't tried it on CM5, <b> but my phone got really slow(MT3G)</b>, Market started failing at downloading, battery drain ...don't even get me started. I am back to stock.. stock CM5, that is.

Ditch the JIT for now.
Have a stable device.

Just my 2c.

When i first enabled JIT it ran fine after first restart then a couple hours later my memory dropped to like 4mb so I uninstalled it but if you read into the installation part of JIT it says that the phone needs a few restarts before it will run up to potential. So I reinstalled it and haven't had any problems since. Emulators run faster, phone is stable for me never had any issues with FC's or reboots. So idk why JIT would make your phone unstable but it seems everyone has a different experience with anything they put on their phones. For instance battery life with 5.0.8 tests (some claim amazing battery life and some claim miserable)

#13 DackParis

DackParis
  • Members
  • 90 posts
  • Twitter:DackParis
  • Country:
    us - United States
  • Location:
    Oklahoma City, Oklahoma
  • Device Model:
    G1 & EVO
  • CM Version:
    6.1 .0 RC1
  • Recovery Image:
    Amon_RA

Posted 13 June 2010 - 03:28 PM

View PostAuthority, on 30 May 2010 - 03:18 AM, said:

...there was an old post on the forums before reset, and the Average was 2.0-2.6.


True. I posted the "Speed" thread. What the general consensus was is that tho higher Linpack scores gives us the warm-fuzzies at the cost of ruining your device with an aggressive overclock or the instability of JIT the "benefits" were all highly subjective and personally I couldn't perceive any real increase in performance.  

I am still hitting 2.2 average on my G1 and I am most pleased on T4 not that the scores are the same but that they are consistently the same. Almost never drops down to 1.0 or 1.2 like it did before.

*shrug

#14 YottaByte

YottaByte
  • Donators
  • 155 posts
  • Country:
    us - United States
  • Device Model:
    G2, N1, and G1
  • CM Version:
    CM7 nightly
  • Recovery Image:
    ClockworkMod

Posted 13 June 2010 - 03:46 PM

View PostDackParis, on 13 June 2010 - 03:28 PM, said:

True. I posted the "Speed" thread. What the general consensus was is that tho higher Linpack scores gives us the warm-fuzzies at the cost of ruining your device with an aggressive overclock or the instability of JIT the "benefits" were all highly subjective and personally I couldn't perceive any real increase in performance.  

I am still hitting 2.2 average on my G1 and I am most pleased on T4 not that the scores are the same but that they are consistently the same. Almost never drops down to 1.0 or 1.2 like it did before.

*shrug
I am getting 3.6's with JIT but with Test 4 my phone seems to be acting slower than ever... I even used cachemate, still slow as hell. Should I increase my 32MB swap? I have swap and compcache enabled.

#15 setaou

setaou
  • Donators
  • 15 posts
  • Country:
    fr - France
  • Location:
    Glénac, Brittany
  • Device Model:
    Magic
  • CM Version:
    CyanogenMod-5.0.8-DS
  • Recovery Image:
    Amon_RA

Posted 10 August 2010 - 12:06 PM

I got 2.11.

Can you explain what is JIT. Can't do a search on it, word is too small.

#16 pbelcomp

pbelcomp

    Cheeseburger?

  • Donators
  • 662 posts
  • Twitter:pbelcomp
  • Country:
    us - United States
  • Location:
    Las Vegas
  • Device Model:
    G2, G2x
  • CM Version:
    N/A
  • Recovery Image:
    ClockworkMod

Posted 10 August 2010 - 05:31 PM

View Postsetaou, on 10 August 2010 - 12:06 PM, said:

I got 2.11.

Can you explain what is JIT. Can't do a search on it, word is too small.

Use Google for the search instead of the built-in search:

http://www.lmgtfy.co...cyanogenmod.com

#17 Gazorninplat

Gazorninplat
  • Members
  • 69 posts
  • Country:
    ca - Canada
  • Device Model:
    |1|2|
  • CM Version:
    CM 6
  • Recovery Image:
    Amon_RA

Posted 10 August 2010 - 07:21 PM

For me, JIT has only been good for bumping up benchmarks.  I have never had any real use advantages with JIT.  Games ran just as slow, some apps became incompatible.  Perhaps for a higher spec machine JIT might make a difference (I have a sapphire).  I have found BFS kernels make a big difference in real world use performance, but doesn't give the same benchmark candy JIT does.

#18 setaou

setaou
  • Donators
  • 15 posts
  • Country:
    fr - France
  • Location:
    Glénac, Brittany
  • Device Model:
    Magic
  • CM Version:
    CyanogenMod-5.0.8-DS
  • Recovery Image:
    Amon_RA

Posted 12 August 2010 - 08:25 PM

View Postpbelcomp, on 10 August 2010 - 05:31 PM, said:

Use Google for the search instead of the built-in search:

http://www.lmgtfy.co...cyanogenmod.com

Thanks.

PS : No thanks for the burger.  ;)

#19 setaou

setaou
  • Donators
  • 15 posts
  • Country:
    fr - France
  • Location:
    Glénac, Brittany
  • Device Model:
    Magic
  • CM Version:
    CyanogenMod-5.0.8-DS
  • Recovery Image:
    Amon_RA

Posted 12 August 2010 - 08:37 PM

View PostGazorninplat, on 10 August 2010 - 07:21 PM, said:

For me, JIT has only been good for bumping up benchmarks.  I have never had any real use advantages with JIT.  Games ran just as slow, some apps became incompatible.  Perhaps for a higher spec machine JIT might make a difference (I have a sapphire).  I have found BFS kernels make a big difference in real world use performance, but doesn't give the same benchmark candy JIT does.

Ok. I believe it's better to use a Class 6 microsd than use jit.

#20 Von Rottes

Von Rottes
  • Members
  • 140 posts
  • Twitter:PLURked
  • Country:
    us - United States
  • Location:
    Columbus
  • Device Model:
    Mytouch 3G 1.2
  • CM Version:
    6.1 RC1
  • Recovery Image:
    Amon_RA

Posted 12 August 2010 - 11:37 PM

5.0.8+JIT+576Mhz=3.92MFLOP
110MB Ram free, No swap, No Compcache, A2SD(ADW+theme internal)
Before JIT i was getting 2.84MFLOP with about 120-122MB RAM