pagepool

computercarl · #1 (**permalink**) 09-01-2007, 04:51 PM

http://forum.xda-developers.com/showthread.php?t=323538

some of us have tried it, and noticed some spee increaces... I personally tried the 12, and it seems to "scoll like butter" in the windows dir...

Can some others do a before test, then apply the 12 meg ver, and then retry? I just want to make sure that its not a "all in my head" thing.

*edit* wm6 only...

C

**schettj** · #2 (**permalink**) 09-01-2007, 05:38 PM

wow... now that is exciting news.

Couple of points/questions - what is the default page pool? Is 12 bigger? What was the memory size after? Will we be able to do this to a kitchen-based rom?

I'd like to understand the performance vs ram size/page pool tradeoffs!

computercarl · #3 (**permalink**) 09-01-2007, 05:43 PM

Quote:

Originally Posted by schettj

wow... now that is exciting news.

Couple of points/questions - what is the default page pool? Is 12 bigger? What was the memory size after? Will we be able to do this to a kitchen-based rom?

I'd like to understand the performance vs ram size/page pool tradeoffs!

Yeah, it looks like if it is indeed doing something, there is an example batch file that could be incorporated into the kitchen bat...

As to the rest of yer q's, I know nothing, but that thread is quite informative, im still reading it...

http://www.modaco.com/Paul-s-guide-t...i-t256280.html is also informative... I'm still reading...

looks like it all depends on the ROM having the PagePool value at 0x37855A. Official WM6 have it there, WM5 have it at 0x3EC8AE but with cooked ROMs one never knows...

pierocksmysocks · #4 (**permalink**) 09-01-2007, 06:03 PM

I'm still not sure if it actually did anything. I mean I did the hard reset after testing it, but I don't know if it survived the hard reset. I'll reflash my phone later tonight to try and see if I notice any changes...

By the way did Capt ever do the benchmarks...?

pierocksmysocks · #5 (**permalink**) 09-01-2007, 08:20 PM

Just thought I'd post and say I'm running benchmarks right now to see if it actually makes much of a difference. Will post back later this evening with the results

pierocksmysocks · #6 (**permalink**) 09-01-2007, 10:52 PM

Test #1: Apache running WM6, default settings
----------------------------------------------------------
Spb Benchmark Indices:
Spb Benchmark index 284.44
CPU index 1335.59
File system index 114.37
Graphics index 2154.17
ActiveSync index insufficient number of tests were done.
Platform index insufficient number of tests were done.

Main Test Results:
Write 1 MB file 788 ms 1299 KB/sec
Read 1 MB file 265 ms 3.77 MB/sec
Copy 1 MB file 801 ms 1279 KB/sec
Write 10 KB x 100 files 2999 ms 341 KB/sec
Read 10 KB x 100 files 605 ms 1.65 MB/sec
Copy 10 KB x 100 files 3015 ms 340 KB/sec
Directory list of 2000 files 2101 ms 0.952 thousands of files/sec
Internal database read 667 ms 1500 records/sec
Graphics test: DDB BitBlt 4.99 ms 200 frames/sec
Graphics test: DIB BitBlt 30.7 ms 32.6 frames/sec
Graphics test: GAPI BitBlt 3.8 ms 263 frames/sec
Pocket Word document open error
Pocket Internet Explorer HTML load 2992 ms 8.27 KB/sec
Pocket Internet Explorer JPEG load 2270 ms 112 KB/sec
File Explorer large folder list 4044 ms 495 files/sec
Compress 1 MB file using ZIP 4672 ms 217 KB/sec
Decompress 1024x768 JPEG file 875 ms 321 KB/sec
Arkaball frames per second 6.42 ms 156 frames/sec
CPU test: Whetstones MFLOPS 5210 ms 0.071 Mop/sec
CPU test: Whetstones MOPS 1599 ms 39.4 Mop/sec
CPU test: Whetstones MWIPS 10336 ms 4.84 Mop/sec
Memory test: copy 1 MB using memcpy 12 ms 83.2 MB/sec
----------------------------------------------------------

Test #2 Apache running WM6, pagepool at 12
----------------------------------------------------------
Spb Benchmark Indices:
Spb Benchmark index 309.14
CPU index 1298.76
File system index 126.4
Graphics index 2176.07
ActiveSync index insufficient number of tests were done.
Platform index insufficient number of tests were done.

Main Test Results:
Write 1 MB file 796 ms 1286 KB/sec
Read 1 MB file 255 ms 3.93 MB/sec
Copy 1 MB file 756 ms 1354 KB/sec
Write 10 KB x 100 files 2784 ms 368 KB/sec
Read 10 KB x 100 files 817 ms 1.22 MB/sec
Copy 10 KB x 100 files 2796 ms 366 KB/sec
Directory list of 2000 files 1766 ms 1.13 thousands of files/sec
Internal database read 708 ms 1412 records/sec
Graphics test: DDB BitBlt 4.89 ms 204 frames/sec
Graphics test: DIB BitBlt 30.6 ms 32.6 frames/sec
Graphics test: GAPI BitBlt 3.76 ms 266 frames/sec
Pocket Word document open error
Pocket Internet Explorer HTML load 3251 ms 7.61 KB/sec
Pocket Internet Explorer JPEG load 2482 ms 102 KB/sec
File Explorer large folder list 4371 ms 458 files/sec
Compress 1 MB file using ZIP 4595 ms 220 KB/sec
Decompress 1024x768 JPEG file 993 ms 283 KB/sec
Arkaball frames per second 6.43 ms 156 frames/sec
CPU test: Whetstones MFLOPS 5213 ms 0.071 Mop/sec
CPU test: Whetstones MOPS 1597 ms 39.5 Mop/sec
CPU test: Whetstones MWIPS 10325 ms 4.84 Mop/sec
Memory test: copy 1 MB using memcpy 11.1 ms 90.2 MB/sec

Conclusion: Do it. The benchmarks don't show a substantial difference, but the user experience is improved. Screens load faster, web pages navigate better, etc.

**Trident** · #7 (**permalink**) 09-02-2007, 12:43 AM

Quote:

Originally Posted by computercarl

http://forum.xda-developers.com/showthread.php?t=323538

some of us have tried it, and noticed some spee increaces... I personally tried the 12, and it seems to "scoll like butter" in the windows dir...

Can some others do a before test, then apply the 12 meg ver, and then retry? I just want to make sure that its not a "all in my head" thing.

*edit* wm6 only...

C

Can you explain what this does? The benchmark tests don't explain it very well to us laymen.
Also, I read the directions but still didn't really get what to do.
And the zip file provided wouldn't unzip. [figured this one out- used winRAR to decompress]
...
I tried it (why do I do this stuff without knowing what the heck I'm doing?].
I only had the balls to try it once- used the 12mb pool. No noticeable difference, but I haven't used it much.
We'll see during normal usage tomorrow what happens.
Good night.

pierocksmysocks · #8 (**permalink**) 09-02-2007, 11:09 AM

Quote:

Originally Posted by Trident

Can you explain what this does? The benchmark tests don't explain it very well to us laymen.
Also, I read the directions but still didn't really get what to do.
And the zip file provided wouldn't unzip. [figured this one out- used winRAR to decompress]
...
I tried it (why do I do this stuff without knowing what the heck I'm doing?].
I only had the balls to try it once- used the 12mb pool. No noticeable difference, but I haven't used it much.
We'll see during normal usage tomorrow what happens.
Good night.

Yeah the zipped file was actually a .rar file. Some bum renamed it or something to a .zip. It was giving me fits yesterday...I didn't understand how everyone else in the world could extract the files and not me heh.

Pagepool = specially reserved space for the system to gain more free RAM for usage instead.

If that makes sense.

**schettj** · #9 (**permalink**) 09-02-2007, 03:14 PM

So, you changed it to 12 from the default - was the default 8? or 10? So the free program memory is 4MB lower? Sounds like the mogul/6800 is using 12 (it's free program memory is about 2-4mb lower then a 6700)

I guess I was wondering about going the other way... whatever the default is, does lowering it hurt performance noticeably?

saqer · #10 (**permalink**) 09-02-2007, 03:14 PM

so if i understand correctly, install cab file first, extract contents of second file onto desktop, create backup, then run the pagepool .bat of my choice?

How does it change settings on my device from the desktop?