AKU 3.3 vs 3.5 Speed with .net CF 2.0
 

After adding .net CF 2.0 to my 3.5 build (actually enabling the correct reg keys to get it working) I noticed a good speed drop. Although I saw this with 3.3 (Cf seems to slow it to a degree). The speed difference between 3.3 and 3.5 seems to be noticable.

is anyone else seeing this? I was not sure if it's just that I have loaded so many different roms (2 build or more a day). I was thinking of maybe reverting back to 3.3 as there's not that much of a change. Speed and CF 2 seems the most important factor to me if all else are equal.

Just looking for others observations and ideas.

-saridnour

I will admit that I have noticed it as well. Last night I flashed to you 3.5+ ROM (first revision) and had to install .net CF2.0 manually. Noticed that, even overclocked to 624 MHz and with regediting glyphecache and such, that it is still a bit sluggish going from portrait to landscape mode.

I do have a lot of crap on here, though, so that could be part of it. In defense of your ROM, I have not yet crashed - when I was running colonel's 3.3 beta 2, I would crash probably once a day. So here's keeping my fingers crossed.

I don't really mind the sluggishness all that much. I am very satisfied so far. I am really really loving Internet Sharing - it is so easy to tether now.

Real quick, I don't want to get too far off topic, but in the second revision of the 3.5+, did you include the WM6 ringtones and alarms? They were not included in the first revision I downloaded. Otherwise, great job dude. Thanks.

PhilR8:The ringtones and alarms should be the same from 3.3 Ill do a dif tonight to make sure. Honestly, I can say for sure as I get what is what so mixed after so many spins of the rom :) Ill look later and correct if that is the case.

I may revert back to 3.3 as my Os of choice and just include the internet sharing app and keys. Other than that and the UI changes there is not that much difference... other than speed. The more I can underclock and keep the speed respectable the better for the battery life. Thats my trade off for direct push at least :)

-saridnour

A little off topic hear, but Saridnour,you keep mentioning using cpuxscaler to underclock to save battery...but with the abcPowerMeter program it indicates running XScaler at all, even underclocking, draws nearly 2 to 3 times the current than without running it. ??? I have a xv6700 with Helmi's AKU 3.3. I too used to xscaler unit I did saw those initial results. Now I don't run it and the battery life has indeed improved. Maybe I have a different setup, but I'd check out abcpowerMeter if you haven't already.

And just to stay off topic, Thanks for all the work you've done as I plan on installing your rom this weekend.

Dunno if this would help, but when you put .net cf2 on there, you should remove the 1.0 versions of the equivalent files as well as the reg entries.

Haven't seen anything yet. However, I change roms in less than a day, so I am not the best test subject. Once mine is out, we should get more feedback.

I stay away from .net cf 2 as much as possible because it slows my 6700 down so much. which also means I have to stay away from really good apps that require cf2. to me the loss in performance is not acceptable. plus its just huge and takes up a ton of space

Ok got one to try. I did a bulk edit of some changes and looks to have sped mine up a bit. trying to narrow it down but I did notice this one variance betweeen my 3.3 and 3.5 builds.

[HKEY_LOCAL_MACHINE\System\StorageManager\Filters\f sreplxfilt]
"ReplStoreCacheSize"=dword:00004096

3.5's default is 0.

This key would affect speed to a degree, but now if it just makes a visable difference by itself. I am not able to hard reset right now and I have some other modified values in my reg for testing along with this one. Just want to get it out there for others to try and give some feedbak.
Mainly:
Do you see a speed change (menu drop down, rotate, etc)

Make sure to soft reset after the change, also make sure to note the speed of thing before re-setting :)
Post back your findings.

-saridnour

As a note, mine was set at 4096. Perhaps that is why I wasn't seeing any differences.

colonel: : That is interesting. Did you apply another "hack" from what the kitchen produced? The 0 value is what it (the kitchen) outputted by default. I did not add any speed "hacks" to the PR1 as I waned to see how it worked out of the box and knew some hacks could degrade it. My PR2 had some in place and I was testing more when I diffed my current settings to that of my last 3.3

That being said though, I still thin 3.3 is faster overall. Making some changes to 3.3 to add in internet sharing and some other tweaks to general system defaults and will see how it compares.

-saridnour