Mobility Today Podcast 111

We'll see what happens, but I do agree that HTC has a horrible reputation for getting it right on initial releases.

"That is true JN, but it does not make sense to me nor do I suspect to you either, that if they have more RAM available they would choose to leave the cache levels the same as they were set when there was less RAM available knowing full well that the system can now afford it.That is true JN, but it does not make sense to me nor do I suspect to you either, that if they have more RAM available they would choose to leave the cache levels the same as they were set when there was less RAM available knowing full well that the system can now afford it."

LOL! This IS, after all, HTC we're talkin' about. :) Like I said, I suspect all the hubbub about delaying of units might be that HTC is quickly making this change to improve performance before the rest of the world device leave the factory.

That is true JN, but it does not make sense to me nor do I suspect to you either, that if they have more RAM available they would choose to leave the cache levels the same as they were set when there was less RAM available knowing full well that the system can now afford it.

You know as well as I do that WM needs at least the 30 MB free program memory for numerous 3rd party apps to run without issues for long periods of time without a reset.

That is why I think that the Diamond Touch Pro will not have these issues out of the gate, because with the extra ~100 MB of RAM they simply can afford the higher cache levels without any concern for low RAM issues.

"Another good podcast. Is there a fund I can contribute to to help Jerry get a broadband Internet connection and leave dial up?"

LOL! In all seriousness, it's not my connection as my cable usually gives me 1.15 mb/s down and .06 mb/s upload. I think it's the crappy Linksys G router that causes problems. :)

Phys,

As evidenced by the various cache hacks and the improvement of TF3 with the hacks enabled, shows that (as we both suspected) the TF is accessing areas of memory that are reserved for system processes which does not in a direct way affect user RAM. However, by increasing cache, it does eat into the amount of free (usable) free RAM. So, in essence, even on a higher RAM device, TF would still have performance problems unless the appropriate cache levels were set.

I have a feeling the recent "delay" HTC is claiming is to modify all the devices with higher cache settings to improve TF performance.

Post your comment here.

JN, of course I am speculating here as I do not have a test unit nor anyone to walk through diagnostic process with. I mentioned that earlier.

The quoted amount of free RAM may be true, the question is how much memory is the TF using when it is the active application? Looking to see how much free memory there is minimizes the TF allowing it to page out some of the code so that you are not getting the accurate amount of memory being used by the application or that is free when the app is active.

If the cache was set too low that is an indication of the lack of RAM which would allow setting the cache to a higher amount without any adverse affects on usage.

Another good podcast. Is there a fund I can contribute to to help Jerry get a broadband Internet connection and leave dial up?

is the lag really killing you? can you live with it considering the additional features you are getting? i just saw a video review on the nokia N78 and was very impressed with a few of the added features like FM transmitter and carl zeiss 3.2 camera. why can't they start focusing on extra features instead of these phones re-hashing the same features over and over. i mean really, what new feature does it have over the att tilt? one year later...

Greetings from Sweden! I love the show and listen to all podcasts as I get them in my Newsbreak RSS reader on my Tytn with XDA-dev. hacked 6.1 rom.. ;-)

I'm very greatful for the indepth review of the Diamond, now I'm even more decided to get the iPhone 3G. Windows Mobile is simply not up to scratch these days. Microsoft has been sitting on it's big ass since first release of Pocket PC. I recall my first device, the original iPaq, was a quite responsive device. It's simply embarrasing to release a top of the line device like Diamod with a UI that is simply put as slow as watching paint dry...

Everything I have seen on the iPhone makes it a superior user experience, where you see that some one has actually thought about the users needs. Also it is super responsive, they way I sure want a device to be.

Windows Mobile is dead in the water long before they hatch version 7. You snooze you loose - even when competing with a 500 dollar device and no keyboard....

Keep up the good work guys!

Best regards,

Johan /Sweden

FYI...there has been some success and File system cache to 8mb and file system filter cache to 131072 to better TF3D performance.

But you are making the assumption that that there is a lack of RAM. As stated several times before, the Diamond had well over 30mb of RAM free, more than enough for any application.

That's true to some degree, but even in the case of a memory leak the more memory you have the more time you have until it becomes a problem. I do not think a memory leak is the issue here at all.

If the code is not optimized to use memory efficiently, it won't matter how much RAM a device has, it will still perform the same way.

Shoey,

It appears that the program is paging itself out of RAM and then back in, which is where I suspect the lag is comming from. The only way to dismiss that variable would be to determine the free RAM while the TF 3D is the active window on top and has reached a point of zippiness. My guess is that as soon as it is minimized to bring another app on top, it pages all or some of itself out of RAM. Then when you go back to it, it pages fully back into RAM, but if you access it continuously for a short time it stays resident in RAM which explains why after a few starts it came up quicker.

I believe that this can be determined from the process viewer which may have a memory usage vs time graph. That could be used to determine the change in memory usage over the time it takes to switch from TF 3D to another app and then back TF then back to the viewer graph.

It is also possible that the 3D component of the chipset uses alot of power, and may be paging that code from the TF out of RAM to keep the battery from draining to fast.

It could be a combination too. Again, I am speculating as I do not have a device to test with, nor someone to walk through some diagnostic procedures. However, I do not believe from what I have seen thus far, that I am far off the mark.

The Pro will definitely not suffer from either scenario regardless, as it has more battery capacity and more RAM.

Pyhsboy, the device with Touch Flo running has about 40 Megs of Program Memory free, this is a good amount of free ram on a WM device. David can confirm it by checking the free memory.

When you power off the device it goes into a kind of suspend mode, when you power it on it comes out, what is most likely happining is the Touch Flo program is probably pulling some stuff to see where it's at and that's the probably cause of the lag. If that's it, then we should see a new version that resolves this problem in the future and I'm sure with further tweaks to the code speeds things up as well.

There could be other tweaks, page pool, touch flo settings ect that could address this as well through community tweaks as is been the case with past devices where people want to squeeze every bit of performance possible out of their devices.

I agree that how the RAM is being used by TF is where the problem lies. I just think that the amount of RAM has something to do with why it is using RAM the way it is using it.

The Diamond Pro has almost 100 MB of more RAM (288 MB), 450 more mAh of battery capacity (1350 mAh), and higher capacity for storage memory, since it is expandable SDHC, than the regular Diamond.

I called it before when everyone was saying that the 128MB WM6 devices were not going to be any more stable than the 64MB WM6 devices. Wrong, out of the box the 128MB devices were more stable with the same OS but more RAM. I am putting it out there now, the Diamond Touch Pro will be the way to go, and will not suffer these issues. A lot of people are going to be happy with the Diamond Touch Pro, and judging from the timeline the US version may even come with the PIE update. ;)

I still am curious about the Android offering though, as well as the ability of the X1 to deliver on the promised goods.

I agree with jerry.

The Diamond has plenty of RAM and Dave had no applications running and well over 35-40 mb of free ram available. The TF problems are not RAM based in that perspective. Putting more RAM in is like putting more "chef's to bake the cake". No, I think the problem is HOW TF is using the RAM.

The Touch Pro has almost 100 MB more of RAM and I bet there is little to no lag issues on it.

Next week, iPhone show, I'm there!!!

Man you guys really hate Sprint...lol...I have Sprint currently and I'm only with them because no other carrier can match my service plan. I have 600 anytime minutes, unlimited mobile to mobile, american roaming feature, unlimited data, unlimited text, N/W @ 6pm, equipment insurance and my monthly bill is around $44 after taxes. But if it's true that Sprint will be getting the Xperia, then I might be able to give a review of that...lol

Anyways, great podcast...too bad I missed the live streaming.

GR8 podcast guys !