Speedy Benchmark Result for PalmRevolt, SkinUI, mySkin and myKbd

[lhandra]

When we are using app that seem always running on front or background process, then no doubt that the processor speed will affected.

So I do some test to check my processor benchmark when using some app like Palm Revolt, SkinUI, myKbd and mySkin using Speedy v6.1 on my Tungsten T3.

The result is :
1. Standard T3 (after HardReset) -> 370 Mhz (can also give me 375 Mhz)
2. PalmRevolt v0.934b Only -> 349 Mhz
3. PalmRevolt v0.942b Only -> 300 Mhz
4. mySkin (last version) -> 349 Mhz
5. mySkin (last version) but you move the mySkin app into card or delete the app -> 370 Mhz
6. myKbd (last version) -> 349 Mhz
7. SkinUI -> 349 Mhz

For the correct result when using speedy you have to do a softreset on the softreset hole on your PDA rather than using softreset software

The point on this thread is if we wanna make our palm comfortable with skin or themes feature so that is the cost that we must have (the processor speed will decrease). But of course i dunno exactly if there is another result when we use another software benchmark beside speedy.

A small hint for the developer of PalmRevolt that maybe need to do some test why there is a different on processor speed when using PalmRevolt 0.934b and 0.942b

Another attachments can see on Post #2

[lhandra]

If we use the combination between those software like PalmRevolt vs MySkin or mySkin vs SkinUI the result is the slower speed, like you see on attachment PalmRevolt 0.942b vs mySkin coz 300 Mhz and SkinUI vs mySkin coz steady at 349 Mhz

[CroCo]

Nice tests! I was wondering how much processing power these "background" programs eat. Luckily running differnet programs together in the background does not seem to consume processing power additively- otherwhise with a few of these apps our PDAs would be back to PalmProfessional processor powers:-)

[pruss]

I wouldn't particularly worry about this. The only time the difference is going to matter is during period of intensive computing, e.g., when decoding video or searching through a large etext or rescaling a picture. However, during intensive computing, if the developer has done her job right, there shouldn't be many events occurring. Since it is only when events are occurring that myKbd/mySkin causes any slowdown, at such times, there should be no slowdown at all.

[lhandra]

Yes that is true alex.

hmm... about myKbd/mySkin
maybe a little question why the processor can be affected by speedy process benchmarking when using myKbd/mySkin ?
or why when i use the mySkin and change the DIA with the new one and then remove mySkin app with the DIA still the same, the processor speed is full. Like the test result 5th on Post #1

And then for latest PalmRevolt, u can see the value of decreasing speed when using speedy to do the benchmarking.

----------------------------------------------------------------------------
This is just for discussion, not for make a decision that this app is best or worse

[pruss]

Yes that is true alex.

hmm... about myKbd/mySkin
maybe a little question why the processor can be affected by speedy process benchmarking when using myKbd/mySkin ?

I don't know how speedy measures things.

or why when i use the mySkin and change the DIA with the new one and then remove mySkin app with the DIA still the same, the processor speed is full. Like the test result 5th on Post #1

mySkin has two aspects: it's a skin loader and then it has some resident functions. When you load an alternate skin, it costs the processor no extra time (or little extra time) to use the alternate skin instead of the built-in one. The OS doesn't really care which one is loaded. The resident functions, though, keep on waiting for certain events, e.g., to compensate for an OS bug that used to cause crashes in the clock popup.

[lhandra]

ic thanks for the answer alex
hmm i still waiting the answer from the developer from PalmRevolt about the big different speed between the older and the latest one

[pruss]

When we are using app that seem always running on front or background process, then no doubt that the processor speed will affected.

So I do some test to check my processor benchmark when using some app like Palm Revolt, SkinUI, myKbd and mySkin using Speedy v6.1 on my Tungsten T3.

It's worth noting that myKbd and mySkin are much faster on the T5/TX/LD than on the T3. That's because the resident code is ARM native on the T5/TX/LD.

Here's what I get on my TX:
With myKbd active: 278.
With myKbd inactive (OS skin, no home button, no RAM icon protection): 275
With myKbd uninstalled: 275

With FontSmoother on: 313 (huh? this I think shows that there is something off on the benchmark, because FS does not speed things up!)

[applejosh]

It's worth noting that myKbd and mySkin are much faster on the T5/TX/LD than on the T3. That's because the resident code is ARM native on the T5/TX/LD.

Just out of curiosity (and because I have no real knowledge of the inner PalmOS workings), why isn't the resident code ARM native on the T3? Is OS 5.2 that much different than Garnet (5.4)?

[pruss]

Just out of curiosity (and because I have no real knowledge of the inner PalmOS workings), why isn't the resident code ARM native on the T3? Is OS 5.2 that much different than Garnet (5.4)?

The DIA code seems to be different. Remember how the T3 uses these separately installed compatibility libraries.

[applejosh]

Ah yes, because they forgot to include the proper versions in the ROM or something to that effect. (Sometimes I really wonder what the programmers at Palm are thinking.)

[pruss]

A better test for code that mainly modifies event handling is Igor Nestorov's PalmInternals, by the way.

[lhandra]

cool alex, different speed on different hardware. like nearly i said before maybe it will be different if we use another benchmark tools or maybe another Hardware like you did.