Run Your iPaq 1910 at 300MHz?

[Janak Parekh]

<div class='os_post_top_link'><a href='http://discussion.brighthand.com/showthread.php?threadid=69594' target='_blank'>http://discussion.brighthand.com/sh...?threadid=69594</a><br /><br /></div>There's been considerable buzz over on Brighthand as of late, where a programmer figured out how to control the XScale's CPU frequency on the iPaq 1910. It seems that Intel has released two XScale CPU's to the PDA market: one unit capable of running from 100MHz to 300MHz, and another unit capable of running from 100MHz to 400MHz. Both support 100MHz increments: think of it as a SpeedStep-like functionality on steroids (ergo the name XScale implying the ability to scale the speed up and down), with a machine instruction to switch speeds. Note that this is not overclocking the processor per se -- if anything, the 1910 seems to have been underclocked from the processor's maximum specs, perhaps for battery life or for other reasons HP didn't disclose.<br /><br />Just as interesting is the fact that the 300MHz-based Pocket PCs are apparently using the exact same chip, but clocked at 300MHz. This includes both the Dell Axim and the ViewSonic V35. This also leads more credence to Christian Torring's <a href="http://www.pocketpcthoughts.com/forums/viewtopic.php?t=7028">thread</a> where he thinks the V35 is the same unit as the Medion, Mitac, Packard Bell PocketGear (and possibly NEC MobilePro) units, with just the Viewsonic scaled up.<br /><br />You can get info from the link above as well as on <a href="http://discussion.brighthand.com/showthread.php?threadid=68127">this thread</a>. Note that: (a) as of this moment, this program is very prerelease--it only lets you scale up to 300MHz and not back down without a hard reset; (b) you're executing this program at your own risk, not ours. It might work perfectly, or you might void your warranty, or your screen might turn psychedelic colors, or your iPaq might sprout wings and fly away, for all I know. ;)

[icatar]

Sounds very cool. I hope the DMCA isn't invoked in this case....

[vincentsiaw]

anybody try this thing yet ?

[Sparkomatic]

Just out of curiousity... Back when I had a HP Jornada, I remember seeing an overclocking program made for iPAQs from Jimmy Software (I think). The last time I tried to track the program down, I couldn't find it. Does anyone know what happened to it? Does it exist anymore? Did it even work?

[Wilbert]

I posted on this earlier.... :?

It works great.

[moaske]

I doubt the usefullnes of this....cause all PPC's actually don't run any faster wether they have a SA1110 or a boosting 400Mhz PXA250.
Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:. I think that only than we will see a significant speed increase.

I use a FSC Pocket Loox with the fastest PXA250, and i DON'T notice ANY bl**dy performance difference with my 'old' Casio E200 that still has the 'old' ARM v4 instruction set in it's 'old' SA1110 processor. :(

[Janak Parekh]

I doubt the usefullnes of this....cause all PPC's actually don't run any faster wether they have a SA1110 or a boosting 400Mhz PXA250.
It does make a difference, relatively speaking. Read the links. As to the SA-vs-XScale discussion, we've covered that from every angle already. ;)

--janak

[Sparkomatic]

Good point on the Microsoft OS. I remember when I had a Palm and used an overclocking hack to boost the speed. It was a very noticible difference. I think that was one of the biggest disappointments when I moved over to a PPC. However, everything that the PPC offers made up for it :)

[Wilbert]

I have been using the mod since lastnight and it is a noticable difference!

Everything runs smoother from video, to apps. The benchmarks prove it.

Now when and IF MS does allow xscale in the next PPC, the 300MHZ 1910 will fly!!

[GO-TRIBE]

Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:.
Hey Microsoft is doing the right thing here. They said all along that PPC would be standardized on ARM4 allowing for processor competition. The problem here is that Intel produced a chip that has little performance gain when running standard ARM4 code. What Intel would like MS to do is make PPC an Intel only OS, and thereby destroy compatibility and competition. Can’t you see MS is supporting competition here (which lowers prices.). Eventually PPC will move to ARM5, but I hope it never becomes Intel only. :wink:

[jngold_me]

Wilbert,

What effect have you seen on battery life so far?
I will most likely go for this app when Immier is able to do on-the-fly clock switching.

I am feeling rather confident that this x-scalar app will be fine because of all that I have reading about the processors ability to "scale" to different processing speeds and that HP supposedly "scaled-down" the 1910 on purpose. If the V35 can do it, so can we! :)

[Jason Dunn]

Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:. I think that only than we will see a significant speed increase.

This has been beaten to death over and over.... :? It's the bus speed that is killing the XScale performance, not the OS. Intel is the one holding us back, not Microsoft.

[Pocket Lee]

hi.

theoratically, if i freeze the cpu, the cpu will run faster. can some one confirm that?

like put the pocket pc in to a ziplop and put the ziplop into the freeze for 1 hour. and take it out.
i am sure the cpu temperature wil be under 20, which mean it would run faster than in the normal temperatur.

right??? try it. and report it.

Lee

[Jonathan1]

Hey Microsoft is doing the right thing here. They said all along that PPC would be standardized on ARM4 allowing for processor competition. The problem here is that Intel produced a chip that has little performance gain when running standard ARM4 code. What Intel would like MS to do is make PPC an Intel only OS, and thereby destroy compatibility and competition. Can’t you see MS is supporting competition here (which lowers prices.). Eventually PPC will move to ARM5, but I hope it never becomes Intel only. :wink:


Now, a lot of people said you couldn't do video on a handheld device, and it was too hard, but with the combination of some great hardware -- thank you, Intel, for the StrongARM processor and in the future for the next scale(Read: X-Scale). I'm not pre-announcing anything. This is the StrongARM processor in the device, and the software from the Windows Media Team at Microsoft, we're able to provide video like this.
-Remarks by Steve Ballmer
Pocket PC 2002 Launch
Oct. 04, 2001

http://www.microsoft.com/presspass/exec/steve/2001/10-04pocketpc2002.asp

[jizmo]

Just out of curiousity... Back when I had a HP Jornada, I remember seeing an overclocking program made for iPAQs from Jimmy Software (I think). The last time I tried to track the program down, I couldn't find it. Does anyone know what happened to it? Does it exist anymore? Did it even work?

It's a tiny stand-alone program called ipaqclk, and on ARM processors it's a must at gaming. 206mhz StrongARM is already the fastest processor on the block, compared to 400mhz XScale, and overclocking it to perfectly safe 236mhz will give it quite a lead. That is, if you have batteries to spare.

/jizmo

[SiGen]

I doubt the usefullnes of this....cause all PPC's actually don't run any faster wether they have a SA1110 or a boosting 400Mhz PXA250.
:(

ahhaha, i doubt the usefullnes of your post then :wink:

a new thread has been openned where all users testing this application are giving the results. increase of 2-5 fps in movies is nothing for you? decrease the time while loading games is nothing for you? or this benchmarks are nothing for you?

http://discussion.brighthand.com/attachment.php?s=&postid=438256

[Jonathan1]

Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:. I think that only than we will see a significant speed increase.

This has been beaten to death over and over.... :? It's the bus speed that is killing the XScale performance, not the OS. Intel is the one holding us back, not Microsoft.

Jason its both. It may be the bus speed but how do you know that optimizing the OS for the X-Scale utilizing the Media Processing Technology (Guessing this is the rough =ent of MMX or SIMD) wouldn' help performance? You can blame Intel all you want but there is no one person who is responsible. It was the OEM's that used the chip, it was MS that knew about this CPU well in advance. It was Intel that implemented the bus speed. You want to point a finger you better have more then 2 hands.

[moaske]

History has taught us that M$ isn't always true to its customers, so i guess there's still a portion of truth in my remark that M$ should have optimized their code fort ARM5. (Thanks for backing me up Jonathan).

Anyway, i must say that the benchmarks look quite impressive with the speed-up of the iPaq. But as my PXA250 is already running at 400Mhz, i guess it wont help me a lot... :(

And as for the fingerpointing: i wish the OEM's, M$ and Intel (and whatever other party) stopped pointing fingers at each other and started working on some serious performance gains here... :evil:
I know the subject has been 'beaten to death', but if i buy a device that says 400Mhz on the box, i WANT TO HAVE 400 Mhz !!! This starts to stink like consumer-misdirection.....

My 2 cents...

[William]

Just out of curiousity... Back when I had a HP Jornada, I remember seeing an overclocking program made for iPAQs from Jimmy Software (I think). The last time I tried to track the program down, I couldn't find it. Does anyone know what happened to it? Does it exist anymore? Did it even work?
See here: http://www.jimmysoftware.com/trial.asp. And yes, it works.

[Jason Dunn]

Jason its both. It may be the bus speed but how do you know that optimizing the OS for the X-Scale utilizing the Media Processing Technology (Guessing this is the rough =ent of MMX or SIMD) wouldn' help performance? You can blame Intel all you want but there is no one person who is responsible. It was the OEM's that used the chip, it was MS that knew about this CPU well in advance. It was Intel that implemented the bus speed. You want to point a finger you better have more then 2 hands.

You'll get no argument from me that:

1) Microsoft bragged about XScale performance gains when they shouldn't have
2) The OEMs didn't push back on Intel for providing them a faulty design

90%+ of this falls back on Intel though. They should never have released a Pocket PC with a SLOWER (by 3mhz) bus speed than previous Pocket PCs.

Microsoft doesn't want to fragment their OS code base by optimizing for a specific processor from one vendor, and no amount of bitching and moaning will convince me that it wasn't the right thing to do. It was. Samsung is using their own ARM processor, and I'm sure there will be others.

When Intel came out with their hyper-threaded processors, should Microsoft have come up with a special version of Windows to take advantage of Intel's HT technology (which is different than normal multi-procs which Windows already supports)? No.

If Intel wants to introduce new products with great new technology, all the power to them - but they need to be compatible with what people are using. If the HT processors performed 30% worse than normal Intel processors, no one would buy them right? Well, that's basically what Intel with the XScale - it performs quite a bit worse in some scenarios than a plain-jane StrongARM, yet Intel still released the processor.

[FredMurphy]

It not uncommon for a manufacturer to produce two versions of the same product by "crippling" the lower spec one intentionally. If it's cheaper to produce 2000 of the higher spec devices (and "cripple" 1000 of them) rather than 1000 low spec and 1000 high spec then they will. Economies of scale mean this is feasible.

You could ask why they don't just sell 2000 of the high spec devices, but having different products can mean the higher spec device has a higher perceived value.

I'm not saying HP has done this, or even that this has happened in the PDA market at all, but it is possible that the 1910 has intentionally been underclocked.

[egads]

Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:. I think that only than we will see a significant speed increase.

This has been beaten to death over and over.... :? It's the bus speed that is killing the XScale performance, not the OS. Intel is the one holding us back, not Microsoft.

There is no reason that a PPC with a 300 to 400Hhz processor with a 60 or 100Hhz system bus should not fly. The screen/redraw times drive me nuts. I have a 400Mhz Axim that replaced a Casio EM500 with a 150Mhz MIP's. I see a minor speed increase, nothing really to hop up and down about. There are no harddisks to slow things down, every thing is solidstate memory with faster access times than harddrives. The only bottle neck I can see is the OS its self.

I love my X5, just wish it was 3 or 4 times faster than it is...

[mterlouw]

It not uncommon for a manufacturer to produce two versions of the same product by "crippling" the lower spec one intentionally. ... I'm not saying HP has done this, or even that this has happened in the PDA market at all, but it is possible that the 1910 has intentionally been underclocked.I agree this is probably what was done. To separate the 1910 from the 5450 they slowed the CPU and disabled serial I/O (which is provided on the XScale BTW, so it should have been trivial to provide it on the Sync port).

Also as mentioned in the original Thought I'm sure battery life played a role in the clocking decision as well.

[SiGen]

Anyway, i must say that the benchmarks look quite impressive with the speed-up of the iPaq. But as my PXA250 is already running at 400Mhz, i guess it wont help me a lot... :(


of course it will help you a lot. maybe not in speed, but what about battery? Immier has reported that PIMs, MP3, reading ebooks, etc. just need 100Mhz to run, and they run good enough. .85 volts at 100Mhz. if you got one of those 3000mAh batteries, u could get about 12 or more hours in mp3 playing (course, screen off) or with low battery life devices (like the iPAQ1910, known to have a very poor battery life) it will help a LOT. (aside the speed it can offer now).

[mterlouw]

u could get about 12 or more hours in mp3 playing (course, screen off) or with low battery life devices (like the iPAQ1910, known to have a very poor battery life) it will help a LOT.I thought it was the 5450 that has the poor batter life... :confused totally:

[T-Will]

Yeah, the 1910 actually has a battery life of almost 5 hours with the backlight on one notch from off.

[Deslock]

I've been running this utility all day; no problems so far. Well, I did take my HP1910 out of my pocket at one point and it had soft-reset itself. But I dunno if that had anything to do with running at 300 MHz (PPC does behave strangely now and then).

I expect battery life to be 20-25% less (depending on exact usage). Today I ran it at max brightness and it ran down to 35% in 1:39. The last time I ran at max brightness, it took 2:01 to get to 35%. But that's only after one trial (for purposes of comparison, when using it with brightness at 50%, I generally get ~ 4 hours at 200 MHz). Good thing the HP1910 has a swappable battery :-)

300 MHz improves video playback considerably. The unit also feels snappier overall. It's still quite sluggish compared to my Sony Clie T665C when switching between apps (even if they're already running on the HP1910) and when populating drop-down lists. But that's more a function of PPC and/or how the individual programs have been written.

[seanturner]

u could get about 12 or more hours in mp3 playing (course, screen off) or with low battery life devices (like the iPAQ1910, known to have a very poor battery life) it will help a LOT.I thought it was the 5450 that has the poor batter life... :confused totally:

well, here are my tests for the batter life of the 5455. Mind you however, that the backlight was on during the entire thing. It was meant to be more of an effect of wi-fi test than a battery test alone.

http://www.pdajunkie.net/wirelessinetspeedtest.htm

[seanturner]

hi.

theoratically, if i freeze the cpu, the cpu will run faster. can some one confirm that?

like put the pocket pc in to a ziplop and put the ziplop into the freeze for 1 hour. and take it out.
i am sure the cpu temperature wil be under 20, which mean it would run faster than in the normal temperatur.

right??? try it. and report it.

Lee

No. No effect. It would, in theory, allow you to overclock the processor more, however, there are currently no overclocking utilities for the XScale so it would have no effect other than condensation under your screen...

[seanturner]

Just out of curiousity... Back when I had a HP Jornada, I remember seeing an overclocking program made for iPAQs from Jimmy Software (I think). The last time I tried to track the program down, I couldn't find it. Does anyone know what happened to it? Does it exist anymore? Did it even work?

The utility does only work on strong arms though. I've installed it on my 5455 and a 3900 series and it has absolutely no effect.

[Terry]

Wings would be very cool. I wonder what an iPaq with wings would bring on eBay? 8)

[seanturner]

Wings would be very cool. I wonder what an iPaq with wings would bring on eBay? 8)

More than a kidney.

[Janak Parekh]

Wings would be very cool. I wonder what an iPaq with wings would bring on eBay? 8)
You're assuming, of course, it wouldn't fly out of an open window. :D

--janak

[juni]

The LOOX comes with a setting to run the processor at slower speed to preserve battery power. I assume this is the same thing that this program does - HP just conveniently "forgot" to implement it as a setting.

[GO-TRIBE]

Microsoft is the one who should solve the problem of slow PPC's by modifying their OS :evil:. I think that only than we will see a significant speed increase.

This has been beaten to death over and over.... :? It's the bus speed that is killing the XScale performance, not the OS. Intel is the one holding us back, not Microsoft.

There is no reason that a PPC with a 300 to 400Hhz processor with a 60 or 100Hhz system bus should not fly
Put simply, if your processor can process data at 300-400 mHz, but data and instructions can only be feed to it over a 60-100 mHz bus, your real effective speed is 60-100 mHz. The only times your processor can really take advantage of it's speed is when it has the data or instuctions cached (i.e. they do not need to be sent over the bus). So there is ample reason why a 300 to 400mHz processor with a 60 or 100mHz system bus should not fly.

[Sslixtis]

Put simply, if your processor can process data at 300-400 mHz, but data and instructions can only be feed to it over a 60-100 mHz bus, your real effective speed is 60-100 mHz. The only times your processor can really take advantage of it's speed is when it has the data or instuctions cached (i.e. they do not need to be sent over the bus). So there is ample reason why a 300 to 400mHz processor with a 60 or 100mHz system bus should not fly.

According to that logic, All PPCs run at the same speed, 206Mhz StrongARM and 200-400Mhz Xscale, all run at 60-100Mhz. :silly:

[Janak Parekh]

Put simply, if your processor can process data at 300-400 mHz, but data and instructions can only be feed to it over a 60-100 mHz bus, your real effective speed is 60-100 mHz.
Of course, it's not that simple :( On the desktop, we have situations where the processor greatly exceeds the speed of the bus. Of course, the bus speed is still much faster, and more importantly, the processor's cache is much, much larger.

--janak

[julrik]

Just as interesting is the fact that the 300MHz-based Pocket PCs are apparently using the exact same chip, but clocked at 300MHz. This includes both the Dell Axim and the ViewSonic V35. This also leads more credence to Christian Torring's thread (http://www.pocketpcthoughts.com/forums/viewtopic.php?t=7028) where he thinks the V35 is the same unit as the Medion, Mitac, Packard Bell PocketGear (and possibly NEC MobilePro) units, with just the Viewsonic scaled up.


I am fairly new to PPC, but am now runing this hack on the Medion mentioned by Christian, and it appears to work. However, I do not have any benchmarking software to be sure. Can anybody point me in the right direction and mention what figures I should be looking for to see if it is running at 300 MHz. ?

Thanks
Jacob