Reply
Thread Tools
Posts: 38 | Thanked: 2 times | Joined on Apr 2008 @ Colombia
#121
on an offtopic question.. the Nokia N95 and N82 also share the omap 2420 processor of the N800 and N810, and run @ 330, do you think they might be overclockable (as for hardware capabilities) as well??
 
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#122
Originally Posted by shockedder View Post
Incredible as it may seem , yes , I did read the *entire* thread .
My bet is that you missed something but I admit I'm too lazy to read whole thread again to be sure so let's just answer this again.

Originally Posted by shockedder View Post
And, have been scouring the net for more information on this issue.
As I've said, there's very little information of any use regarding this.
There is enough information
http://focus.ti.com/general/docs/wtb...emplateId=6123
clearly says "The OMAP2420 includes an integrated ARM1136 processor (330 MHz), a TI TMS320C55x™DSP (220 MHz)"

Originally Posted by shockedder View Post
If you have no useful additions Antiles, it might be more useful to refrain from posting rather than making snarky, uncalled for comments.
You also had no useful addition and still you posted your question ;-)

Anyway, it was said few times (here and elsewhere) that for N8x0 tablets Nokia gets 'speed sorted' chips that are able to run ARM CPU at 400MHz with one little complication - DSP must run at 133Mhz.

Similar situation is with 770, OMAP 1710 says 220 but we have 'overclockable' chips running at 252MHz (both ARM and DSP)
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.
 

The Following 4 Users Say Thank You to fanoush For This Useful Post:
Posts: 3 | Thanked: 2 times | Joined on May 2008
#123
Yes. I've found some more specs on the ARM 11 ( 1136 ) used by TI Omap 2420 . Here's a quote from the ARM site :
The ARM1136J-S core and the ARM1136JF-S core are targeted at a wide range of consumer, wireless and networking applications such as set-top boxes, digital cameras, 2.5 and 3G mobile phones, voice-over-IP (VoIP) infrastructure equipment, broadband modems, residential gateways, WLAN access points and security devices. The ARM11 microarchitecture features a high-performance pipeline and high-bandwidth memory system which, when combined with clock gating throughout, results in an extremely compact and power-efficient microprocessor core. With a typical operating frequency of over 533MHz in 0.13u process, these two new cores are the only licensable microprocessor intellectual property (IP) to deliver performance of over 600 Dhrystone MIPs (un-optimized) at under 200mW.
with some strategically placed bolds.
I'm going to see if I can digg some more up from their technical spec ( freely available ).
Now if only I could find those frequency multiplyers.
 
Benson's Avatar
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#124
Interesting; it's worthy of note that the 2420 seems to be 90nm process: http://www.semiconductor.com/resourc...05_11_03_1.asp
But, of course, that doesn't really tell us anything.
 
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#125
Originally Posted by shockedder View Post
Yes. I've found some more specs on the ARM 11 ( 1136 ) used by TI Omap 2420 . Here's a quote from the ARM site
Omap 2420 != ARM 1136

What ARM site tells you about clock speed, power requirements and manufacturing process is interesting but not directly applicable to Omap 2420. I'm not sure if this is clear (sorry if it is) but TI does not buy ready made ARM 1136 chips from ARM and somehow stick them into their OMAPs. They get licence for the design and incorporate it into design of their own chip with their own limitations, features and then manufacture it themselves. ARM site is good for telling you about instruction set or other generic ARM 1136 core features (cache size, instruction timings), but for other stuff (like clock speed) it is better to search TI.
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.
 

The Following 3 Users Say Thank You to fanoush For This Useful Post:
igor's Avatar
Posts: 198 | Thanked: 273 times | Joined on Jan 2006 @ Helsinki, Finland
#126
Originally Posted by shockedder View Post
Since a leap from 330Mhz to 400 was possible , I think that we are far from tapping out the chip's performance.
A hardware spec sheet for the chip would be great.
Cheers
We do safe overclocking as "supported by TI".
Nothing prevents you from going further (and potentially frying the omap) and that actually would be _real_ overclocking since it would really go beyond the specified limits.

Note that to do it properly, you need also Menelaus TRM, the power companion chip, for getting the accuracy and ranges of the voltage regulators.

And no, sorry, can't pass it over. Check with TI.

Depending on your luck and skill, you might actually get something more.

The way i would proceed is to select 1 OP, either OP1 or OP0, not change the ratio for dividers and instead bump up a little bit the DPLL ratio.

This will not give you radical changes, but maybe a still perceivable improvement.

I wouldn't touch the voltage setting and play it safe.

The only risk is that you will have to take your tablet to the service point for coldflashing, but at least it won't fry.

Finally, one last piece of advice: wait for Diablo code to be released - the final version. The WiMAX tablet uses a different SDRAM silicon (single die vs dual die) that required fixes in the calculation of memory bus timing.
This doesn't seem to affect older hw, but you never know. Overclocking might expose the bug on existing units as well. The bug was fixed after leaked code came out.
 

The Following 7 Users Say Thank You to igor For This Useful Post:
Posts: 20 | Thanked: 3 times | Joined on Sep 2007
#127
check this out:

http://linux.omap.com/pipermail/linu...ry/008895.html

very interesting...

Wish someone compiled preemptible kernel, with frequency scaling, and higher max_speed_hz...

Heck it looks like vcore voltages can be manipulated too - but does not even seem necessary...

netC

I only know enough to be dangerous...
 
Reply

Tags
omap2420

Thread Tools

 
Forum Jump


All times are GMT. The time now is 02:23.