1500MHz lock?


43 posts   Page 1 of 2   1, 2
by Rsnubje » Wed Feb 19, 2014 11:29 am
Hi there,

Currently I'm working on overclocking the Raspberry Pi (B revision) and I managed to get it to 1500MHz with a voltmod which allows me to apply any voltage I want. Getting to 1500Mhz was pretty easy, but even if i go only 1MHz above 1500MHz, it will not boot into OS. I'm pretty sure the ARM processor can go faster so I think there must be a lock of some kind. Does anybody know if there is such a lock and maybe where I could look to unlock?

Thanks in advance.

Image
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by dom » Wed Feb 19, 2014 1:35 pm
What does:
vcgencmd measure_clock arm

report when you've booted at 1500MHz?
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5055
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge
by ulysess » Wed Feb 19, 2014 1:51 pm
Mis apuntes de...Raspberry Pi: http://misapuntesde.com
Google Docs RPi: http://goo.gl/Iwhbq
User avatar
Posts: 300
Joined: Thu Aug 02, 2012 6:35 am
Location: Spain
by Rsnubje » Wed Feb 19, 2014 3:14 pm
I will check the real frequency, altough I believe it does really run at 1500MHz. Might be worth to try another OS?
Just have to say this, I'm pretty much a noob when it comes to using Linux, so I'm still learning to get along with it.

I will report back when I have checked the real frequency.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by Jessie » Thu Feb 20, 2014 6:19 pm
User avatar
Posts: 1754
Joined: Fri Nov 04, 2011 7:40 pm
Location: C/S CO USA
by Rsnubje » Thu Mar 06, 2014 10:36 am
Is it possible to use processor.ignore_pcc=1 for this? If so, where should I put it on raspbian?
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by jamesh » Thu Mar 06, 2014 10:41 am
Rsnubje wrote:Is it possible to use processor.ignore_pcc=1 for this? If so, where should I put it on raspbian?


I don't believe so.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 16959
Joined: Sat Jul 30, 2011 7:41 pm
by GTR2Fan » Thu Mar 06, 2014 11:10 am
I stumbled across this very handy BASH script just a few minutes ago that you might find useful...

https://gist.github.com/ecampidoglio/5009512

It gives you a readout of multiple parameters in this format...

Image
Pi2B Mini-PC/Media Centre: ARM=1GHz (+3), Core=500MHz, v3d=500MHz, h264=333MHz, RAM=DDR2-1200 (+6/+4/+4+schmoo). Sandisk Ultra HC-I 32GB microSD card on '50=100' OCed slot (42MB/s read) running Raspbian/KODI16, Seagate 3.5" 1.5TB HDD mass storage.
User avatar
Posts: 1601
Joined: Sun Feb 23, 2014 9:20 pm
Location: South East UK
by GTR2Fan » Thu Mar 06, 2014 12:31 pm
I don't know if this still applies with your custom volt mod, but...
current_limit_override Disables SMPS current limit protection when set to "0x5A000020". Can help if you are currently hitting a reboot failure when overclocking too high. May set warrany bit.

http://elinux.org/RPi_config.txt#Overclocking_options
Pi2B Mini-PC/Media Centre: ARM=1GHz (+3), Core=500MHz, v3d=500MHz, h264=333MHz, RAM=DDR2-1200 (+6/+4/+4+schmoo). Sandisk Ultra HC-I 32GB microSD card on '50=100' OCed slot (42MB/s read) running Raspbian/KODI16, Seagate 3.5" 1.5TB HDD mass storage.
User avatar
Posts: 1601
Joined: Sun Feb 23, 2014 9:20 pm
Location: South East UK
by Rsnubje » Thu Mar 06, 2014 1:05 pm
So that should be "current_limit_override=0x5A000020" right? Haven't tried this yet. I've seen some stories about bios_limit but i don't think this applies for the pi.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by shuckle » Thu Mar 06, 2014 1:36 pm
viewtopic.php?f=28&t=44904

"Latest rpi-udpate firmware now has current_limit_override enabled by default"
And that was a long time ago.
Posts: 565
Joined: Sun Aug 26, 2012 11:49 am
Location: Finland
by GTR2Fan » Thu Mar 06, 2014 1:53 pm
In that case, can someone please update the info on that page to reflect this? :)
Pi2B Mini-PC/Media Centre: ARM=1GHz (+3), Core=500MHz, v3d=500MHz, h264=333MHz, RAM=DDR2-1200 (+6/+4/+4+schmoo). Sandisk Ultra HC-I 32GB microSD card on '50=100' OCed slot (42MB/s read) running Raspbian/KODI16, Seagate 3.5" 1.5TB HDD mass storage.
User avatar
Posts: 1601
Joined: Sun Feb 23, 2014 9:20 pm
Location: South East UK
by jamesh » Thu Mar 06, 2014 3:03 pm
GTR2Fan wrote:In that case, can someone please update the info on that page to reflect this? :)


It's a Wiki - can't you do it yourself?
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 16959
Joined: Sat Jul 30, 2011 7:41 pm
by GTR2Fan » Thu Mar 06, 2014 3:42 pm
I could do, but I thought it would look better in the page's edit history if it was a name more familiar to the Pi community, especially as that exact page is quoted as a source of reference in the last line of Raspbian's 'config.txt' file. :)
Pi2B Mini-PC/Media Centre: ARM=1GHz (+3), Core=500MHz, v3d=500MHz, h264=333MHz, RAM=DDR2-1200 (+6/+4/+4+schmoo). Sandisk Ultra HC-I 32GB microSD card on '50=100' OCed slot (42MB/s read) running Raspbian/KODI16, Seagate 3.5" 1.5TB HDD mass storage.
User avatar
Posts: 1601
Joined: Sun Feb 23, 2014 9:20 pm
Location: South East UK
by bryanandaimee » Tue Aug 05, 2014 10:29 pm
Any idea if this voltage mod is legitimate? Looks like the guy is a writer for an online tech news site so I'm inclined to give him the benefit of the doubt. Also it looks like he did some benchmarking.

http://us.hardware.info/news/38967/hardwareinfo-pro-oc-overclocks-raspberry-pi-to-15-ghz
http://hwbot.org/submission/2503213_rsnubje_hwbot_prime_broadcom_bcm2835_795.18_pps

From the second link:
"rsnubje says:
155 days ago – Power is given on Varm readout point. L5 resistor removed. "

Does that sound like a possible way to increase the Vcore? I'm tempted to try it. (Yes I know it will void the warranty and likely let out the magic smoke.)

I'd vote for increasing the 1500 MHz limit if it's possible. I'd like to see if he can push it any farther.
Posts: 31
Joined: Fri Sep 14, 2012 4:32 pm
by Rsnubje » Wed Aug 06, 2014 6:55 am
Hi,

Since I'm one of the few that actually tried this mod, let me help you with this. See below images(click to enlarge).
I highlighted the areas that you need. For your external supply you can use the blue highlighted ground.
Just make sure it's a solid supply with not to big of a vdroop.

If you need more help, let me know.

Image
Image
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by rpdom » Wed Aug 06, 2014 8:03 am
bryanandaimee wrote:Any idea if this voltage mod is legitimate? Looks like the guy is a writer for an online tech news site so I'm inclined to give him the benefit of the doubt. Also it looks like he did some benchmarking.

"rsnubje says:
155 days ago – Power is given on Varm readout point. L5 resistor(sic) removed. "

Does that sound like a possible way to increase the Vcore? I'm tempted to try it. (Yes I know it will void the warranty and likely let out the magic smoke.)


Yes, it does look like a possible way of changing the core voltage. It is also very risky if you don't know what you are doing.

By removing the L5 inductor, you are disconnecting the VDD_CORE inputs from the VDD_OUTn, then by supplying external power across C16 you supply directly into the core. You'd need a very stable and accurate PSU, and I'd be concerned about losses and noise picked up from the wires leading to the supply - it isn't for nothing that part of the circuit is very small and close to the BCM2385.

The probability of blue smoke is high with this one :(

I wouldn't even attempt it.
User avatar
Posts: 10579
Joined: Sun May 06, 2012 5:17 am
Location: Essex, UK
by dom » Wed Aug 06, 2014 1:14 pm
I can confirm that this sort of mod is real. We did some experiments in the lab and found overclock increased by ~100MHz per 100mV.
I believe we tested to about 2.3V which is way out of spec and it didn't die. I wouldn't like to comment on a chip's lifespan at that voltage.

If someone does want to experiment with this then report what voltages you have tested, what the max stable overclock was, and what "vcgencmd measure_clock arm" reported.

It may be possible to increase the 1500MHz limit, but I'd like to sure exactly what limit is being hit.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5055
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge
by Rsnubje » Wed Aug 06, 2014 1:23 pm
When I was testing it, I hit the 1500MHz wall with just over 1.8v and was pretty stable at that point. The only thing that would make the system unstable were the RAM, Core and gpu clocks. This benchmark: http://hwbot.org/submission/2510494_rsn ... 819.02_pps checks the Arm freq like the way you mentioned, so the 1500MHz is real.

When you go for the external PSU mod, just be sure to keep the wires as thick and short as possible to keep the voltagedrop low.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by bryanandaimee » Wed Aug 06, 2014 4:00 pm
Wow Dom, that is a lot higher than I would have thought to go. With 100 MHz per 100mV, 1500 MHz doesn't sound too hard. That's what I see on mine too. Going from 1.2 to 1.4V, I get 950 to 1150 which is right at 100 MHz per 100mV. So on mine that would translate to over 1500 MHz at 1.8V just as reported.

Thanks for the pics Rsnubje! I have a couple B+ coming, (B+'s?) so I might have to give this a shot with my old model B. Any reason a battery or wall wart and an LM317 would fail misserably? I understand that there are much better ways to do this but I figure simple and cheap would be in keeping with the Pi ethos.

Bryan
Posts: 31
Joined: Fri Sep 14, 2012 4:32 pm
by Rsnubje » Wed Aug 06, 2014 4:05 pm
Might depend on what battery you intend to use. If it's some sort of LiPo battery it might work, because these have much more power than just an AA nimh cell or something alike. Just make sure the output is in the 1.0-2.0v range ;) But IMO it would be better to use a constant power source and not something that loses voltage over time as the power drains.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by dom » Wed Aug 06, 2014 5:41 pm
Here is a test build. It's hard to test (without external voltage), but it should go up to 2GHz. It is a fixed 256M/256M memory split.
https://dl.dropboxusercontent.com/u/366 ... t_2ghz.elf

I'd be interested to hear how far you can get.
I'm expecting either a report of "vcgencmd get_clock arm" returning 2GHz, or a photo of an exploded Pi!

(obvious disclaimer: this will void your warranty, and shouldn't be attempted by anyone who doesn't have a spare Pi on standby).
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5055
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge
by Rsnubje » Wed Aug 06, 2014 5:43 pm
Nice work. Will try this as soon as I find some time to do it.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by Rsnubje » Fri Aug 08, 2014 7:18 am
Maybe a stupid question, but how do I use your file? I'm failry new to linux to be honest.
Posts: 12
Joined: Wed Feb 19, 2014 11:24 am
by rpdom » Fri Aug 08, 2014 7:40 am
I'd say rename it "start.elf" and put it in /boot, after renaming the original file to something else like "start.old".
User avatar
Posts: 10579
Joined: Sun May 06, 2012 5:17 am
Location: Essex, UK