Last Updated: Nov 05,2010
[KERNEL] Dirrk v01 Voodoo lagfix OC/UV - xda-developers
before (with 0.2 voodoo lagfix already installed)
Linpack 8.128 MFLOPS
after installing D01 1200LV:
Linpack 7.946 MFLOPS
Hot off the presses.
I am self-imposing a 24 hour waiting period on any flashing when something drops over at xda. I learned a valuable lesson with the 1st voodoo lag fix. Besides at the rate these things deploy, much like the OC Kernels last week, sometimes and again bless the dev's there are still tweaks not ironed out.
So while in practice I usually jump on a new flash immediately. Im now going to wait things out for a bit; just to be sure.
But don't let that stop any of you from taking the plunge of course.
i know what you mean, but i have been sitting on the sidelines the past week while all you guys were playing....i finally got CWM installed properly, went for the Voodoo lagfix, and then when this popped up i was like yeah...lets try it
Hmmm with the previous OC Kernels from last week my Linpack went from around 8.1-8.2 to 9.7-9.8 [literally the 20% increase in cpu speed]. The fact that yours didnt is suspect to me? Did you auto detect in SetCPU and slide the slider to 1200?
yeah, initially when i started SetCPU it showed 800Mhz max 100Mhz min, then after a second or two it showed 1200 max and 100 min.....might reboot and try again
For safety sake dont engage the set on boot check mark in SetCPU [its unchecked by default]. That does of course mean each and every re-boot you will have to auto detect and slide the sliders but thats a safety method. Because if you get crashing at 1200 you dont wanna be booting at 1200.
You dont need to reboot your phone per se. If its 1200 then its 1200. Usually the numbers reflect 1200 better though in Linpack. Another fun and easy test is the app called Speed Pi [free in Market]. If you take 6xx ms you're probably still at a Ghz if you are in the 5xx ms you're probably at 1.2Ghz
now im getting 6.4's in Linpack......rebooted ran it, got 6.4, said meh, flaked out...ran again....6.4.....ran Quadrant, got 1481....
looks like the clockspeed isn't sticking.....
thats the thing about the auto-detect.....after rebooting, first thing i did was go into SetCPU, and it showed 1200max and 100min, no autodetect was required.....set on boot has never been checked...
can't wait getting my Sf on wednesday
Hmmm, well maybe this new kernel boots at 1200 then. I know a lot of the OC Kernels from last week were booting at 800Mhz. 6.4 is more like an 800Mhz score anyway. For the hummingbird basically every 100Mhz should net you about 8mflops. So 800Mhz ~ 6.4Mflops. 1Ghz ~ 8mflops and 1.2Ghz ~ 9.6mflops. Sounds like you're at 800Mhz to me now. This could explain the Quadrant decrease as well
That will be good time because by then Im sure all these Kernels will be worked out. You definitely should have a working and stable 1.2Ghz voodoo kernel if not a 1.4
here are screens showing even after running Linpack and Quadrant the cpu spent nearly no time at 1200Mhz
You could try running SetCPU in performance. That locks it at the maxspeed. Its also a good stability tester.
And yeah those screenies show its really doing it 800. See if it even can lock in at 1200
Is this the Kernel you've been telling me about, Saps? How much space does its lagfix take up of your 2GB?
So this is voodoo and oc kernel rolled into one thing...correct?
Also jamesnmandy and actually everyone. While we all want to run the lowest possible voltage since it will increase battery life and reduce heat, the whole nine yards. Most of the problems people have are because of too little voltage. Now of course too much and you're fired. But when these kernels come out I always go for the standard voltage ones first because the likelihood is it will be more stable than a LV model. Now as always YMMV and one cpu might be able to run 1.2Ghz on 1200mv whereas another won't be able to run 1.2Ghz even on 1300mV. cpu's are individual creatures in that regard. Now sure the law of averages places many and most in the same ballpark, but there will be exceptions out of the standard range both high and low.
This is why I was so drawn to jt's original 1.2Ghz kernel he actually overvolted slighted [1325mV versus the stock 1300mV] and that made complete since to me and quite honestly it probably the exact reason it was so stable for me and the most stable for others. Even people who did have FC or freezing issues as it was better than some of the other ones out there.
If I were to grab an voodoo OC kernel right now it would be the SV one and not LV. Now admittedly, why would Dirrk release an unstable kernel? He would not...based on the testing he had available to him. We need to remember these incredible dev's dont actually hundreds and hundreds of SF's to test their kernels on. In fact, some dev's don't even have the devices they write for. [Not the case for Dirrk and jt]. So yes Dirrk's LV kernals might be stable on the SF's it was tested on but its hard to make blanket statements like this kernel is safe and stable for everyone; when you expand beyond the default.
Now Im on record stating I believe 1.6Ghz is possible with enough voltage. This seems confirmed by the fact Dirrk already floated out a 1.6Ghz model but took it back because the cpu was not holding the 1.6Ghz due to lack of voltage. Now of course you must tippy toe your way up the voltage ladder since as soon as you're too far you're fried. But if the stock cpu is 1Ghz@1300mV and I was stable as a rock at 1.2Ghz@1325mV from jt. I have to assume 1.4Ghz is attainable with maybe 1350mV or 1375mV; maybe even less. The Epic4G has a renowned OC kernel which is 1.2Ghz on 1250mV.
Samsung would not release a 1300mV stock kernal if there wasn't a bit of play-room left over. I have to assume 1400mV is a tolerable voltage. And if the same hummingbird is doing 1.2Ghz on 1250mV what is possible with 1400mV?
Maybe Im just too PC influenced by as others know you put that voltage up and then once you find or near the ceiling then you can start taper back. Often times its a compromise. But in all cases you must have adequate voltage for stability's sake.
I think Im rambling now so I'll get off the soapbox. But again for me Im going to run too much voltage rather than not enough. Especially when too much is still the factory stock voltage. It would be a different matter once we are over the stock voltage but outside of jt's 1.2 [and again that was 25mV] we arent there yet
read saps post below, sorry!
Yes . Less than 24 hours after jt got us voodoo. Dirrk gave us voodoo plus 1.2Ghz. As aforementioned it would not stun me if by the time you get your SF however, if the voodoo and/or the OC top speed is revised again.
This arena is literally changing by the hour. In fact last night it was voodoo .1 that came out only to be replaced an hour later by voodoo .2.
Im sticking to my guns and not flashing anything today. Call me a coat tail rider but after yesterday's fiasco Im letting others beta test for me for a change. [Although every ounce of my being believes in my heart that Dirrk 1.2 stock voltage is GTG]
No, no there's new kernels Dirrk put up an hour or so ago: [KERNEL] Dirrk v01 Voodoo lagfix OC/UV - xda-developers
In his own words: "All of these kernels have the voodoo fix"
ok cool thanks for clearing things up; both for D13 and me!
so the 1200sv is a bit better...it actually runs at 1200 sometimes...lol still want to try on demand scaling
I think the lagfix only takes like 200mb, to respond to SirKronan
I'm with this guy.
Be very careful if you do that. I have asked Dirrk to share his exact voltage profile the way jt did but he's likely so busy making these kernels he hasnt yet.
How are the numbers with 1200sv? Linpack quadrant et al.
Seriously I wish everyone would grab either Speed Pi or Benchmark Pi. These are nostalgic throwbacks to the days when SuperPi was actually something PC OC'ers benched with. Plus it really is a cpu tester with nothing from the I/O or Gpu
Sorry SK I missed this early. No I think when I first messaged you it was because we had actual OC kernals without the voodoo. A few days later we got voodoo and now we have them both...although the initial impression seems to be mixed and I for once am not being a guinea pig on this one
Separate names with a comma.