Any disadvantage using the 21 multi on a 920?

Associate
Joined
25 Jun 2008
Posts
637
I have a 920 + Asus P6T se.

Currently I am playing around with the best way to get 4ghz stable.

I have found that with turbo disabled and using the 21multi fixed with the speed step disabled I have it rock steady at 21 x 190 for 4ghz @ 1.23V.

This is perfectly stable on a titan fenrir.

However if I drop the multi down to 20 and the clock to 200 I cannot get it stable under air, it gets to the 90's after 15 mins and eventually crashes. Also takes more volts to do this, 1.275 instead of 1.23.

Currently I have benched 3dmark 06 and noticed next to no advantaged in not using the turbo multi. I am hoping I am not losing anything performance wise running the 21 multi?

Thanks for advice, this is sooooo different to the old cd2 stuff.
 
Associate
Joined
16 Sep 2009
Posts
144
Location
Milton Keynes, Bucks
for some reason some 920s dont like the 200x20 multiplier, mine doesn't..

I can run 191x21 all day long, but 200x20 just fails hard without a lot more volts.. i was tempted to try something silly like 267x15 but i wasnt sure it would be safe :)

200x19 is very safe and stable for 3.8ghz though :/
 
Caporegime
Joined
14 Dec 2005
Posts
28,071
Location
armoy, n. ireland
Im currently running 21x183 for 3.84ghz, turbo and ht enabled, if i7 is anything like s775, lowering the multi puts more strain on the memory controller (which is on the cpu on i7's) many of the i7 clocking guides ive read reccomend avoiding the 20x multi as its harder to get stable, i cant confirm this myself as i havent tried it yet.
 
Soldato
Joined
22 Dec 2008
Posts
10,370
Location
England
I am having considerably greater luck with x20 than x21. The system is completely unreasonable at 210 bsck with turbo on, with the processor peaking at 65 degrees and passing many runs of ibt it freezes randomly. Sometimes it takes 5 minutes, sometimes it takes 40 minutes. More vcore pushes it towards 40 minutes, but doesn't solve it. This includes freezes in the bios and in memtest.

I've had it do this at 200x21, though it is much less frequent. Without a reliable trigger I can't judge at what point stability vanishes. However I do know that turning turbo off gets rid of this infuriating effect.

I am thrilled to put more vcore through it to hold 200x20 rather than 191x21 since it gets rid of the system randomly locking up on me (though it didn't lock up at 191x21). Currently at a very, very stable 200x20, waiting on another 6gb of ram before further overclocking endeavours.

So I will say your milage may vary. Either this effect is unique to my system/misdiagnosed, or the only reason people are loving the x21 multiplier is because none of them are going over 200 bsck.

I would be thrilled to hear any and all theories on why my system is locking if it isn't turbo. The list of things which do not resolve it are vcore, qpi, pll, vdimm, timings, multipliers, power saving options. Clock skews are a remaining hope, but as turning turbo off resolves the issue this is the approach I will take for the time being,
 
Associate
Joined
11 Oct 2008
Posts
123
Location
North West
Not played with the x21 multi properly yet, figured 200x20 for the moment and for memory reasons but will be looking for 4.2 over weekend.. Will give 210x20 and 200x21 a shot to see the difference but reckon may get turbo throttling whilst trying at 4.2 with high temps. (Have a moded Asus bios which is available from the Xtreme Systems forum guys which eliminates the turbo throttle, so may end up using that !!)

That sounds odd JonJ678, must be a bit frustrating.. What mobo do you use and I take it your on watercooling ?
 
Last edited:
Soldato
Joined
19 May 2009
Posts
3,113
Location
Cannock
From what i've experienced with my own chip x19 or x21 are the multis to use, x20 doesn't seem to be as stable for me, however a few overclocking sites do recommend turning off the turbo. I've just left the turbo on at x21 though and used a lower bclk instead, perfectly stable here at 4Ghz with very respectable voltages.
 
Last edited:
Associate
Joined
25 May 2009
Posts
1,696
Location
Derbyshire
i've been stable at 200 x20 with turbo on using p06 bios (bloodrage). NOticed that as you approach and breach 200 more volts are required and stability is harder to attain. I've been in win at 210 x 21 but got rounding errors in ibt.

I'm out of touch having been back at stock for the last two weeks but doesn't increasing the bclk increase the qpi speed? IIRC past 200 takes it above the speed of the extremes so I'd presume your starting to get to the limits of the memory controller on the chip. Obviosly some are better than others.

I could be spouting utter b*****s mind!
 
Soldato
Joined
22 Dec 2008
Posts
10,370
Location
England
What mobo do you use and I take it your on watercooling ?
It's using the gigabyte UD5 and corsair ram, cpu is under water but board isn't. And yeah, frustrating is right. Can't wait to repeat the experience with 12gb, that'll be easier... :)

I don't think my problems were related to turbo throttling, though I didn't test with throttling enabled. Perhaps I should do, cheers.
 
Soldato
Joined
6 Dec 2008
Posts
2,692
Location
Burghead, Elgin
191x21 seems to be the simplest way to get 4GHz on these chips, but I personally prefer to do 4GHz without using Turbo (200x20), it feels as though I'm cheating when using Turbo Boost for the 21x multi!!

It took me a while to get 200x20 stable, and it took quite a lot of fiddling around with my voltages to get there as well..

If I want to go beyond 200 Bclk, I have to raise my Vcore and Vtt up by quite a bit for it to become stable.
 
Associate
Joined
11 Oct 2008
Posts
123
Location
North West
Still working at this but so far had a go over the weekend at getting to 4.2 with the x20 and x21 multi. As expected with the x21 multi it's a damn sight easier !! Whilst trying 20x210, I had to up the vcore to 1.4 in bios & 1.98 pll which just let me run the 32M speed test in wPrime (v1.55 with 8 threads set) and it would BSOD (101) on the 1024M stability test only 1% in !! The temps were rather high at idle and shot up to 71C during the 32M run. No need to try LinX as it's pretty obvious what was going to happen and I didn't fancy putting more volts through my chip.

With the x21 multi vcore has only needed to be 1.35v but this caused turbo throttling which meant I wasn't at a steady 4.2.. Am only on air cooling and understand that high temps and power draw will cause the Asus P6T Deluxe V2 to turbo throttle, whilst I understand what the turbo throttle is about and why it does what it does, from reading around it seems that only Asus have decided not to put the option in their bios to disable it.

There's a great thread over on the XtremeSystems forum about this and they have some Asus modded Asus BIOSes available plus the latest RealTemp has an i7 Turbo program with it that lets you monitor and see if throttling is happening.

So today I've crossflashed to the modded P6T Deluxe 0006 bios and now there's no more turbo throttle. Temp wise at 1.3500 vcore using LinX max memory and 10 passes my temps peaked at a scary 82C (in an Antec900 with side panel off and house was pretty cool) but average was 79C, can see I'm going to have to look at water cooling at some point.

Will run Prime95 overnight and see what happens.

Some settings from bios : 1.35vcore, 1.94pll, 1.35qpi/dram, 1.24ioh, 1.20ich, llc enable.





It's all good fun really and no doubt I'll end up crossflashing back to the normal 0610 Asus bios and sitting at 4.00 (20x200) untill I can get watercooled :p
 
Last edited:
Back
Top Bottom