Hi
not too bad for first try
Hi
not too bad for first try
That is awesome :up: What batch is it?
You have low GFlops as for 5GHz+. I have 62GFlops+ with 4,5GHz.Quote:
Originally Posted by beast200
http://img202.imageshack.us/img202/8459/testy1load.jpg
Maybe yours CPU is throttling during the test. How did you set Short and Long Turbo Power limits as well as Core Current Limit?
http://img641.imageshack.us/img641/1...0211055641.png
BR,
Hi
you are correct
I have now set at 150 each
Ket can you advise setting for ram as i have the samebut get BSOD but v @ 1.56
is 1.6 safe with setup
cheers
Up to 1.65v will be safe. You should be able to get at least 1866MHz out of the kit, but it depends what ICs your kit is using I may of just lucked out with my kit being able to do 2133MHz+. If you need a little more detail about voltages Clicky :up:
ED - Oh, and about power draw. I'm not entirely certain if the mainboard just draws X wattage specified regardless of if the CPU actually needs to draw that amount of power, but I deffinately wouldn't set long/short power limits and current limit higher than is necessary. 250/250/250 should be plenty for 5GHz @ 1.45v. You could likely even use 200/200/200 without any problems.
Hi thats a bit better
I don't really understand the workings of the current limit settings to be honest but mine will run a stable 5ghz with 150/150/150 at 1.4V loaded. Can anyone describe what these long/short and current settings really do?:shrug:Quote:
Originally Posted by Ket
These are Turbo Power limits. Overclocking Sandy Bridge is about overclocking those Turbo states. It means how far you can go with Turbo on all 4 cores. So if you put higher limits, the less likely that it will try to throttle down. Very high limmits doesn't mean that CPU or mobo will consume more energy.
Basically, if you don't set long/short power draw high enough the CPU will throttle itself under load. The current setting specifies the maximum current the AI core has going through it at any given time. So far, my testing hasn't revealed it actually helps when OCing but my results aren't conclusive yet.Quote:
Originally Posted by ClockZ
Out of curiosity can somebody run IBT and let me know how hot your cores get? My 2500k seems to be a bit weird, cores 0&3 top out around 73c, but cores 1&2 top out around 83c. I looked at the IHS before I put the CPU in and it looked pretty flat so I'm just wondering if a 10c delta is common. Both Core Temp and Real Temp report the exact same. HSF is a lapped Hyper 212 Plus. I'm 99% certain the issue isn't with the HSF because I used one on a Phenom2 955BE that had more volts pumped through it than this 2500k and temps topped out at about 60c.
61, 66, 68, 67 - are what mine run at in IBT using AVX.Quote:
Originally Posted by Ket
Hi
Mine are 61 69 70 66
Damn. Looks like I'm going to have to lap the IHS. Compared to you guys it does very much seem like the IHS on my 2500k is pretty concave, which just sucks. Maybe I'll take the lazy mans route and buy a H50 instead and use one of my sickleflow fans with it.
////////For my reference, don't mind me/////////
4.6GHz - 1.35v actual [UEFI 1.36v]
4.8GHz - 1.38v actual [UEFI 1.405v]
5GHz - 1.4v actual [UEFI 1.425v], 1.43v [UEFI 1.45v], 1.45v [UEFI 1.48v] ALL FAIL
///////End////////////////////////////////////////
Based on those results it seems like 5GHz fails not because of a lack of voltage, but the concave IHS makes cores 1&2 overheat. With a flat IHS 4.6GHz stable is probably achieveable with ~1.32v actual. Thanks guys, your results are what made my numbers make sence :D
Quote:
Originally Posted by Ket
I've lapped every chip I've had the last few years. I haven't done this one yet. The contact looks good but the temps indicate mines not flat either. With 8 deg between cores I think mine is in need of a good lapping too.
No matter what im running mine seem's to have a 4to5c delta between all cores with a EK supreme hf. Much better than my i5 750 was.
watercooling,got a bad bord no dual mem run 2x slots is dead
here whit single mem. beta bios 1.33a
I7 2600K 5.120GHZ ATI 6870 1100/1250
http://img816.imageshack.us/img816/8325/screen014y.jpg
http://img696.imageshack.us/img696/4800/screen015y.jpg
so what is the max safe vcore without killing these things? 1.45v? 1.50v?
hmm ?
whit watercooling idel temp 29/32c
load team 50/54c
wprime 1024m
I7 2600K 1.535v 5.120GHZ
yea but at some point voltage is killer regardless of temps. Intel says max 1.52v. Is there any reason we should be worried about 1.4, 1.45, 1.50?? degradation? anyone have problems or kill one yet?Quote:
Originally Posted by Martin.v.r
Hard to say at this point, I myself am keeping mine 1.35ish. for 24/7 4.7/4.8 ghz is plenty. I'll let the hardcore guys test the water with high voltages for long periods and then i'll maybe turn it up some.
Max 24/7 use is supposed to be 1.4v from what most users say. I think I saw in another thread here that someone pointed out that Intel isnt talking about the actual vcore being 1.52 max usable on these processors but it is just a number in their table as they have done with other processors that goes well above what is the actual maximum
It would seem so. After watching temps for a few mins I found that the "worst" core on my 2500k is 12c hotter than the coolest core. Thats just crap manufacturing on intels part. I mean seriously... £180 and intel can't even put a even metal plate on the core? :down:Quote:
Originally Posted by dnottis
Somebody hasn't read Page 4, have they :p: I've heard about plenty 2600/2500k CPUs dying with 1.55-1.6v. For 24/7 use I would not go higher than 1.45v. Its common scence, 45nm CPUs intel said absolute safe max voltage is 1.45v but of course that envelope was pushed a bit - 1.475-1.5v, 32nm makes sence absolute safe max 24/7 voltage would be around 1.42v, but again pushing the envelope a bit around 1.46v shouldn't do any harm, assuming you are using a good cooler.Quote:
Originally Posted by dnottis
Well I've read it... but then I see all the screenies here of 1.6v lol.Quote:
Originally Posted by Ket
Yea I dropped back from 5.0 @ 1.44 to 4.8 @ 1.35 just makes sense, I dont want to kill it.
Yeah it amazes me that so many people actually think 1.5-1.6v is safe to run through their 2500/2600k CPU :rolleyes: 32nm is at the point now of being extremely delicate. 1.45v is a absolute max safe voltage beyond that you are going to start hurting these little CPUs :( Then when people do hurt their CPUs with 1.5-1.6v they cry foul, don't accept their own stupidity and want to RMA :slap: 22nm will be the same story, people will pile 1.5v+ through the poor little things and whine when the CPU blows. 22nm you can easily expect max safe voltage to be 1.38-1.4v. The question to be answered when a die shrink that small does come is can intel make those CPUs at least clock as far as the current 32nm CPUs? If not, the current CPUs will likely still be the better choice for OCing.
FYI folks, its already well noted that even water isn't needed to see the maximum of of these new CPUs, a quality air cooler with a lapped IHS will be able to get the same results. Gone are the days of high voltage and extreme cooling with these chips. There is a technical explanation for this, I forget the details but I shall divulge what I do remember, (and this has always been true of all CPUs) below a certain temperature you actually start to hurt the CPU die because the electrons can't flow as well and start to break down.
Oh, and dnottis, I'd try setting a 1866MHz divider for your memory, I'd be extremely surprised if your kit can't handle it these Asrock boards absolutely love G.Skill kits ;)
On the contrary, my tests give me more the impression that this settings are important. If too low, the CPU throttles so it's an important parameter IMHO :yepp:Quote:
Originally Posted by Ket
Hi Ket
now got mem @2133 @1.6 v
have you got your mem @ 1 or 2 T
are you cooling the mem with a memory cooler and does it need it
now i have got past 5 Ghz am lowering to 4.8 and see now low the voltage i can do
The options that prevent the CPU from throttling under load are the long and short power draw options, not the option letting you change how much is put through the AI core ;) In fact, I've run 4.8GHz with long and short power draw set to 200 and the standard 150 for the AI core, theres been no throttling at all. Nomatter how I use the AI core option, it just makes no difference. :shrug:Quote:
Originally Posted by Olivon
Currently I'm running 2133 @ 8-10-8-25 1T 10-80-7-7-7-25 rock solid stable @ 1.65v (UEFI set to 1.665v to allow for vdrop). Those settings have now almost passed the 1000% point in HCI Memtest :yepp: Whatever ICs are used on my Ripjaw kit are pretty damn awesome :up: I do wish Asrock would ad a extra memory divider or two though, I've seen fairly crap boards even offer up to a 2400 divider.Quote:
Originally Posted by beast200
I will do more tests regarding short period and long period but those settings are important with big overclockings for sure.Quote:
Originally Posted by Ket
Thank you Ket for your timings, seems to work well for me too :up:
I'm just glad Asrock listened to me about the memory compatibility of their boards. They are seriously cracking out some awesome numbers and while some of that will of course depend on what ICs the memory uses, a good chunk of it is also down to having a good memory table as well. Its amazing how these Ripjaw kits just keep letting you lower the timings like they do. I wonder if G.Skill actually fully knew just how good these ICs were/are, if they did, surely they would of released at least a 2000MHz CL8 kit.
FYI guys I've completed my testing of UEFI 1.42 vs UEFI 1.5 on the Extreme4 and here are the final conclusions;
These observations have been made comparing UEFI 1.42 to UEFI 1.5 on the Extreme4. Below are issues still present or added with UEFI 1.5 along with recommendation for future UEFIs for the Extreme4.
Legend:
Green - Minor issue
Orange - Elevated issue
Red - Serious issue
Purple - Removed options
Black - Recommendations
Still present in UEFI 1.5:
- Full screen logo [Disabled] option is not saved into profiles
- Using Clear CMOS button results in all saved profiles being wiped!
Added problems with UEFI 1.5:
- Adjusting BCLK frequency results in the board failing to POST
- Disabling Spread Spectrum results in the board failing to POST
- When installing a USB2 Quickcam Pro 4000 connected to the rear USB3 port(s) the system will BSOD, pointing to the etronxhci.sys driver.
Removed from UEFI 1.5:
- "Good Night LED" option
- "Onboard Debug Port LED" option
Recommended improvements for UEFI 1.6:
- Put "Good Night LED" and "Onboard Debug Port LED" options back in!!
- Add DRAM voltage monitor to H/W Monitor page in UEFI
- Increase Vcore range from 1.52v to 1.6v (Yes this is possible, UEFI 1.43A turns this up to 1.7v. It also introduces the bugs in UEFI 1.5 sadly)
- Fix outstanding bugs in UEFI 1.5 (of course)
I'm really hoping Newegg just replaces them for us, instead of us having to deal with ASROCK. I don't want to be down for 2 weeks. In fact I was just going to order another one from Newegg, then RMA this one and sell the replacement. Figured that way I'm not down at all.Quote:
Originally Posted by burebista
I'm wondering if I should bother RMAing my Extreme4 when the new boards are out at all. I tested SATA2 port3 and it passed without error and as we know, if one port is affected they all are. As mine passed, seems safe to say my board is fine. Can't be arsed with a week downtime for no reason.
Intel states a failure rate of 5-15% after 3 years. I don't see how you can say your board is safe after it has only been out for a couple months. I've got all of the SATA2 ports occupied and they work fine for now, but I'm definitely replacing the board as soon as the new ones are available.Quote:
Originally Posted by Ket
I state what I did because quite a few people around here found their SATA2 ports borked right out of the box, so the fact mine are fine.. bleh. I doubt there will be any problem with them in the forseeable future. Its also 5-15% during 3 years, not after. At least thats what the document read I was sent.
I'm 120% sure I won't own this in 3 years. LOL I upgrade my CPU / Mobo as frequently as Intel updates them, every year to 18 months. My 860 / p55 was about a year old. My Q9550 went 18 months. I don't even really consider this a problem, I'm using 4 of the SATA 3GB/Sec. 2 x Sammy F3s in Raid 0, a WD black 640gb for storage and DVD RW.
Same here. As much as I love my Extreme4 I doubt it'll make it much past May as I'll probably pick up a Z68.. if their any good.
Any news on potential new uefi versions?
The 1.52v vcore limit on 1.5 sucks :\
1.43A will give you a vcore limit of 1.7v, but be advised 1.43A also introduces all of the bugs 1.5 has. I'm sure Asrock will churn out more UEFIs, from their perspective though they have only been back at work since Monday because of the chinese new year in which they have 2 weeks off.
Ah, well ok, thought chinese new year was already last week :) My bad!
I've been to 1.43A before, but it would be nicer to just have a new version with all the goodies included!
Pretty good job on creating that list above, did you inform Asrock about it?
That's some serious issues. No problem with Extreme 6, hopefully.Quote:
Originally Posted by Ket
It's misleading. It actually generates dual boot on every power on.
Okay, now I understand, thanks radier :DQuote:
Originally Posted by radier
Yup i can confirm that. Same here with 1.43beta.Quote:
Originally Posted by radier
Yep I've already emailed Asrock and linked them to this thread. I'd expect a reply from them in a couple days after they manage to catch up on things.Quote:
Originally Posted by Cy4n1d3
Dual boot on power on issue I didn't have with 1.43A or 1.5. Its the only bug Asrock actually managed to fix but in the process introduced a crapload of others lol :D 1.42A I've found to be extremely good. Take away the dual boot issue at power on and it would be pretty perfect.Quote:
Originally Posted by gurneyhck
What is the current "best bios"?
1.43A or 1.50, depending on your liking.
help volt set to run 5.4 test,not to 24/7 I7 2600K L041B213 Asrock p67 Xtreame6
PLL
QPI/VTT
PCH
?turbo voltset ????????
link to 5.4GHZ superpi32m an 01
http://hwbot.org/community/submission/2125864_
http://hwbot.org/community/submission/2125861_
Again, kindly host just the link and not the file else there goes the accurate file count/feedback, pw is Mac... 1.51 with LED options. :)
http://www.ziddu.com/download/13822043/151EXT4.rar.html
I tried CL7 2133, no go Ket :(Quote:
Originally Posted by Ket
Ripjaw kit I have doesn't do CL7 2133 either, it does do 2133 @ 8-10-8-25 1.65v all day long though :yepp:
UEFI 1.42A hands down. By far has the fewest bugs.Quote:
Originally Posted by Bei Fei
Got a changelog for that? Know if Asrock fixed any of the issues I highlighted HERE? If they haven't its another lemon like 1.5.Quote:
Originally Posted by MacClipper
i have this same batch and i stuck at 5,2 for bench, what PLL/QPI/PCH you use for 5,4? My LC with this speed is to week:( i think to try o/c on SSQuote:
Originally Posted by Martin.v.r
Nice , thanks ket
Does anyone know how the following might work out:
http://www.abload.de/thumb/5000_1352_linxrgmw.png
now, the problem is - how the hell can I be stable 1 1/2h using newest linx with max mem but games like Black Ops crash right away after about 10 mins of gaming?
BSOD code is 124 always. I already tried fixed vcore up to 1.41v but to no avail :\
PCIe controller is integrated, so I thought raising vtt might help. I used up to 1.21v VTT, 1.86v PLL but Black Ops still crashes on me. I can't help but wonder if BO really needs over 0.06v additional vcore over linx/prime?
Any idea why this happens? System is below...
Reduce the intensity of LLC to level3, you may need to raise vcore slightly to compensate. As a rule of thumb, just because your system passes a stability test never just assume it is stable. Stability tests offer a very limited scope.
I don't get this - the graphs in the BIOS seem to indicate the voltage decreases as load increases - is this correct? Why would I want this to happen? Temps?Quote:
Originally Posted by Ket
Alright, I'm going to lower LLC to 3 (2 atm).
Question coming up is: if I raise vcore to compensate loss of vcore from LLC2 I'm going to be over 1.4v during idle.
I've read that higher idle vcore is not as harmful as vcore applied under load?
Is this (to some extent) correct?
Surely stress tests don't cover all aspects, I totally agree on that. But still i'm wondering why Furmark + Linx w/ max mem running at the same time don't cause any crashes but Black Ops (~40% cpu load, 80% gpu load) steadily kills my system within a few rounds of gaming if vcore is 'too low'.
(Well of course I also know that 5 ghz might not be needed 24/7 :))
This is by design to reduce wattage(temps) under load and protect the cpu from voltage spikes (changing from idle to load and vice versa).Quote:
Originally Posted by NicePants42
tech explanation:
http://www.thetechrepository.com/showthread.php?t=126
It would be nice if the 7seg dbug lcd would display the cpu temp after boot.
I am guessing that if you get the - after post that means the cpu is not able to boot with that multiplier.
With that multiplier or that vcore :)
Enable Internal PLL Overvoltage and increase vcore to test maximum boot.
I was 100% stable under the normal stress testing but also had to increase V/core for stability in Black ops and DirT 2, These chips got somthing a little different going on as far as stability than ive had in the past. Before if I was stable under LinX/P95 etc. I was usually stable in my games, not with this cpu though, they seem to need a little more core voltage for gaming. Really doesnt make much sence as the cpu usage in a game like Black Ops is not crazy high. Im game solid at 4800 with 1.35V load, plenty Im sure unless someone is using some very serious multi gpu setup.
SATA 3 ports will not recognize the WD 1TB..
Any ideas?
Quote:
Originally Posted by Cy4n1d3
same problem here, I documented in a thread.
4.8 Ghz @ 1.340v passed IBT with latest linpack (avx), prime 4 hours small FFT and Linx 20 passes. Crashed COD BO in mins. Needs 1.36v to be stable in games.
These chips are different than generations past. apparently they are easier to get stress stable and need more volts for apps.
Test prime95 with Blend. Linx and IBT was rockstable for me but prime in Blend crashes after couple of seconds !
Just arrived today:
G.Skill Ripjaws F3-12800CL9D-8GBRL 8GB (2x4GB) DDR3
Intel Core i7 2600K
ASRock P67 Extreme4 Motherboard
It looks like I'll be joining the discussion, when I have something worthwhile to say.
Alright, thank you guys :)
Started to think it's just me or something...
If you use Linx to test stability make sure you install win7 SP1, so Linx will use AVX. This way your Gflops will double and the stress on the cpu is higher as well (which is the most important obviously).
I'm waiting for the 22nd as it's coming out officially then :)
Hey guys I'm trying to get a system up for the kids and the WD 1TB just won't work on SATA3, you know the good ports, any ideas?
Make sure using the Intel SATA3, not the Marvell ones.
I'm using the Asrock P67 Pro3, it's only got two SATA3 ports (white) and four SATA2 ports (blue)
Is the hdd working on the sata2 ports?
Try another cable!
Yeah it doesQuote:
Originally Posted by Cy4n1d3
You mean another sata cable?
Jep, one of my cables didn't work for me..
Did you set SATA Controller in to AHCI Mode?
No worries, just fixed it by taking the DVD drive out then connecting the hard drive and putting the DVD back inQuote:
Originally Posted by Cy4n1d3
Bit quirky, it just wanted them connected in a certain order, both running on SATA3 now, thanks for the help.
Ah great :)
Had a similar problem, swapping around cables did the trick for me, might be about the order you're talking about!
FYI guys I had some movement from Asrock today, they are going to report my findings to the engineers to check out. Hopefully we will have a real nice polished UEFI for the Extreme4 soon :up:
That would be -SO- awesome!
One issue that nobody seems to have mentioned yet: Sleep mode doesn't work when Internal PLL Overvoltage is enabled. You will wake to a black screen.
Most P67 boards had this issue, but they have since been addressed by the other manufacturers through UEFI updates. The issue is still present on this board as of 1.50.
Thats probably because nobody on XS uses the sleep function lol :D I'll mention it to Asrock.
Think Pro 3 users will get an update too? I can't imagine the bios is much different between the Pro 3 and extreme 4.
It works fine for me. I am running an i5 2500k @ 4.8, (with PLL enabled) and it goes in and out of sleep mode fine. I have always found that if an overclock is not stable, the pc will wake to a black screen. Just my experience.Quote:
Originally Posted by Poohead
I7 2600K asrock Xtreame 6 bios 1.33a
losing performance Vantage 3Dmark in the CPU test ´5.1GHZ 32XXX 5.2GHZ 29XXX 5.3GHZ 28XXX 5.4GHZ 27XXX
everything is tried in the BIOS, can not fix it up, others who have tried the same thing?
bios 1.4 have the same error
Yours CPU is throttling. Set power limits to 255/255/300.
Hmm my OC is perfectly stable (Linx using AVX instructions on Win7 SP1 runs for hours, no issues with memtest, and gaming for hours never crashes either). I assumed the sleep issue was due to PLL enabled since the board sleeps fine at stock speed, but not when I overclock to 4.9 GHz (only requiring enabling PLL and bumping the CPU voltage in the UEFI). Perhaps a higher multiplier is the cause then.Quote:
Originally Posted by mean6
big thanks to radier ;):up:
Ojck I use watercooling
here is the bios set :rolleyes:
Hi Martin
what are the 2 voltages VCCSA & DDR
are you running @ 1.565 24/7 or just benching
which bios are you using the beta or 1.4
is it the same about the max 0.5 v rule between VVT & DDR volt as the i7 920 x58
Runing level1 LLC is a bad idea. It will most probably cause random 101 BSODs, the CPU simply doesn't like it. Water or not, your CPU probably won't live very long with almost 1.6v either.Quote:
Originally Posted by Martin.v.r
Check out my voltages:
http://farm6.static.flickr.com/5258/...8033ebf1_b.jpg
Use offset voltage instead of fixed.
BTW use XMP Profiles for memory.
http://i.min.us/ilyKhO.BMP
http://i.min.us/ilyHZG.BMP
@Radier, nice dude! Impressive to say the least. Can you show me a cpu-z screenshot under full load?
Can anyone exsplain how the offset core voltage deal works? Been running fixed on mine but maybe this offset control is a better way:shrug: And for me level 2 LLC seems to be best, running anything lower LLC than 2 and you get big v core drop under load.....how would that work for anyone? I guess I need to spend more time with these new bios, too much old school in me:D
Has anybody done any stress testing using the offset vcore? I can't get any stability using the offset vcore, I've tried multiple configurations to no avail. My vcore drops too much and causes stress testing to fail. I need about 1.376v @ 4.8ghz but using the offset will have my vcore jumping around from 1.288v to 1.384v.
Here is what I have so far...Quote:
Originally Posted by Mysterfix
http://i304.photobucket.com/albums/n...Capture-11.jpg
http://i304.photobucket.com/albums/n...Capture2-2.jpg
http://i304.photobucket.com/albums/n...n/Capture4.jpg
http://i304.photobucket.com/albums/n...n/Capture3.jpg
http://i304.photobucket.com/albums/n...0219122209.jpg
http://i304.photobucket.com/albums/n...0219122228.jpg
http://i304.photobucket.com/albums/n...0219122239.jpg
http://i304.photobucket.com/albums/n...0219122247.jpg
http://i304.photobucket.com/albums/n...0219122330.jpg
I'm using Offset Voltage at the moment. 1.376 under heavy load, using LinX / Prime, also gaming works - and no BSOD during idle phases :)
Elloquin, your G-flops are low for 5ghz if im seeing right,should be around 70. I'll bet you have some core throttling going on. Mine was doing that aswell and needed more core voltage, I could see a core drop out and downclock by using T-monitor.
Edit, I see you have a 2600K and running HT,does that change the G-flops in linX? Maybe you are fine?
Sweet, thanks for the response guys. Elloquin, maybe the setting "additonal turbo voltage" is the key. I haven't tried that yet.
Edit: Ok guys I figured it out, bios 1.42a was the problem. I switched to 1.51 and now stress testing is running without a problem. 2 loops of LinX 20 runs each passed with flying colors. I still need to test with prime and games but it's looking good so far.