Simon
How do I find this out as its just the out of the box BIOS.
Also PLL voltage must not be changed, it causes issues.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
i7 2600k Sandybridge may be dieing...
- Thread starter Loko
- Start date
Simon
How do I find this out as its just the out of the box BIOS.
Also PLL voltage must not be changed, it causes issues.
Soldato
We are using BIOS 0402
Associate
If it's out of the box it will be 400 or something like that, which is not even on the Asus site! To check either boot into bios or run the Asus Update in windows.
The change in the 1053 beta bios is that they have added a PLL overvoltage option (AUTO, ENABLED, DISABLED) which is there to aid overclocking.
Associate
hmmm i read on other forums that chips just got all weird even without touching the oc tools at all, and that worries my a lot.
btw what's wrong with using turbo boost?
btw what's wrong with using turbo boost?
Its amazing how something can be taken from a handful of failures, how many actual people here have had a CPU die, I think 1 so far, other more like teething setup issues. We've sold over 500, then plus what competitors have sold I don't think Sandybridge has any issue its mainly down to teething issues with setups, nothing more.
Were working to try and find out what can causes such issues hence the high level of interaction from OcUK staff and our own in-house testing to try and find the do's and don'ts with Sandybridge.
Soldato
got to love ocuk feed back its exellent...
Associate
I think its called vccio on asus boards... Don't wuote me though. Maybe someone who has and asus can comment?
Associate
Reset my Asus P8P67 bios back to 1033 and tested the RAM using memtest 4.2 DOS Booted from CD.
Patriot 2400 2x2GB completed test at 2133 9,11,9,27 1T at 1.5v
OCZ 1600 low latency 2 x 2GB completed test at 1600 6,8,6,24 1T at 1.4v
For the record the OCZ's have just completed at 1866 6,8,6,24 1T at 1.575v (ah but failed 2nd pass)
Both sets of ram say 1.65v as their Standard.
Patriot 2400 2x2GB completed test at 2133 9,11,9,27 1T at 1.5v
OCZ 1600 low latency 2 x 2GB completed test at 1600 6,8,6,24 1T at 1.4v
For the record the OCZ's have just completed at 1866 6,8,6,24 1T at 1.575v (ah but failed 2nd pass)
Both sets of ram say 1.65v as their Standard.
Last edited:
Unlikely I think. This adds the option for internal PLL over voltage and is disabled by default. It's to help in extreme overclocking. In all my testing the 1053 bios has been the most stable.
Its amazing how something can be taken from a handful of failures, how many actual people here have had a CPU die, I think 1 so far, other more like teething setup issues. We've sold over 500, then plus what competitors have sold I don't think Sandybridge has any issue its mainly down to teething issues with setups, nothing more.
Were working to try and find out what can causes such issues hence the high level of interaction from OcUK staff and our own in-house testing to try and find the do's and don'ts with Sandybridge.
Fully agreed. Chinese wispers and forum'itis where people are suddenly experts on processor micro-architecture can yield a damaging image.
The initial impression from the events was that the CPU was degrading, as the crashes appeared more frequent, though this may have just been random. My concern was the multiplier not going to x38 with Turbo on under load. Perhaps the BIOS was playing safe after so many identified system crashes (overcloking failed report on POST - it think this as running forced RAM speed/timings).
Now I think it is clear that my problem certainly seems to be a memory compatibility issue / motherboard / BIOS issue. As has been stated, it's on the approved list and even states 1.66v.
At 2133MHz, standard timings and this voltage the system crashes randomly. Mostly what appears to be mundane tasks. Interestingly it can run Crysis at 4.8GHz and full memory speed for 1hr, yet jump out to desktop and crash whilst browsing the web. Madness.
Another point to note, the Asus P67 boards with EPU, when this is enabled, Windows 7 treats the PC as a laptop and will have various power management profiles. The default one is "Balanced Power" and this by default comes with go to standby after 30mins. I couldn't understand why the pc was off when I woke up in the morning after stability tests over the w/e at stock/auto settings!! It turns out that the BIOS doesn't pick up that 100% is not idle!!
Mine ran Prime overnight fine, still running now at 4.5GHz, RAM at 1600MHz 7 8 7 24. All voltages forced at stock bar Vcore at 1.26v (best guess, haven't tried lower yet). Temps stable at 58deg C.
Will check if it's still running ok when I get back from work. My CPU has been running at 4.8GHz fine for a good lenght of time during tests over the weekend in Crysis, so I think my issue is resolved.
Need an improved BIOS and tinkering with timings to get good memory bandwidth, running around 21Gb/s now. Was at 28Gb/s (if I recall right) when at 2133MHz.
Getting there. Thanks for the help all.
Good to know. Hopefully system will still be running when I get home. If so, can conclude that CPU is ok and it is a memory compatibility issue.
Now I know we are barking up the right tree, I will focus my investigations on memtests at different speeds, timings and voltages. Certainly RAM rated speed, voltages and timings aren't liked by the machine. Though it is strange, becuase I've had it running at rated RAM settings and 4.8GHz in Crysis fine! Then Prime kills it. Slightly reduced clock tighter timings seems the way forward.
Associate
This is probably because its the speedstep jumping the multi and voltage. In Prime or a Game like Crysis the CPU will spend all of it's time at the top multi / voltage where as doing mundane tasks like using the web will make it switch between spedstep modes.
Associate
Glad you got it working!!! Got mine back to 1.28vcore @ 4.4ghz running 2133 by upping the QPI. So I think it is just a few people running into issues with the higher rated ram and aupport for that...
This was something I suspected a couple of hours ago but couldn't recall what ram you were running. us poor folk who bought slow ram might be fine
HI there
Right memory at 1.65v seems absolutely fine, still not a single issue with our rig, still going strong in prime.
I think the issue is coming from anyone trying a combination of the following:-
1. High Vcore
2. High frequency memory
3. High Vdimm
I think when your trying to push its just too much for the CPU to deal with voltage wise.
These chips to get the best results shoud ideally be set at:-
1. 1.35-1.40v Vcore for maximum overclock and ideally 1.30v for 24/7 running.
2. Keep memory frequency ideally around 1600MHz and tighten the timings up.
3. Keep memory voltage sub 1.55v
4. Keep chipset/pll at default and BCLK at 100MHz.
I think the above setting will yield 4.6GHz-4.8Ghz happily but with complete stability and a system that run around the 75c mark.
Right memory at 1.65v seems absolutely fine, still not a single issue with our rig, still going strong in prime.
I think the issue is coming from anyone trying a combination of the following:-
1. High Vcore
2. High frequency memory
3. High Vdimm
I think when your trying to push its just too much for the CPU to deal with voltage wise.
These chips to get the best results shoud ideally be set at:-
1. 1.35-1.40v Vcore for maximum overclock and ideally 1.30v for 24/7 running.
2. Keep memory frequency ideally around 1600MHz and tighten the timings up.
3. Keep memory voltage sub 1.55v
4. Keep chipset/pll at default and BCLK at 100MHz.
I think the above setting will yield 4.6GHz-4.8Ghz happily but with complete stability and a system that run around the 75c mark.
Ok boys, just got back from work, was a long day.
PC still happily running Prime at 4.5GHz, thats 16 hours, satisfies my stability quibble.
Definitely agree on Speedstep, Turbo ratio and DRAM speeds not working right! Hopefully BIOS will sort that.
PC still happily running Prime at 4.5GHz, thats 16 hours, satisfies my stability quibble.
Definitely agree on Speedstep, Turbo ratio and DRAM speeds not working right! Hopefully BIOS will sort that.
Associate
Thinking of buying a Sandybridge i7 2600k with Asus P8P67 Pro or deluxe Motherboard.
Im going to leave it at stock though for the time being so would it be ok at stock/auto or should i still set the Back clock to 100 just incase, even though il be keeping it at stock?
Sure i saw on a post that memory works fine when left at stock/auto?
Thanks
Im going to leave it at stock though for the time being so would it be ok at stock/auto or should i still set the Back clock to 100 just incase, even though il be keeping it at stock?
Sure i saw on a post that memory works fine when left at stock/auto?
Thanks
Nope you need to define the values, there is a question mark over auto settings.