Is your CPU stable on OCCT's Linpack @ Max setting?

[pajama]

All my machines are linpack stable.

I always run OCCT with linpack on my systems, I personally feel if they pass then I have a stable O)C. Plus I know how high my temps will get under the worst conditions.

 

[Studabaker]

Tomorrow I'll build my E8400 in the same mobo as you have.

I love gigabyte motherboards

Me too, it's my first and the best experience I've had since my ages old Abit board. I like it more especially now that I feel like I've got a grip on how to best configure the board. I'll find out just how much I know the board when I try for 3.8 again. I'm going to try raising only the VTT as much as possible without raising the Vcore.

 

[mudkip]

Me too, it's my first and the best experience I've had since my ages old Abit board. I like it more especially now that I feel like I've got a grip on how to best configure the board. I'll find out just how much I know the board when I try for 3.8 again. I'm going to try raising only the VTT as much as possible without raising the Vcore.

I've never had C2D overclock experience. I went from a P4 to a Core i7...

So tomorrow I'll learn many new things

 

[IINexusII]

i got a e7300 at 3.6 stable using intelburntest, not the occt version of linpack. is it the same? 3.8 gives me a bsod after 10mins :S need to upgrade the P5Q se, northbridge cooler is shite

 

[Studabaker]

i got a e7300 at 3.6 stable using intelburntest, not the occt version of linpack. is it the same? 3.8 gives me a bsod after 10mins :S need to upgrade the P5Q se, northbridge cooler is shite

yeah linpack is intelburntest. you just need to tweak the settings a little bit. take a lesson from my screenshot, most stuff doesn't even need more voltage, one or two things could use a little bit less.

 

[dalekdukesboy]

Nice, I just woke up and my system is now Linpack stable for nearly 10 hours

whoa...I saw this and the message after it saying his pc rebooted after a few hours (assumably at least 3 + hours)...guys, I can honestly tell you that is excessive, you obviously are free to run your machines however you wish, but I'm just telling you if you're using max memory you probably don't need near that long for a solid overclock...of course I'm not using OCCT anymore, I found another program which rather than uses 90% free memory as occt does I found one that uses 100% and it shows you how many runs it's made as well as how long it's run for, about 30 runs is all you need to show very good stability, 40-50 means unless you run your whole rig in an oven @ 400 degrees 24/7 you're rock stable...for the speed I am running at 32 runs is usually 40-50 minutes, again this is 100% memory use not 90 so perhaps a couple hours at most is what I'd call sufficient...plus that's a heck of a lot of wear and tear on a system while overclocked with no beneficial use by you the owner. Again, just my opinion and how I look at it but I'd be surprised @ the guy who got a few hours with assumably no errors stable before rebooting if it can pass that I'd bet in normal use over even several hours I'd be surprised if it ever showed a hint of instability...linpack certainly is NOT normal usage it's truly a torture test.

 

[dalekdukesboy]

Nice, I just woke up and my system is now Linpack stable for nearly 10 hours

yeah linpack is intelburntest. you just need to tweak the settings a little bit. take a lesson from my screenshot, most stuff doesn't even need more voltage, one or two things could use a little bit less.

DEFINITELY! I figured this out when I got that run I posted earlier, I used to think I needed 1.256 volts for 4.0 ghz...after finding this program and hours of tweaking I use that voltage to get 4.3 ghz stable, this is definitely a program that forces you to explore your bios (albeit sometimes in desperation) to get an overclock to pass the test and usually some voltages go down, some up, but usually vcore in particular isn't as high as you think you need to crank it if you just figure out what the rest of your bios settings really do...

 

[IINexusII]

you just need to tweak the settings a little bit. take a lesson from my screenshot, most stuff doesn't even need more voltage, one or two things could use a little bit less.

heres my current settings a 3.6ghz

fsb termination: 1.3v
vcore: 1.3
northbridge: 1.3v the problem is that i cant read the temps on this mobo
ram v: 1.9 (OCZ Gold 4gb 800mhz)

 

[niko084]

Bringing up kinda an older thread here but anyone else notice linpack making their chip run at 100%ish for a bit and then back down and then back up?

Trying to see if my clock is perfectly stable, I'm having some goofy issues with Vista all of a sudden *clean install due to the issues also* and supreme commander, game locks up religiously after we get lots of units going, Vista boots in and doesn't load coretemp or rivatuner for a few minutes, shows my network connection as not connected but it is and it works, near no cpu usage and the rest of my taskbar comes up about 2 minutes later.

 

[farlex85]

Nope, sure doesn't. Doesn't even pass 10 minutes of Orthos. Runs cool though and I've never had a crash I didn't provoke (by stressing the hell out of my cpu for some odd reason) so I imagine I'm getting along.

 

[niko084]

Ya, one of my old machines would lock up on orthos at about 6-7 minutes and I could do anything on it for days except Supreme Commander, about 2-2.5 hours into a BIG game the game would crash.

 

[LagunaX]

I'll do Linpack if I ever move to watercooling.
Otherwise OCCT/Prime 95/Orthos blend is good enuf for me (don't need to see 70+ deg C)

 

[Bundy]

Bringing up kinda an older thread here but anyone else notice linpack making their chip run at 100%ish for a bit and then back down and then back up?

Trying to see if my clock is perfectly stable, I'm having some goofy issues with Vista all of a sudden *clean install due to the issues also* and supreme commander, game locks up religiously after we get lots of units going, Vista boots in and doesn't load coretemp or rivatuner for a few minutes, shows my network connection as not connected but it is and it works, near no cpu usage and the rest of my taskbar comes up about 2 minutes later.

yes it does go up and down a little. Have a read over the other thread at our clubhouse (in sig). For me, it took a lot of head scratching and a dose of humility to get mine stable. In simple terms, the OC I originally thought was stable, wasn't. My system has worked perfectly since I got the voltages right.

Oh and for the record - I found out what Studabaker discovered. More stability can often be gained from keeping the mobo voltages (and temps) as low as possible. Adding extra "for stability" can be counter productive.

 

[Mussels]

Ya, one of my old machines would lock up on orthos at about 6-7 minutes and I could do anything on it for days except Supreme Commander, about 2-2.5 hours into a BIG game the game would crash.

I had the same. sure it lasted 3 hours of supcom, but it would crash one game in 3 (almost always at a lan ) - linpack tests NB and ram too (which is why it ups and downs CPU usage, as it loads up a new set of ram data), in my case my ram was overheating.

I'll do Linpack if I ever move to watercooling.
Otherwise OCCT/Prime 95/Orthos blend is good enuf for me (don't need to see 70+ deg C)

if you dont want to see 70C load temps, lower your OC. that makes as much sense as saying you'll never game on your PC because it heats your video card up.

 

[LagunaX]

Nah, I hit 62 load with a TRUE and I do game. Linpack hits mid 70's quickly. Intel rates 45nm to 72 C. Don't need 45nm degradation just to prove my rig can be a toaster oven.
Crysis is the only game I've seen that might crash an overclock when prime/occt doesn't =)
It's not like I'm doing heavy duty folding or anything corporate...besides - what was the benchmark of stability for everyone b4 linpack? Prime 95/Orthos/OCCT.

 

[BababooeyHTJ]

I'm personally not a big fan of linpack.

With my old wolfdale I could pass 40 runs of Intel Burn Test and then go and fail Prime small fft in 45 minutes, usually just tweeking the GTLs would fix that.

I find a couple of custom Prime runs to be much more stressful on the MCH as well, while showing more errors. Once again I was IBT "stable" on my Q9650 and 8gb of Gskill pc8500 for about 40 runs and long runs of Memtest 86+, 3.8, and prime small fft and blend and failed this custom prime run in two minutes.

Linpack puts more of a load on your cpu than you will ever see in real world useage and the increaced vdroop that you will also never see from the increaced load can make your rig bsod or just fail. It is also more difficult to find out what is causing you instability and what you need to tweek with linpack than it is with prime.

 

[infrared]

I can pass it at 4.4GHz for an hour (no need to go any longer IMO).

I'm lazy and hate to wait for the results of a stability test, so I prefer 30mins to an hour of linpack over Prime95.

 

[Mussels]

Linpack hits mid 70's quickly. Intel rates 45nm to 72 C. Don't need 45nm degradation just to prove my rig can be a toaster oven.

since when has intel rated that? whoever told you that, made it up. ever heard of TJmax? that is the max temp for your chip, at which it throttles. heat DOES NOT kill intel chips, by design. (it can make them unstable, but it WONT damage them)

I'm personally not a big fan of linpack.

With my old wolfdale I could pass 40 runs of Intel Burn Test and then go and fail Prime small fft in 45 minutes, usually just tweeking the GTLs would fix that.

you're mistaking intel burn tests use of linpack for OCCT's use of linpack the OCCT one uses a lot more ram and therefore the NB as well. its a far better test.

 

[hat]

He said it failed the Prime small fft test which takes next to no ram at all

 

[Mussels]

He said it failed the Prime small fft test which takes next to no ram at all

he also said he was comparing to intel burn test, when this thread is about the OCCT version of linpack.

 

[LagunaX]

since when has intel rated that? whoever told you that, made it up. ever heard of TJmax? that is the max temp for your chip, at which it throttles. heat DOES NOT kill intel chips, by design. (it can make them unstable, but it WONT damage them)

Hey Mussels I respect your knowledge and experience but Linpack is an extreme stability test that 99% of users and "general" overclockers will never reproduce. Prime 95 blend/OCCT still remains a decent test of stability for everyday and occasional intense usage.

On another note, 45nm degradation does happen - I've done it before. The 2 culprits? Sustained VTT >1.5v and sustained insufficient cooling > 72 C under load. You won't notice any performance hit on stock settings but it'll drop your overclock 0.1-0.2ghz or require significant increase in Vcore. This accounts for a portion of the used 45nm chips on ebay: poor overclockers, abused ex-overclockers, and proven overclockers. On my old C0 chip the electromigration degradation did occur, to quote, "These processors are built on a new 45nm High-K process that invariably makes them predisposed to accelerated degradation when subjected to the same voltages used with last-generation's 65nm offerings."

Anandtech did a great job with their article "The Truth About Processor "Degradation" here:
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3251&p=6

OCCT Linpack is a definitive test though - but just not necessary for everyone

 

[Mussels]

let me put it to you another way, laguna.

The entire point of these tests it to make sure that it will never, ever crash. they are designed to be worst case. it might take you four hours of linpack to know your system is stable, or it might take you four weeks of bumming around in 2D using it for office tasks - the question is simple - Do you ever want your machine to crash on you, yes or no

 

[LagunaX]

Thread: "Is your CPU stable on OCCT's Linpack @ Max setting?"

Gotcha - Not going to run it so I don't belong in this thread =)

No more comments from me here...

 

[Mussels]

Thread: "Is your CPU stable on OCCT's Linpack @ Max setting?"

Gotcha - Not going to run it so I don't belong in this thread =)

No more comments from me here...

no one forces you to agree to our way of thinking
I dont think anyone minds people posting alternate views, so long as it doesnt derail the thread.

 

[Tau]

Hey Mussels I respect your knowledge and experience but Linpack is an extreme stability test that 99% of users and "general" overclockers will never reproduce. Prime 95 blend/OCCT still remains a decent test of stability for everyday and occasional intense usage.

On another note, 45nm degradation does happen - I've done it before. The 2 culprits? Sustained VTT >1.5v and sustained insufficient cooling > 72 C under load. You won't notice any performance hit on stock settings but it'll drop your overclock 0.1-0.2ghz or require significant increase in Vcore. This accounts for a portion of the used 45nm chips on ebay: poor overclockers, abused ex-overclockers, and proven overclockers. On my old C0 chip the electromigration degradation did occur, to quote, "These processors are built on a new 45nm High-K process that invariably makes them predisposed to accelerated degradation when subjected to the same voltages used with last-generation's 65nm offerings."

Anandtech did a great job with their article "The Truth About Processor "Degradation" here:
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3251&p=6

OCCT Linpack is a definitive test though - but just not necessary for everyone

775 Core based degredation is all based on VTT/PLL/GTT voltages being out of wack. NOT heat related. Considering that MOST 45nm start to throttle at 100*C, now also assuming that there is lets say a 40% fudge factor here over say 4-8 seconds. (What most ICs use for example, and still applies to CPUs) you would need to hold your 45nm CPU w/ tjmax100 @ ~130*C for 4-8 seconds to cause perminant damage. and considering how FAST the throttling actually kicks in you would be lucky to break 110, and definitly not for more than 4 seconds.

I have run plenty of CPUs are VERY high temps without any degredation, but the moment you load them with the wrong voltage/settings you can cause instant degredation. the socket939 AMDs were VERY bad for this (first onchip memory controllers) if you were even slightly off you would cause memorycontroller degredation nearly instantly.