Amlajm Wrote:After reading your post tittle the old and very famous quote came into my mind that "Old is Gold". I think same thing is happens to you in real world therefore you like the old cpu as compare to modern one.
[insert appropriate lolwut image here]
I would try without the overclock. When you don't give enough voltage, it will still show up as 4.3GHz, but the performance just won't be there ...
That makes no sense. Voltage does not effect frequency or performance. It affects stability and heat.
Well I have seen a difference in performance in my case, jsut by adding +0.004V. I do have speedstep enabled though and I'm guessing it was because my "Turbo" voltage at 4.4GHz wasn't high enough.
248mm2 CPU is fatter than 160mm2 CPU, more at 11.
What does that have to do with his performance?
(04-19-2013, 10:47 AM)NaturalViolence Wrote: [ -> ]That makes no sense. Voltage does not effect frequency or performance. It affects stability and heat.
It's actually correct as over in the overclock.net forums a guy named Idontcare, who's done a great deal of significant work logging data on the new ivy bridge chips, has found that due to internal error handling Intel chips may appear stable on the outside but internally they're running into tons of errors and spitting out worse performance due to instability. They suggest if someone has an OC they think is stable to open up the Event Viewer and check for WHEA errors under System Events>Administrator. I was running 1 hour custom tests just fine no errors in Prime95 etc, but when I checked the Event Viewer I was finding tons of parity errors. These could potentially degrade performance and often do.
But yes the number one cause of performance degradation is a thermally throttled chip for sure.
(04-20-2013, 05:33 AM)DaRkL3AD3R Wrote: [ -> ] (04-19-2013, 10:47 AM)NaturalViolence Wrote: [ -> ]That makes no sense. Voltage does not effect frequency or performance. It affects stability and heat.
It's actually correct as over in the overclock.net forums a guy named Idontcare, who's done a great deal of significant work logging data on the new ivy bridge chips, has found that due to internal error handling Intel chips may appear stable on the outside but internally they're running into tons of errors and spitting out worse performance due to instability. They suggest if someone has an OC they think is stable to open up the Event Viewer and check for WHEA errors under System Events>Administrator. I was running 1 hour custom tests just fine no errors in Prime95 etc, but when I checked the Event Viewer I was finding tons of parity errors. These could potentially degrade performance and often do.
But yes the number one cause of performance degradation is a thermally throttled chip for sure.
Idontcare is also in the anandtech forums
Here's a case where one member on the forums was stable (Intel Burn Test, Prime) except for dolphin ->
link
Must be becasue Dolphin stresses mostly two cores and differents instructions sets, which in turn require more voltage at the same clock speed
(04-20-2013, 06:10 AM)davidthemaster30 Wrote: [ -> ]Here's a case where one member on the forums was stable (Intel Burn Test, Prime) except for dolphin -> link
I had a very similar experience with my 2500k. OCd to 4.5 with stock voltage and all stress tests indicated that it was stable. Dolphin would crash within a couple of min. Had to bump the vcore +.04 to get it Dolphin stable. I was really hoping to keep voltage stock but oh well my temps are still great and I'm not too worried about longevity. Ill likely upgrade on broadwell or at the latest skylake.
(04-20-2013, 05:33 AM)DaRkL3AD3R Wrote: [ -> ] (04-19-2013, 10:47 AM)NaturalViolence Wrote: [ -> ]That makes no sense. Voltage does not effect frequency or performance. It affects stability and heat.
It's actually correct as over in the overclock.net forums a guy named Idontcare, who's done a great deal of significant work logging data on the new ivy bridge chips, has found that due to internal error handling Intel chips may appear stable on the outside but internally they're running into tons of errors and spitting out worse performance due to instability. They suggest if someone has an OC they think is stable to open up the Event Viewer and check for WHEA errors under System Events>Administrator. I was running 1 hour custom tests just fine no errors in Prime95 etc, but when I checked the Event Viewer I was finding tons of parity errors. These could potentially degrade performance and often do.
But yes the number one cause of performance degradation is a thermally throttled chip for sure.
Could you link the thread that you're referring to?
That's very interesting. I had never considered that but it makes sense. It helps explain a lot of odd benchmark results and OCing experiences that I've seen with recent chips.
davidthemaster30 Wrote:Idontcare is also in the anandtech forums Big Grin
Here's a case where one member on the forums was stable (Intel Burn Test, Prime) except for dolphin -> link
Must be becasue Dolphin stresses mostly two cores and differents instructions sets, which in turn require more voltage at the same clock speed
Xalphenos Wrote:I had a very similar experience with my 2500k. OCd to 4.5 with stock voltage and all stress tests indicated that it was stable. Dolphin would crash within a couple of min. Had to bump the vcore +.04 to get it Dolphin stable. I was really hoping to keep voltage stock but oh well my temps are still great and I'm not too worried about longevity. Ill likely upgrade on broadwell or at the latest skylake.
That's because for some odd reason over the years OCers have begun spreading the idea that if your cpu is prime95 stable it must be stable under every scenario/workload. Which is of course ridiculous.