Ivy Bridge OC Results Came in (Engineering Sample)

[Khato]

I'm now beginning to wonder whether the 'new manufacturing process' issue that was cited as a cause for the IVB delay may be related to these odd temperature results. Namely, not an issue with the 22nm process itself, but something with assembly/packaging.

Why? Take a look at the difference in power/temperature results between the overclocked 3570k and 3770k. Total system power consumption is only off by 1W idle, 4W load... but there's a 10C idle and 21C load temperature difference? It would have been nice if they'd provided power consumption for a CPU only load, but there definitely seems to be something off with that result. Yes, given same power draw IVB should run hotter than SNB due to the smaller die size, but the increased thermal conductivity should be somewhere between the process scaling (47%) and die size reduction compared to SNB (74%.) In the Tweaktown testing, comparing non-overclocked to overclocked power/temperature delta (assuming that the entire delta of power is due to CPU) the 2600k is at 2.8 W/C (2.8 watts causes a 1C temperature rise) while the 3770k is at 1.3 W/C and the 3570k is at 1.7 W/C. So compared to the 2600k, the 3770k is actually slightly worse than process scaling at 46%, whereas the 3570k is pretty much right in the middle of expectations at 60%. So maybe their 3770k sample is older and suffered from the 'manufacturing process' issue, while the 3570k is newer and fixed?

 

[Idontcare]

^ excellent post regarding the discrepancies in thermal dissipation between the two packages.

Also IDC lapping is good . BUT . If its true that intel will replace for free the 3700K 1 time if you destroy it . I would be sure that lapping doesn't remove the 1 free replacement offer.

Lapping does invalidate the performance replacement warranty, someone pointed me to the fine print before when I mentioned having lapped my 2600K

But really you can't expect Intel, or anyone, to be willing to take on the open-ended risk of covering physically augmented CPU's for warranty purposes. That's a fool's errand for the manufacturer.

 

[Nemesis 1]

Looks like they were not kidding you. Read his review, he says he got 5.2 on his 2600k, and couldn't get higher than 4.8 on his 3570k and 3770k due to heat on air.

http://cdn5.tweaktown.com/content/4..._with_the_core_i7_3770k_and_core_i5_3570k.png

http://www.tweaktown.com/articles/4...e_i7_3770k_and_core_i5_3570k_cpus/index8.html

That review really says nothing as the voltage shown true or not is rather good for 4.8 ghz. If thats the correct voltage and it should be as its using a Z77 M/B

 

[Nemesis 1]

I'm now beginning to wonder whether the 'new manufacturing process' issue that was cited as a cause for the IVB delay may be related to these odd temperature results. Namely, not an issue with the 22nm process itself, but something with assembly/packaging.

Why? Take a look at the difference in power/temperature results between the overclocked 3570k and 3770k. Total system power consumption is only off by 1W idle, 4W load... but there's a 10C idle and 21C load temperature difference? It would have been nice if they'd provided power consumption for a CPU only load, but there definitely seems to be something off with that result. Yes, given same power draw IVB should run hotter than SNB due to the smaller die size, but the increased thermal conductivity should be somewhere between the process scaling (47%) and die size reduction compared to SNB (74%.) In the Tweaktown testing, comparing non-overclocked to overclocked power/temperature delta (assuming that the entire delta of power is due to CPU) the 2600k is at 2.8 W/C (2.8 watts causes a 1C temperature rise) while the 3770k is at 1.3 W/C and the 3570k is at 1.7 W/C. So compared to the 2600k, the 3770k is actually slightly worse than process scaling at 46%, whereas the 3570k is pretty much right in the middle of expectations at 60%. So maybe their 3770k sample is older and suffered from the 'manufacturing process' issue, while the 3570k is newer and fixed?

Looks to me like a bad install of heat sink. as thats the normal range for a poor fit. Looking at the Whole preview it kinda sucks reguardless. After all isn't the only differance between the 2 cpus HT

 

[Nemesis 1]

^ excellent post regarding the discrepancies in thermal dissipation between the two packages.



Lapping does invalidate the performance replacement warranty, someone pointed me to the fine print before when I mentioned having lapped my 2600K

But really you can't expect Intel, or anyone, to be willing to take on the open-ended risk of covering physically augmented CPU's for warranty purposes. That's a fool's errand for the manufacturer.

In 10 years I have had to lap only 1 cpu for personnel use All the others were flat. But than again I normally have more than 1 cpu to choose from.