Ivy bridge overclocking is disappointing

[Shonumi]

(03-26-2012, 01:24 AM)Starscream Wrote: That actually slowed things down even more for me, approx 5% slower on my AMD chip.

Well that's what you get for not having a Sandy Bridge chip :p

[AnyOldName3]

I was going to post that, but was too intelligent to fake-make-fun-of an administrator.

[NaturalViolence]

Quote:Yeah I was surprised with the results too. The incresed temps are more likely due to higher levels of resistance/impedance caused by using shorter wires (due to smaller die). Perhaps we've reached a point where the thermal benefits of requiring less voltage are outweighed by the higher electrical resistance. It will be very interesting to see what happens with the move to 14nm, but at least we'll have one more architecture revision to boost performance before then.

Anyway, regardless of the reasons, all testing so far points to one thing.....we're going to have to live with higher temps with ivy bridge.

That also makes no sense. Shorter wires would LOWER impedance. If the electrical resistance had increased than it makes no sense that the temps are much lower at stock settings compared to sandy bridge.

[bret emerald]

(03-26-2012, 05:21 AM)NaturalViolence Wrote: That also makes no sense. Shorter wires would LOWER impedance. If the electrical resistance had increased than it makes no sense that the temps are much lower at stock settings compared to sandy bridge.

That's very true in fact, don't know why I was thinking shorter would increase impedance. Unless the wires are longer relative to the transistor size....Ah forget it, disregard my last post completely.

Something doesn't add up with these 'Tweaktown' previews, there has to be a reason for those overly high temperatures (hopefully it has nothing to do with that stupid new graphics core). I also think their OC methods may be questionable. If you look at the cpu-z screens showing the OC clocks, the vcore voltages appear to be very low...suggests to me they didn't enable load line calibration. Plus I can't help but think increasing the bclk so much as they did with the 3570k, can't be a good idea. I understand that they were using stepping E0 and the first retail chips will be E1 I think, so hopefully those will give us much better results......had my heart set on upgrading my i5 750 to a 3570k.

Btw any chance you could point me to a graph comparing stock temps with sandy bridge, can't remember seeing one (it's a shame 'Anandtech' didn't include one in their preview).

[NaturalViolence]

Quote:Btw any chance you could point me to a graph comparing stock temps with sandy bridge, can't remember seeing one (it's a shame 'Anandtech' didn't include one in their preview).

You don't need one actually. Intel has published the TDP values, the high end quad cores are 77w TDP while sandy bridge is 95w TDP. So unless intel is lying to OEMs (or the stock cooler is way worse than sandy bridge which wouldn't make any sense) ivy bridge will run cooler.

[DefenderX]

I hope, you're right.

I have saved money for Ivy Bridge

[bret emerald]

You don't need one actually. Intel has published the TDP values, the high end quad cores are 77w TDP while sandy bridge is 95w TDP. So unless intel is lying to OEMs (or the stock cooler is way worse than sandy bridge which wouldn't make any sense) ivy bridge will run cooler.
[/quote]

Yes logic would suggest that you would only need to look at the TDP of each, however I had another look at Tweaktown's preview and found this



It seems that the 3770k is running 11 degrees hotter under load than the 2600k despite supposedly needing to dissipate 18 watts less of thermal energy.

I don't have much experience with that particular site, can anyone here confirm how reliable their test results tend to be?

[lamedude]

Engineering sample leaked to website != retail chips

[bret emerald]

(03-28-2012, 02:28 AM)lamedude Wrote: Engineering sample leaked to website != retail chips

Hmmm, and where did you come across this fine snippet of information? Personally I see no evidence to suggest that they did not obtain the sample directly from Intel.

The chips they used were E0 stepping (same as all the other previews), and it happens to be a very recent revision. In fact had the Ivy Bridge launch not been pushed back E0 could well have been the first retail revision. Hell, E0 could still be what we see at retail.

These things are due to launch in a week or 2 (not 6 months from now), and I think it's quite naive to dismiss these test results simply because they are not from retail.

[dannzen]

ivybridge has a large gpu unit :'(
heat...
space...
...