i7 920 vs i7 860 direct test comparison

[Otis11]

HT is a big jump in efficiency compared to when it is off. it does a much better job of keeping the cores under full load. thats why its hotter. the wider you make a cpu, the harder it is to keep under load.

Yeah, that's my point. With HT you're sending two sets of commands through the same pipeline, meaning the transistors in that pipeline are busy a higher percent of the time.

I'm assuming Win7 is using the HALT command in the core to lower power consumption, but since any given transistor is less likely to be idle with HT, there are less watts saved because the each cycle is being more effectively used!

That's why I don't see it saving as much power as they say with how we use our computers... But I could be wrong. (that happens more often than I would like... )

[Chumbucket843]

the execution units and cache are the only part of the cpu that are shared when HT is on. thats why its so effective in the atom, because there is no out of order execution it basically doubles throughput. having those transistors used is better than letting them do nothing. they consume the same amount of power either way.

i dont see how win7 would have a reduction in power when its under full load though.

[David_L6]

What's the final verdict for a cruncher - 920 or 860?

[jcool]

Did we ever determine the difference between 4ghz and 3.8 for ppd comparison? Can't remember and I am a bit lazy this week.

I am starting to think the same as Emu. I could have sworn I was running my X3440 @ 4ghz@1.36v, but now I need at least 1.4v or I have lost my head. It still runs fairly cool on air, but cooler is always better. I do not really like the fact of running 1.4v either.

3.8/4.0 shouldn't really make much of a difference. my 920 produce comparably 3.8/4.0

According to my formula, you get roughly 30k PPD at 4Ghz, while you get around 28,5k at 3,8Ghz. So the difference is 1500 PPD, or 750 points per every 100mhz more that you manage to squeeze out of your chip.

However, don't forgt that you will nearly double the CPus power draw if you go from 3,6-3,8Ghz at low Vcore (1,1-1,2V) to 4,2-4,5Ghz high Vcore (1,4-1,5V).

As for Win7 using less power when crunching.. I don't think so. We'd have to set up the same rig with different OS to be sure though.

[Jaco]

What's the final verdict for a cruncher - 920 or 860?

Good question , I'm not so sure either. Correct me if I'm wrong here

If you'd like to upgrade to 6core cpu's , take the X58 platform
If you really need 2 x pcie 16x (for SLI or crossfire) , take the X58 platform

If you don't need any of the above , I'd take whatever is the cheapest.
P55 boards and ram are supposed to be a little cheaper. Don't know about the cpu's (the i7 920 is pretty cheap if you look around)

[WhiteFireDragon]

has anyone done just a very slight OC to 3ghz on the 920 and 860? if not, what do you think the lowest stable vcore will be and power consumption of each chip at that vcore? i'm looking for the lowest wattage chip at 3ghz

[lkiller123]

Xeon L-series is your best bet for low wattage.

[WhiteFireDragon]

well anything that won't kill my budget. i'm guessing between the 860 and 920, the 860 will win in terms of overall cost platform and power consumption? unless the 920 can be significantly undvolted to match the 860 in power consumption

[jcool]

has anyone done just a very slight OC to 3ghz on the 920 and 860? if not, what do you think the lowest stable vcore will be and power consumption of each chip at that vcore? i'm looking for the lowest wattage chip at 3ghz

Not sure about 3Ghz, but I have been selling a few rigs with 920's @ 3,4Ghz (passive water cooling w/o any fans) and undervolted them.. "average" core voltage needed for 3,4Ghz on a D0 is about 1,1V load, sometimes less. 3Ghz should be doable at 1,05V or even 1,0V. Not sure if an 860 can go that low...If you ask me though, going "ULV" only makes sense if you are REALLY limited cooling-wise. Undervolting is nice, but I'd rather use the potential these chips have. Go for 3,4-3,6Ghz and you can still undervolt a 920, but get a huge boost in performance/PPD.
On the other hand, when overclocking a CPU too far, there is a point of diminishing returns where you need an 0,1 to 0,2V increase in core voltage to gain 100-200mhz. Need to find the perfect balance for each chip individually I guess.