Thanks.
If you could, an idle temp would help as well, as it gives a differential for comparison.
Yes, and it will be in the i7 based iMac as well (source).
It communicates via DMI rather than QPI.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
My Quad Core 27-inch i5 has arrived! (includes pictures and Benchmark)
- Thread starter Carressa
- Start date
- Sort by reaction score
If you could, an idle temp would help as well, as it gives a differential for comparison.
It communicates via DMI rather than QPI.
But for a lot less coin than the base mac pro too remember.
AND they come with a 27" LED panel...
that said, i use the 2.6 quad for work and its great but for an 'enthusiast' home computer, simply too expensive.
Just to let anyone know I have a hackintosh now before all the fan boys get upset, because I don't have a real mac I just wanted to let you know how my core i7 920 scored. In 32-bit mode it scored 8235 with 6Gb of 1066 DDR3 in triple channel running 10.5.8. Just for fun at 3.35 Ghz it scored 10478. Should give you a good idea if the upgrade to the i7 is worth it.
CPU @ 2.67 Ghz
Geekbench 2.1.4 for Mac OS X x86 (32-bit)
Integer Processor integer performance 6754
Floating Point Processor floating point performance 13097
Memory Memory performance 4393
Stream Memory bandwidth performance 4094
CPU @ 3.35 Ghz
Geekbench 2.1.4 for Mac OS X x86 (32-bit)
Integer Processor integer performance 8535
Floating Point Processor floating point performance 16676
Memory Memory performance 5618
Stream Memory bandwidth performance 5312
I've been using OS X since it was in beta, and have had many macs its just a fun side project.
^.^
but the Core i7 in iMac 860/LGA1156/Dual channel
but yours is different socket and over clocked so not related but thanks anyway
but the Core i7 in iMac 860/LGA1156/Dual channel
but yours is different socket and over clocked so not related but thanks anyway
You think that's night and day ... I'm looking at getting an i7 iMac as well (probably after Christmas), but I'm currently using an 11 year old 266MHz beige G3 PowerMac running Mac OS 9 with a 19" CRT display ... that must mean I'm going from night to a day next year.
Thank you all I was trying to show was the difference between the OP core i5 and the core i7 socket size really makes no difference, more I was trying to show the difference between having HT and triple channel memory even though the i5 only has dual channel.
OP's iMac
Geekbench 2.1.4 for Mac OS X x86 (32-bit)
Integer Processor integer performance 4967
Floating Point Processor floating point performance 10436
Memory Memory performance 3630
Stream Memory bandwidth performance 3791
Geekbench Score: 6496
My hackintosh
CPU @ 2.67 Ghz
Geekbench 2.1.4 for Mac OS X x86 (32-bit)
Integer Processor integer performance 6754
Floating Point Processor floating point performance 13097
Memory Memory performance 4393
Stream Memory bandwidth performance 4094
Geekbench Score: 8235
the i7 iMac only has dual channel RAM does it not? *checks*
yea it must, cauz the 860 has "down graded" back to dual channel.
Geekbench is essentially a synthetic benchmark when it comes to the memory side of it. The Core i7 860 iMac is going to churn out CPU scores like the Core i7 920 and maybe a little faster. The Bloomfield is going to yield a higher synthetic memory benchmark score but in real world applications it's not going to mean much on a single socket for something like Handbrake.
Real world apps will see little to no performance difference, only if they use HT. It comes down to if you think your money is well spent on the processor vs the rest of the computer, or treat your friends to a good dinner. I just want that resolution 2560 x 1440 better then my 27.5 in @ 1920 x 1200
a few benchmarks i read showed that video performance (encoding etc) had a pretty nice improvement - the most significant of the lot. somewhere around 10%->15%.?
p.s. thanks for memory clarificaiton
Just a reminder that the Core i7 860 and Core i7 920 both have Hyper Threading.
Only bad thing of the 8xxs is that they have DMI instead of fast QPI.
Tom's hardware is really good for benchmarks. With handbrake I see little difference with HT and with out it. Even though with HT the processor will go to 800% vs 400% with out it
what sort of performance hit is there from QPI->DMI?
ssticky: my mistake, i was referring to the i5 vs i7 imac chips.
comparing the 860 to 920 was around the same IIRC.
Just a reminder that the Core i7 860 and Core i7 920 both have Hyper Threading.
Correct but the core i5 does not.
QP1 reaches 6.4GB/s and DMI reaches 2GB/s
Edit - BTW, DMI reaches theoritically up to 2GB/s, true speed is 10Gb/s
Lynnfield has an onboard memory controller and PCI-Express 2.0 one as well.
The problem is that you only get a few lanes over DMI for stuff like USB 3.0 or SATA 6 Gbps. QPI connects to the X58 chipset which handles 36 PCIe 2.0 lanes for GPU and crazy I/O peripheral goodness.
HT has its positives and negatives i guess. minor performance hits, but meh.
really!? wow i didnt know it was that bad. ugh. not that most people would reach the full potential of this bandwidth though
oh up to 2GB/s. 1.25GB/s actual?
Exactly. Which is why I am not buying the 1156 socket chipset, its a bit too limited for my taste.
Really how much of a difference would it make, when I overclock my QPI from 4.8 to 6.4 in cinebench there is no difference. I've timed both I believe 54 seconds on the image.
There isn't much difference in a 4.8GB/s vs 6.4GB/s because of memory clocks and such.
But in an 1156 system there is, because DMI is clocked up to that specification. Sure you can overclock, but for that case, just go with the 1366 socket which allows greater overclocking.