i5-750 vs i7-920 TEST

[namegt]

Hello.
i5-750 vs i7-920 Comparison Test
Instead of hard links to a lot of data link

http://namegt.tistory.com/224

Test System 1

CPU: Intel Core i7 920 Nehalem
Mainboard: EVGA X58 SLI Classified [E759]
RAM: CORSAIR DOMINATOR 12800 2x6G
VGA: EVGA GeForce 9800 GT
HDD: WD SATA2 500G
Power: Ultra X3 1000W
Cooler: WATER
OS: Microsoft Windows 7 Utimate K 64bit

Test System 2

CPU: Intel Core i5 750 Lynfield
Mainboard: ASUS P7P55D Deluxe
RAM: CORSAIR DOMINATOR 12800 2x4G
VGA: EVGA GeForce 9800 GT
HDD: WD SATA2 500G
Power: Ultra X3 1000W
Cooler: WATER
OS: Microsoft Windows 7 Utimate K 64bit

[scottsche]

Excellent benchmark collection.

4.5 GHz overclock

Bring the 870's.

[bot@xs]

wow, improved memory latency and power consumption. that is great and thats with the mid-range, non-HT version i5. just wow

also great work done there with the benchmarks and the summery.
thanks

[lordmagnum]

I don't like the rads of the mobo.

[dinos22]

thats a lot of testing there
good work
looks good

[zanzabar]

is turbo on or off

[WaterFlex]

I`m just looking at this...and thinking... I will move only to I9, in 2010.

[Assassin48]

I`m just looking at this...and thinking... I will move only to I9, in 2010.

+1
on that

I have no reason to go to an i5 when I have a good 920

[PatRaceTin]

i think it will replace Q9xxx model

[LinkinParkBoy]

The bandwith benchmark only proves that the higher bandiwith from the trichannel configurations its a myth :P

wow, improved memory latency and power consumption.

True

[lordmagnum]

is turbo on or off

It is turned on.

[zalbard]

Great job! In fact, quite surprising results for me. Triple channel doesn't seem to offer much if at all in terms of performance... Thanks for sharing

[CedricFP]

750 looking good.

Not much need for HT anyway.

Would like to see some H.24 encoding comparisons.

[zanzabar]

It is turned on.

if turbo was on, then this whole thing is broken since with the i5 u get +4 or 6 multi single core and +4 for multi threaded so ur comparing in essence a 3.4-3.6ghz cpu to the i7 that he has that will do 3.0ghz with turbo from the 2.8, atleast to XS

[LinusTech]

if turbo was on, then this whole thing is broken since with the i5 u get +4 or 6 multi single core and +4 for multi threaded so ur comparing in essence a 3.4-3.6ghz cpu to the i7 that he has that will do 3.0ghz with turbo from the 2.8, atleast to XS

That's like saying "if SSE* optimizations are turned on, then the P4 vs K8 comparison is false"

No. The P4 actually outperformed the K8 in many SSE* optimized applications because AMD's implementation was slower than Intel's. That's life.

We're comparing an older chip to one with a new, improved feature mix. If turbo was off, this comparison would be completely irrelevant.

[zanzabar]

i wanted to see the i7 to i5 clock for clock as having both at 2.8GHZ would imply but the i5 has more turbo so it will win in almost everything with turbo on (from the 200-400mhz more) if they are the same clock pre turbo or both had the same turbo multis but they dont

[Chri$ch]

look at the screens, both CPUs working with 2.8Ghz. If the higher turbo were enabled then would CPU-Z show higher multipliers idle

for higher Turbo/CPU multiplier you must enable "C-States" in Bios, if this is not enabled then you have only a +133Mhz turbo.

look at this screen, left = c-state enabled / right = disabled

[zanzabar]

thats good then

[chispy]

excellent comparisons , great job. Thanks for sharing.

[Drwho?]

look at the screens, both CPUs working with 2.8Ghz. If the higher turbo were enabled then would CPU-Z show higher multipliers idle

for higher Turbo/CPU multiplier you must enable "C-States" in Bios, if this is not enabled then you have only a +133Mhz turbo.

look at this screen, left = c-state enabled / right = disabled

Without C6 enable, you will not see Turbo, because the energy saving of C6 is what trigger Turbo from a PCU point of view ...

I wish you guys could wait for the party (end of the NDA) to start ....

and remember the end of the NDA date is under NDA ... ;-)

you guys will have to dig pretty hard to figure out if bloomfield is better than Lynnfield on pure performance ... the answer is not trivial

I wish you guys can start looking at Wattage class , like the 200Watts PC and Below ... the 500watts PC ... and the 1500Watts ... and Above ... and see how you guys can Max out the Bench scores based on this. That could be a good contribution to the OC community for the fight against climate changes ... Just my few cents ... I am back in the milky way, let s talk.

Francois

[Leeghoofd]

Isn't the Classie a tad slower at stock then most mobo's ? Hence why it clocks as good... But still finding that difference in Everest latency very very strange...

[namegt]

C-STATE of the error?
I found that was strange.
First is the C-STATE sangtaebyeol Everest memory benchmarks.
READ WRITE COPY maenwitjul levels of memory do I have your attention.


C-STATE ON - AUTO


C-STATE ON - C1


C-STATE ON - C3


C-STATE ON - C6


C-STATE OFF


Therefore, the difference in power consumption C-STATE ...

[namegt]

i5-750 DDR3-2150MHz

[donmarkoni]

+1
on that

I have no reason to go to an i5 when I have a good 920

+1 more on that. No reason to go for worse.

The bandwith benchmark only proves that the higher bandiwith from the trichannel configurations its a myth :P

No. It's not a myth. Check with MythBusters if you don't believe me. You obviously don't have i7 with triple chanel.
Those benchmark results are either false, or somthing's very wrong with test setup.

[Chewbenator]

So, If you were waiting for the Core i5 release it was still a good idea because you can get basically the same performance for a lower platform price. But, if you went with an i7 it wasn't a bad deal because you still have a solid platform that can only improve in the future with higher DDR3 clocks. Plus, you've had it a good 9 months earlier. Good to see comparable performance but it makes me think that Intel has too many choices in the same performance range.