Intel Core i7 Application Results @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,534
Intel Core i7 Application Results - We take the Intel Core i7 965EE, 940, and 920 processors, and run them through six content creation applications stretching from Premiere Pro CS4 to encoding MP3s. Clock-for-clock comparisons are supplied as well as a look at what two extra cores get you compared to the "old" dually E8500.

The Intel Core i7 looks to be doing very well in the encoding and content creation arena. Simply put, these are areas that can benefit greatly from 10% to 20% faster encode times. While this may not be realized on the “normal” consumers’ desktop, there is no doubt in my mind that some of our “file sharing enthusiasts” as well as content creation professionals will see huge benefits in gaining an edge on current efficiency problems.
 
so the E8500 got smacked around by the i7 920, and the $1000 QX9770 put up a good fight against the same. Sounds worthy to me. :D

Nice numbers.
 
So an overclocked (3.4 to 3.8 ghz) non extreme CPU (Q9550, Q9650) could potentially put up a fight with the i7 920? Interesting!
 
Yea but the 920 can be overclocked too. Too early to tell how well.
 
I was expecting gains at 30 to 40 percent improvement. Still not worth the upgrade of motherboard, cpu, and ram imo.
 
fantastic news but look at the 920 here
1225687542NjsvdgF65P_1_2.gif


whats that all about?
 
Very encouraging results. All I'm waiting for is some more in-depth reviews with some overclocking tests thrown in!
 
I think that is the graph Intel secretly planted for AMD.
 
It's not that it sucks. It's just that you're seeing the scenario where all the goodies i7 has to offer (ie. the gains in multithreaded efficiency and insane memory bandwidth) are being thrown to the way side by an app that cares more about clock speed then cache or bandwidth.

The slower clocked C2Qs show the same thing.
 
So going by the Core2 benches, some are showing some big changes going from 1333 to 1600. Is that DDR2 for 1333 and DDR3 for 1600? If so WOW! :eek:

EDIT: NM I think I understand now. The 1333 is the C2D and the 1600 is the Quad! Oops! Damn rough Monday mornings! :p
 
Rudimentary question here...(RAM always screws me up)...Can someone explain to me why the 920 is listed at 1066 but the 940/960 are at 1600? I thought the processors were identical but for the clock speeds.
 
lol. the graph i posted? is it an error? surely the i7 920 doesnt suck that bad?

If you look at the clock speeds and how LAME scales it is quite logical as to the score the way I see it.

The i7-3.2 is 19% faster in clock speed than the i7-2.6 and it is 17% faster in this single threaded benchmark. And obviously no coding has gone on here to leverage Nehalem's architecture.
 
Rudimentary question here...(RAM always screws me up)...Can someone explain to me why the 920 is listed at 1066 but the 940/960 are at 1600? I thought the processors were identical but for the clock speeds.

The 920 has a RAM speed of 1066. The STOCK 940 and 965EE RAM speeds are 1333, but are tweakable to 1600 by changing ratios where the 920 is not, or al least it has not been supported in any BIOS I have seen yet.

So yes, we overclocked the RAM bus on the 965EE and 940. That is what the OCP stand for. ;)
 
The LAME graph **DOES NOT START FROM ZERO**

Zoomed in graphs are like lying. I thought [H] was better than that. Look, you've confused jonneymendoza into thinking the i7-920 is terrible.
 
Looks like my 3.2GHz C2Q should last me longer.

Edit: I saw quite a few sites are using SSD drive for their review. Would the performance gain be affected with a slower harddisk drive?
 
The LAME graph **DOES NOT START FROM ZERO**

Zoomed in graphs are like lying. I thought [H] was better than that. Look, you've confused jonneymendoza into thinking the i7-920 is terrible.

I didn't even catch that. Thanks for bringing it to my attention and sorry for not catching your sarcasm first time around. Edit for the win! It is fixed now. :)
 
Looks like my 3.2GHz C2Q should last me longer.

Edit: I saw quite a few sites are using SSD drive for their review. Would the performance gain be affected with a slower harddisk drive?


Not unless you are using IO dependent benchmarks to test processor power. :eek:
 
I would like to see an E8600 5GHz core 2 DUO go up against an overclocked core i7 in some encoding/gaming benchies.
 
5Ghz vs 5Ghz?
That would be fun

Id kind've like to see the X6800 vs 965 myself. Best of a tock vs best of the new tock.
 
I know performance comes for a premium price but this is ridiculous. Guess I'll stick with AMD for another year :/
 
I wish there was some way to consistently benchmark distributed computing on processors. Folding @ Home of course, but projects that don't have GPU clients like Rosetta @ Home this would be interesting.

But I realize this would be extremely difficult, since at least with Rosetta there are often different types of calculations done by varying batches of work units.

Honestly, distributed computing is the only real reason I would upgrade from a dual core ATM. For games like the [H] says it's all about GPU.

edit: typo
 
Photoshop should have been included.

Looks promising. Now who long before an affordable combo hits the streets.
 
I would hate to ask for more programs, but here goes:
With the release of the Core i7, x264 (H.264 encoder) will soon see its fair share of optimization patches. Would it be possible to include some x264 benchmarks with the current builds (r1016 for example) and the soon-to-be-released ones that will include Core i7 optimizations?
 
Excellent as always, but could I make a suggestion?

Is there any way to possibly augment the digital video workstation content creation benchmark (Adobe Premiere Pro CS4) with an digital audio workstation content creation benchmark?

http://www.dawbench.com/benchmarks.htm

I will look into this. Thanks for the heads up. Always looking to better that suite of real world testing.
 
I would hate to ask for more programs, but here goes:
With the release of the Core i7, x264 (H.264 encoder) will soon see its fair share of optimization patches. Would it be possible to include some x264 benchmarks with the current builds (r1016 for example) and the soon-to-be-released ones that will include Core i7 optimizations?


I used the latest software available and will certainly be keeping our ear to the ground in regards to i7 specific updates.
 
Is that not tweakable at all, or is that not tweakable appart from OCing the FSB equivalent but rising in proportion to it. the former could be a deal breaker if there's a significant amount of ram speed dependence in performance. If the 3.8g OC you reported in one of your other threads is viable long term the latter wouldn't be a major problem since the CPU clock OC would lift the ram clock with it.

I do not understand your question.

We took the 2.66-1066 920 part and overclocked it to 3.8-1520 easily.
 
What is the deal with these CPU's being hard wired so that the non extreme versions cannot dissipate more then 130W? Are there work arounds for this in the bios or is this something we have to deal with, if that is the case that would explain why the OC results for the 920 vs 920 stock speeds were virtually identical!
 
What is the deal with these CPU's being hard wired so that the non extreme versions cannot dissipate more then 130W? Are there work arounds for this in the bios or is this something we have to deal with, if that is the case that would explain why the OC results for the 920 vs 920 stock speeds were virtually identical!

I am not sure about that but I do have a mail into Intel asking them about this. Until we get an answer on that, let me share with you a couple of 920 screen shots at 3.8GHz.

First, one of the CPU 100% loaded while being overclocked to 3.8GHz. CPUZ is dynamic and does record clock changes in real time. (10 minutes later as of posting this, it is still showing the same clockspeed.

FullCPULOad-920_at_3800.jpg


And the second sporting a Sandra CPU score of over 90,000.

sandra2009-920_at_3800.jpg


Is this addressing any of your concerns? Of course these are samples from Intel but we did receive a specific 920 part. We were not just asked to downclock to other speeds as we usually are. As is the case this time with the 965EE and 940 being the same CPUs just clocked differently for testing.
 
As far as the 130W limit on the CPU, this can be changed in BIOS on the intel board. PCPer has an overclocking walkthrough that explains the bios options here is an excerpt:

TDC current limit override and TDP power limit override: These settings determine when the CPU will no longer enable Turbo Mode multipliers and by raising these you can increase the envelope that the CPU has to continue to increase the Turbo multipliers. Just make sure you have proper cooling of the CPU as you raise these two limits.
 
What kind of cooling is necessary for 3.8Ghz?


Used water, but have not tried air. Dan got the 965EE up to 3.8+ on air. I could get the 920 to 4GHz + on water, but not stable and I could get the 965EE up to 4GHz on water and it was stable.
 
Are these tests specified someplace? I'd like to try to run them on my existing rig to set a baseline, and learn how an i7 system might compare. I'd also like to try and repeat them when I build my i7 rig.

I notice that these tests are I/O involved, but use an SSD. I suspect that the vast majority of the readers of the site are using mechanical drives. Will they be able to see proportional increases in performance? That is, to what weight are the tests performed here I/O-bound?
 
Back
Top