New Intel CPUs rock

[Groucho2004]

Here are a couple of reviews which also use x264 for benchmarking (ok a pretty old version but still...):

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3634
http://techreport.com/articles.x/17545

I think the most impressive feature is the power efficiency of the new processors and P55 mainboards.

Edit: Xbitlabs are using a newer version of x264 (r1158):
http://www.xbitlabs.com/articles/cpu/display/core-i7-870_10.html

[burfadel]

These scores come from the newer, faster version 0.59.819 of the x264 executable. http://techreport.com/articles.x/17545/10

Is that revision 819 they're using? Thats piss poor if you ask me!

[Groucho2004]

Thats piss poor if you ask me!

I know. Someone here has already tried to convince the guys at Techreport to update to a new version but no luck so far.

Anyway, I don't think the relative performance difference would be that huge. I could be wrong though and the Lynnfield specific optimisations do make a big difference. The guys who would know best are akupenguin and Dark Shikari.

[Groucho2004]

Thats piss poor if you ask me!

Actually, it would be so simple to put together a new benchmark. All it needs is some HD trailer, a simple Avisynth script and a batch file. The results could be posted/hosted here.

[nakTT]

I know. Someone here has already tried to convince the guys at Techreport to update to a new version but no luck so far.

Anyway, I don't think the relative performance difference would be that huge. I could be wrong though and the Lynnfield specific optimisations do make a big difference. The guys who would know best are akupenguin and Dark Shikari.
Wonder why is it so darn hard for them to change to a new version. I know that will make their older review not directly comparable to new one but still, we have to move forward.

http://techreport.com/articles.x/17545/10

Is that revision 819 they're using? Thats piss poor if you ask me!
Yeah, its too old already.

[Groucho2004]

Wonder why is it so darn hard for them to change to a new version.

Lack of interest, I suppose. Most people are interested in games benchmarks. Some of those are probably not even aware that a computer can be used for something other than games and browsing.

[nakTT]

Lack of interest, I suppose. Most people are interested in games benchmarks. Some of those are probably not even aware that a computer can be used for something other than games and browsing.
That is too bad. Then again, it is the reviewer the one who must keep up with the technology no just doing benchmarketing using a very old version.

[Selur]

also keeping your old version allows you to compare the result with old test results from other pcs, so laziness of redoing tests might also be a factor for them to not update ;)

[Groucho2004]

benchmarketing

Very funny. Did you misspell this or is it intentional?

[LoRd_MuldeR]

However there were various optimizations that only effect the newer CPU's (especially Nehalem). Not including those simply gives "wrong" values, because in reality everybody would use an up-to-date build.

IMHO comparing your "old" results with "new" results, that just don't represent that real performance any more, is kind of useless.

They should either re-do the benchmark with an up-to-date x264 build on all CPU's shown on the graph (that's not too hard, if you have those machines anyway) or they should skip the x264 graph entirely.

[foxyshadis]

I don't think it matters. It conveys what's really needed, the rankings would still be the same. The winners would just win by even more. Sites that still use three different versions of 3dmark might be cajoled into using another x264 version.

Now back to the actual topic of the thread: i7 is amazing, but I'll probably have an i5 in my laptop long before I upgrade the Q6700 encoding box. I don't know what I'd do with the old guts. Also, I'm moving this into hardware discussion, that's where benchmarks go. Groucho, we have actually had exactly that benchmark (http://forum.doom9.org/showthread.php?t=135923) on the forum for a long time. In fact, I should have stickied it a long time ago...

[nakTT]

Very funny. Did you misspell this or is it intentional?
More like the latter :D

Sorry for this off topic post.

[Blue_MiSfit]

I've harassed techreport several times about updating their x264 benchmarks :) but foxyshadis makes a good point about it not mattering that much.

Yup. Dual Nehalem Xeons are the last word in encoding performance ATM, but the new Lynnfield CPUs (LGA 1156 i5's and i7's) are very impressive as a mainstream product, particularly regarding their energy required to complete a given task.

I'm still pretty happy with my Q6600 @ 3.4 GHz, and my dual quad-core xeon servers are still pretty durn fast as well.. :)

Still, if I had the cash, I'd switch up to a Lynnfield i7 right away. I'm due for a "freshening" - it's been about 2 years since I did a full system upgrade! Maybe around christmas when hopefully we'll have Radeon HD5k series cards making waves...

@foxyshadis: Make your old guts into a media-center, or a NAS ;)

~MiSfit

[foxyshadis]

But my encoding box is my silent media center, too. :p I'd have to get another tower & psu and give it to someone, I'm sure.

[Blue_MiSfit]

A silent Q6700? Nice :) My workstation at work is a first-gen Q6700 that gets inconceivably loud when at 100%, and temps are in the mid 70c range :D

Sounds like you've got a nice cooler!!

[saint-francis]

A silent Q6700? Nice :) My workstation at work is a first-gen Q6700 that gets inconceivably loud when at 100%, and temps are in the mid 70c range :D

Sounds like you've got a nice cooler!!

Want silent? Just cool with water and plumb your radiator and pump to the basement. (http://www.xtremesystems.org/forums/showpost.php?p=3891334&postcount=328) ;)

IMO a better way to spend your money (if you already have a good core2 system) would be to just get water cooling and overclock more. You can use the same water cooling gear through multiple systems and keep getting great cooling and overclocking. Good water cooling will cost MUCH less than a new i7 system.

[LoRd_MuldeR]

I don't think it matters. It conveys what's really needed, the rankings would still be the same. The winners would just win by even more.

Well, that the latest CPU generation is faster than the previous one isn't a surprise. Also that higher clock-speed within the same generation gives better performance should be clear.

So the ranking doesn't really matter. IMO the real interesting question is: How much faster is the new generation compared to the previous one and does the speed-up legitimate an upgrade ???

Therefore the Nehalem-specific optimizations may make the important difference between "just faster" and "significant faster" or even "fundamentally faster" ;)

[foxyshadis]

I bought it from Costco cheap, and replaced all the fans and power supply when I got tired of the whining. It works great.

I just saw the prices of the just-released i7 820. $299 + $200 for the mobo & memory. That's not as bad as I was expecting, but I'll still hold off until the holidays at least. I have to figure out whether I'm going to quit and move again or not, plus Canon just released a new series of Powershot cameras, and I've been wanting to get one of those for four years now to replace my old Kodak. Decisions...

Mulder, that's a good point, and I wonder how much of that difference would show up. That version of x264 doesn't even have most of the core2 optimizations.

Francis, that's the most awesome ghetto watercooler I've ever seen. I remember someone whose water block terminated in a bucket, when temps started to rise he'd dump ice in. :D How much power does the pump use, have you ever measured?

[Astrophizz]

DS mentioned some code efficiency numbers on different processors once - maybe you could scale the results with that for a quick and dirty estimation. Still doesn't solve the issue of using old software for a review presented to the general public who doesn't read Doom9.

[Groucho2004]

IMO the real interesting question is: How much faster is the new generation compared to the previous one and does the speed-up legitimate an upgrade ???

Therefore the Nehalem-specific optimizations may make the important difference between "just faster" and "significant faster" or even "fundamentally faster" ;)

I edited my first post. Xbitlabs are using a version of x264 that already has (most of) the Nehalem optimizations.

[saint-francis]

Francis, that's the most awesome ghetto watercooler I've ever seen. I remember someone whose water block terminated in a bucket, when temps started to rise he'd dump ice in. :D How much power does the pump use, have you ever measured?

The electric bill didn't even increase $1 when I got the pump. I have a Kill-A-Watt but haven't bothered to measure because the power usage seems negligible compared to my wife leaving lights on everywhere. I have another heater core on there now too and some crazy 5' long passive radiant turbulator. The only problem I have is the CPU getting below ambient temperatures and creating condensation.

[divengrabber]

ya you are right Intel CPUs always stand top of the chart.