More Woodcrest benches

[DrMrLordX]

Sure, Woodcrest and Conroe aren't exactly the same, but it's still a good showing for the Core 2 architecture:

http://www.gamepc.com/labs/view_content...87CC08E82653397AE3DA1B5D2E4DA8AFB55932

Of course, this is also linked by DailyTech. Note the encoding and rendering benchmarks where Woodcrest dominates at a low clock speed. Also, we see once again someone testing Core 2 in Sciencemark with lackluster results. I wonder why?

 

[classy]

Originally posted by: DrMrLordX
Sure, Woodcrest and Conroe aren't exactly the same, but it's still a good showing for the Core 2 architecture:

http://www.gamepc.com/labs/view_content...87CC08E82653397AE3DA1B5D2E4DA8AFB55932

Of course, this is also linked by DailyTech. Note the encoding and rendering benchmarks where Woodcrest dominates at a low clock speed. Also, we see once again someone testing Core 2 in Sciencemark with lackluster results. I wonder why?

Hmm it squeaked out a couple of wins, but still lost the big one rather handily, the Apache benchmark. If they can't unseat AMD in web and database, they can sell their chips for any price and people still won't buy them. The rendering is very impressive, but Intel had that edge with P4.

 

[Furen]

Power draw is remarkably close. I think this will be AMD's if you consider that power draw will drop a bit with LGA1207 (and if people use more than 4 DIMMs).

 

[dexvx]

Originally posted by: Furen
Power draw is remarkably close. I think this will be AMD's if you consider that power draw will drop a bit with LGA1207 (and if people use more than 4 DIMMs).

Doesnt that use FB-DIMMS?

 

[Furen]

Socket F? Nah, it uses Dual-DDR2, so Intel will be at an "AMB" disadvatange with regards to power. Supposedly Socket F will eventually use FBs but not quite yet.

 

[ahock]

for me, somehow I agree with the comment of the reviewer. mid-range performance is somewhat the similar if not better to top range opteron.

 

[zsdersw]

Hmm it squeaked out a couple of wins, but still lost the big one rather handily, the Apache benchmark. If they can't unseat AMD in web and database, they can sell their chips for any price and people still won't buy them. The rendering is very impressive, but Intel had that edge with P4.

Also keep in mind that those were mid-range Woodcrests... and they beat or were very close to the fastest (and most expensive) Opterons.

 

[Furen]

Originally posted by: zsdersw

Hmm it squeaked out a couple of wins, but still lost the big one rather handily, the Apache benchmark. If they can't unseat AMD in web and database, they can sell their chips for any price and people still won't buy them. The rendering is very impressive, but Intel had that edge with P4.

Also keep in mind that those were mid-range Woodcrests... and they beat or were very close to the fastest (and most expensive) Opterons.

That's on Workstation/Desktop Benchmarks. The sole "server" benchmark at gamePC (Apache) they lost to even the 2GHz Opteron, it wasn't even close.

 

[zsdersw]

That's on Workstation/Desktop Benchmarks. The sole "server" benchmark at gamePC (Apache) they lost to even the 2GHz Opteron, it wasn't even close.

The Xeon 5140 will cost approximately $455. The Opteron 270 can be had for $458. Seems pretty close to me. Price/performance is nearly identical between the two with regards to that one benchmark at gamePC.

 

[imported_ITguy]

Originally posted by: zsdersw

Hmm it squeaked out a couple of wins, but still lost the big one rather handily, the Apache benchmark. If they can't unseat AMD in web and database, they can sell their chips for any price and people still won't buy them. The rendering is very impressive, but Intel had that edge with P4.

Also keep in mind that those were mid-range Woodcrests... and they beat or were very close to the fastest (and most expensive) Opterons.

Also, keep in mind that those were new still unreleased Woodcrests against old Opterons that have been out for years........and were not tested against new Rev F Opterons and 65nm Opterons which will be out this year as well.

The true important benchmarks are Apache and 64bit based benchmarks. In 2007 and on the 64bit benchmarks will rule the day.

 

[zsdersw]

Also, keep in mind that those were new still unreleased Woodcrests against old Opterons that have been out for years........and were not tested against new Rev F Opterons and 65nm Opterons which will be out this year as well.

Comparisons are always made between what you can get your hands on. If someone can get a hold of a Rev. F or 65nm Opteron they would've made comparisons with them.

Woodcrests will be available before either of those, though, so the point is moot.

 

[MDme]

based on these benchies, it seems that the current K8 core needs a 400Mhz clock advantage to be equal to/ or superior to the woodcrest/conroe core. Although the server bench apache shows that the K8 isn't done yet. when AMD launches 65nm rev G K8's AMD might be able to narrow the gap using increased clocks. In fact, If AMD can maintain a 400Mhz clock advantage at equal price points with conroe, then AMD will still be competitive. But by saying that AMD needs a 3.4 Ghz FX (which would be the FX-68) to match the x6800 and a 3.06 Ghz+ FX-64 to fight the E6700 (2.66Ghz). Also, note that these are 90nm opterons vs 65nm woodcrests, if optys can get better clocks and lower power with 65nm SOI then it will be more competitive.

 

[zsdersw]

All of that is definitely feasible... but so too is scaling for Conroe/Woodcrest. The fastest Conroe/Woodcrest at launch may not be the fastest available when 65nm Opterons debut.

 

[coldpower27]

Meh, let's see what happens when they get a ahnd of the 3.0GHZ Woodcrests The 5140 did fairly well though for a mid range part.

 

[Dubb]

personally I don't give a rat's ass about apache. I want a chip that's fast at rendering and multitasking.

 

[DrMrLordX]

Before you crown Opteron the now-and-future King of Apache, do recall that Anandtech did their own Woodcrest review (albeit of 3 ghz chips) here:

http://www.anandtech.com/IT/showdoc.aspx?i=2772&p=6

Notice how Woodcrest does extremely well here? Honestly, I think the GamePC review that I linked above is more useful for determining Woodcrest's potential as a workstation processor. The difference in performance between Anandtech's and GamePC's Apache results don't make much sense, but in the end, I'd have to go with the data provided by Anandtech.

 

[Accord99]

I think GamePC uses the built-in benchmarking utility of Apache, and is only making the same requests for a single static page. Anandtech's review is using a more real-world setup that comprises dynamic and static content and database queries.

 

[DrMrLordX]

That would probably explain the difference then. Thanks. I'll definitely stick with Anandtech's conclusion that Woodcrest is good for Apache.

 

[Viditor]

Originally posted by: Accord99
I think GamePC uses the built-in benchmarking utility of Apache, and is only making the same requests for a single static page. Anandtech's review is using a more real-world setup that comprises dynamic and static content and database queries.

Actually, the difference could very well be that Johan used a 64 bit Gentoo Kernel 2.6.15-gentoo-r7, while GamePC used WinXP64...

 

[Accord99]

Originally posted by: Viditor

Originally posted by: Accord99
I think GamePC uses the built-in benchmarking utility of Apache, and is only making the same requests for a single static page. Anandtech's review is using a more real-world setup that comprises dynamic and static content and database queries.

Actually, the difference could very well be that Johan used a 64 bit Gentoo Kernel 2.6.15-gentoo-r7, while GamePC used WinXP64...

They're not even using the same benchmark. GamePC mentioned using the integrated Apache benchmark utility, which would be "ab"and most likely just requesting a single static page. Johan had a custom script for his test.

 

[ND40oz]

Not a bad showing for woodcrest, be nice if they also had faster chips to review so we could see how it scaled in comparison to clockspeed. I like how Woodcrest is spelled on CPUZ 1.34.1.