Thoughts on Dual-Core XEON article?

[fatty4ksu]

Looks like Intel has put out the best chip.

 

[TraumaRN]

Yay for trolls.....:roll:

 

[Cooler]

Only one two test were done?

 

[Leper Messiah]

you mean a pre-production, top of the line chip that is coming almost a year from now is competive with AMD's processor that has been out for 4 or 5 months. 'grats intel. :disgust:


EDIT: Not to mention the 1.5 to 1.8x power difference.

 

[Brunnis]

Originally posted by: fatty4ksu
Looks like Intel has put out the best chip.

Are you kidding?

- Only two tests performed. Performance identical in one and close in the other.
- Power consumption 1.5-1.8x higher.
- Product should probably be compared to new Opterons coming soon.

As far as I can see, we're going to get a CPU that performs about the same and wastes copious amounts of energy for no good reason. I don't know about you, but I think one would have to be a crazy fanboy to think that this is "the best chip".

 

[stevty2889]

The performance didn't look to bad in the limited tests they ran, but the heat and power consumption are rediculous.

 

[cubeless]

dog pile!!! and there was no mention of price... and it's a new chipset...

 

[Avalon]

The heat and power consumption didn't look worth it at all.

 

[Markfw]

troll

So as said above, a chip not released barely stays up in two tests, and takes a lot more power. Lets see a real benchmark when it actually comes out. By then AMD will be able to walk all over it. And thats not fanboy. I know Intel will come back, but they have at least a year before that happens I think.

 

[AnandThenMan]

My question is, who is going to buy these things? The power consumption is downright insulting. As for performance, 2 hand picked tests are hardly enough to evaluate the overall performance of the platform.

 

[Targon]

The key to server performance isn't about how a synthetic benchmark does, but how well it handles a real life server environment. From that perspective, a database benchmark may not be the proper test for a LOT of reasons.

Servers are generally intended to have MANY MANY MANY data requests thrown at them at the same time, and it's how well the server can respond that decides which platform is better. A web server is a perfect example of this, where you have MANY people trying to request information at the same time. The networking setup and environment, disk access, and CPU demand are the keys in that situation. For databases, if you think about it, there will generally be a front end where information is requested, and the back end, which is where the database is located.

Now, one thing that needs to be said here is that in a server environment, if 64 bit is available on the platform, it MAY help performance by quite a bit due to the amount of data being handled by the server. Opteron in a SERVER environment, where legacy applications arn't an issue should be run in 64 bit mode. Also, sticking with an individual database such as Microsoft SQL doesn't provide a proper benchmark. Where are the Oracle and MySQL benchmarks? How about an Apache benchmark?

 

[Leper Messiah]

Originally posted by: Targon
The key to server performance isn't about how a synthetic benchmark does, but how well it handles a real life server environment. From that perspective, a database benchmark may not be the proper test for a LOT of reasons.

Servers are generally intended to have MANY MANY MANY data requests thrown at them at the same time, and it's how well the server can respond that decides which platform is better. A web server is a perfect example of this, where you have MANY people trying to request information at the same time. The networking setup and environment, disk access, and CPU demand are the keys in that situation. For databases, if you think about it, there will generally be a front end where information is requested, and the back end, which is where the database is located.

Now, one thing that needs to be said here is that in a server environment, if 64 bit is available on the platform, it MAY help performance by quite a bit due to the amount of data being handled by the server. Opteron in a SERVER environment, where legacy applications arn't an issue should be run in 64 bit mode. Also, sticking with an individual database such as Microsoft SQL doesn't provide a proper benchmark. Where are the Oracle and MySQL benchmarks? How about an Apache benchmark?

It was called "First Look" for a reason guys. I'm sure intel didn't give them alot of time with this platform.

 

[AnandThenMan]

Originally posted by: Leper Messiah
It was called "First Look" for a reason guys. I'm sure intel didn't give them alot of time with this platform.

Why not? They don't have enough to go around? Why not let Anandtech run a battery of benches. After all, it is Intel's latest and greatest so it must be better than a 3 year old architecture right?

 

[dexvx]

Why is anyone surprised? The primary reason Dual DC Opterons were dominating Dual DC Xeons was because of the Xeon's shared FSB design (at 800 Mhz); it was a huge bottleneck for a CPU that needs to be fed lots of data. The new platform was specifically designed to eliminate the #1 bottleneck.

The main point of interest, in reality, is the chipset, and not the Xeons.

Originally posted by: Leper Messiah
you mean a pre-production, top of the line chip that is coming almost a year from now is competive with AMD's processor that has been out for 4 or 5 months. 'grats intel. :disgust:

The Opteron 280 was released at the end of September, 2005. Dempsey is going to launch in the next 2 months along with Blackford, right after Yonah and the 65nm P4's.

Originally posted by: AnandThenMan
Why not? They don't have enough to go around? Why not let Anandtech run a battery of benches. After all, it is Intel's latest and greatest so it must be better than a 3 year old architecture right?

You realize NetBurst is 6 years old?

 

[AnandThenMan]

I don't care if NetBust is 100 years old, it is being pushed out the door by Intel as their best technology.

But I get your point.

 

[TuxDave]

Originally posted by: stevty2889
The performance didn't look to bad in the limited tests they ran, but the heat and power consumption are rediculous.

Agreed. But it's a one at a time type process. For the past year AMD has dominated in both the power envelope and whatever benchmarks was used. Assuming the benchmarks in the article are accurate, at least you can't argue on the grounds of performance anymore, only power.

 

[Avalon]

Originally posted by: TuxDave

Originally posted by: stevty2889
The performance didn't look to bad in the limited tests they ran, but the heat and power consumption are rediculous.

Agreed. But it's a one at a time type process. For the past year AMD has dominated in both the power envelope and whatever benchmarks was used. Assuming the benchmarks in the article are accurate, at least you can't argue on the grounds of performance anymore, only power.

In the tests done, true. I'd still like to see something a little more expansive later down the road.

 

[Leper Messiah]

Originally posted by: dexvx
Why is anyone surprised? The primary reason Dual DC Opterons were dominating Dual DC Xeons was because of the Xeon's shared FSB design (at 800 Mhz); it was a huge bottleneck for a CPU that needs to be fed lots of data. The new platform was specifically designed to eliminate the #1 bottleneck.

The main point of interest, in reality, is the chipset, and not the Xeons.

Originally posted by: Leper Messiah
you mean a pre-production, top of the line chip that is coming almost a year from now is competive with AMD's processor that has been out for 4 or 5 months. 'grats intel. :disgust:

The Opteron 280 was released at the end of September, 2005. Dempsey is going to launch in the next 2 months along with Blackford, right after Yonah and the 65nm P4's.

Originally posted by: AnandThenMan
Why not? They don't have enough to go around? Why not let Anandtech run a battery of benches. After all, it is Intel's latest and greatest so it must be better than a 3 year old architecture right?

You realize NetBurst is 6 years old?

right....Q1 can mean anything from january to march. With Intel's track record on releases recently...you get my drift? Bump opty's to 2.6GHz and they'll be basically right where they were before, cept intel still gets raped on power.