Dell to use AMD chips!! (official)

[Gubbi]

Woodcrest previewed...
http://www.techreport.com/etc/2006q2/woodcrest/index.x?pg=1
So far it would appear intel will be faster than amd untill amd has a real answer in both server and workstation markets, and by no small margin either.

I don't think Intel will be faster in the 4 socket market (lack of system infrastructure), which is where Dell will be using AMD.

Everywhere else... Intel will be faster.

Cheers

 

[Moloch]

I don't think Intel will be faster in the 4 socket market (lack of system infrastructure), which is where Dell will be using AMD.

Everywhere else... Intel will be faster.

Cheers

Time will tell

 

[Blazkowicz]

I don't think Intel will be faster in the 4 socket market (lack of system infrastructure), which is where Dell will be using AMD.

Everywhere else... Intel will be faster.

Cheers

Dell is making 4-way servers. so I see that as 2 socket machines with dual core opteron

 

[Gubbi]

Dell is making 4-way servers. so I see that as 2 socket machines with dual core opteron

I don't think Opteron will be competitive with the new Intel CPUs in the 2 socket space, so can't see Dell even bothering with 2 socket. In the four socket space AMD enjoys good performance compared to shared-bus Intel chipsets and a clear cost advantage.

Cheers

 

[Frank]

Everywhere else... Intel will be faster.

Cheers

... than the current AMD line? Definitely. Support for DDR2 for AMD processors might change that and close the gap a fair bit. But I still wouldn't buy a Conroe if it's twice or more as expensive as a Sempron, especially if I had to buy expensive FB-DIMMs as well. What would be the use?

As for servers, dual or quad Opterons have quite a lead as it stands. Synthetic benchmarks and rendering are very nice and all, but not what servers do in general. I would like to see some nice benchies about that before I make up my mind.

 

[Accord1999]

... than the current AMD line? Definitely. Support for DDR2 for AMD processors might change that and close the gap a fair bit.

Seems to be mainly a small boost, and only if you pay for the most expensive memory available.

But I still wouldn't buy a Conroe if it's twice or more as expensive as a Sempron, especially if I had to buy expensive FB-DIMMs as well. What would be the use?

What if said Conroe is upwards of 50-100% faster in single-threaded applications? FB-DIMMs are for servers only.

As for servers, dual or quad Opterons have quite a lead as it stands. Synthetic benchmarks and rendering are very nice and all, but not what servers do in general. I would like to see some nice benchies about that before I make up my mind.

Here are some server benchmarks:

http://www.intel.com/performance/server/xeon/database.htm

Woodcrest is only 2S though, 4S Intel will have to make do with Netburst-based Tulsa.

 

[chavvdarrr]

no doubt Conroe based servers will be much better than current P4-based, but THEY ARE NOT AVAILABLE YET.
Add few months in order to see MB's with revision 1.02 , also note that in 4S designs Opterons still will have higher bandwidth. Server market is where AMD has better positions, its the desktop where they are threatened and even there chjeap sub-100$ Semprons will hold the line as I doubt we'll see Core2 chips at that price point...
IMHO AMD has 4-6 months window before really feeling pressed.

 

[Frank]

Here are some server benchmarks:

http://www.intel.com/performance/server/xeon/database.htm

Woodcrest is only 2S though, 4S Intel will have to make do with Netburst-based Tulsa.

Ok.

Let me restate my question: I would like to see some independent benchmarks of typical server applications, like running lots of queries on a large SQL server.

 

[ANova]

... than the current AMD line? Definitely. Support for DDR2 for AMD processors might change that and close the gap a fair bit.

The move to DDR2 isn't going to make any difference at all.

But I still wouldn't buy a Conroe if it's twice or more as expensive as a Sempron, especially if I had to buy expensive FB-DIMMs as well. What would be the use?

Yes well the Conroe will be quite a bit faster than a Sempron as well, up to 70%.

 

[Moloch]

The move to DDR2 isn't going to make any difference at all.


Yes well the Conroe will be quite a bit faster than a Sempron as well, up to 70%.

It makes difference.. just not much.
You dont have to say isn't going to.. the reviews are up

 

[ANova]

The difference is so small you might as well not even mention it. However, the move to DDR2 and AM2 has increased available bandwidth by alot, which should help down the line with quad core and greater chips.

 

[Moloch]

The difference is so small you might as well not even mention it. However, the move to DDR2 and AM2 has increased available bandwidth by alot, which should help down the line with quad core and greater chips.

some games benefits by a wopping 4%

 

[Accord1999]

Ok.

Let me restate my question: I would like to see some independent benchmarks of typical server applications, like running lots of queries on a large SQL server.

That's what TPC-C is, arguably the most important DB benchmark available. Results are conducted by each manufacturer who tune their systems to the utmost.

 

[SugarCoat]

some games benefits by a wopping 4%

Problem is the Latency of DD2 kills the increased bw. Thats why AMD went for DDR2 800 support rather then stopping at 667 like intel did, except for their EE chips of course.

DDR2 500 does nothing (can even be slower), DDR2 667 does about 5% on a good day, DDR2 800 and DDR2 1000 are a bit better topping out at about 10-12%. DDR3 is where its at . Solves the latency issue and continues to boost speed while running on less juice.