Here is an interesting Xeon vs. Opteron review

Goatsan

Member
May 30, 2003
123
0
0
they might fall back again, which wouldn't be a surprise comming from amd, but you never gave thought that the opteron is a 64bit cpu, which can do alot of difference.
 

Socio

Golden Member
May 19, 2002
1,732
2
81
According to this article the Nocona will arrive in June it also hints that the 3.2 GHz Xeon may be the end of the Xeon line because they are having trouble at 3.4 GHz with the Northwood process.

Link
 

Megatomic

Lifer
Nov 9, 2000
20,127
6
81
I'm not really heavy on server benchmarking so that was a difficult read for me. But the summaries and conclusion were easy to read. Things look good for the Opteron.
 

alexruiz

Platinum Member
Sep 21, 2001
2,836
556
126
Originally posted by: lookouthere
Opteron will lose if they don't use 90nm process next summer or before that.

Check out thise.....Nocona based on 90nm with 800FSB
This could throw Opteron into trash.....who know?? If the Opteron up the FSB to 1000, it throw the Xeons in there(trash)

How many times it has been said that the opteron DOESN'T have a FSB?
 

Sahakiel

Golden Member
Oct 19, 2001
1,746
0
86
Originally posted by: alexruiz


How many times it has been said that the opteron DOESN'T have a FSB?

A cpu without a FSB must have a pretty damn hard time fetching from memory...
 

Jeff7181

Lifer
Aug 21, 2002
18,368
11
81
Originally posted by: Sahakiel
Originally posted by: alexruiz


How many times it has been said that the opteron DOESN'T have a FSB?

A cpu without a FSB must have a pretty damn hard time fetching from memory...

There is no Front Side Bus as it exists right now. Right now, the FSB is the link between the CPU and the northbridge. The memory controller is part of the northbridge. With the Opteron and Athlon-64's, the memory controller is now part of the CPU, so the CPU has a direct connection to the RAM (doesn't need to talk to any other chips to access RAM), which is one reason for the increased pin count.
 

jhu

Lifer
Oct 10, 1999
11,918
9
81
i want to see benchmarks showcasing the opteron, itanium2, and ppc970. when's this going to happen? if i had the money, i'd do it myself
 

alexruiz

Platinum Member
Sep 21, 2001
2,836
556
126
Originally posted by: Sahakiel
Originally posted by: alexruiz


How many times it has been said that the opteron DOESN'T have a FSB?

A cpu without a FSB must have a pretty damn hard time fetching from memory...


A very nice smooth time as shown by the benchmarks..... ;)

Jeff already explained the main difference, but a reading of the AMD64 and hypertransport white papers wouldn't hurt.
 

alexruiz

Platinum Member
Sep 21, 2001
2,836
556
126
Originally posted by: Acanthus
i want to see AMD-64 come up against the Itanium 2 design.

What does it have to do with the opteron-xeon topic......?

addresing yoiur question, exactly how.....? An itanium 2 box will beat an opteron is single and dual configurations with properly compiled software..... for 4 and 8 way machines, I wouldn't be that sure..... not to mention the huge difference in price.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: alexruiz
Originally posted by: Acanthus
i want to see AMD-64 come up against the Itanium 2 design.

What does it have to do with the opteron-xeon topic......?

addresing yoiur question, exactly how.....? An itanium 2 box will beat an opteron is single and dual configurations with properly compiled software..... for 4 and 8 way machines, I wouldn't be that sure..... not to mention the huge difference in price.

because people keep playing "AMD will take the lead for 64-bit" card, and that doesnt apply in the server arena because thats where the Itanium is marketed. Intel specifically markets the chip for high end servers and even says on their own website that Itanium is better than Xeon for high end servers, offering a 30-40% performance increase for a 10% increase in cost.

Opertron 148 - 1405 Int 1393 FPU
Itanium 2 1.5ghz 6MB L2 - 1332 Int 2119 FPU
 

alexruiz

Platinum Member
Sep 21, 2001
2,836
556
126
Originally posted by: Acanthus
Originally posted by: alexruiz
Originally posted by: Acanthus
i want to see AMD-64 come up against the Itanium 2 design.

What does it have to do with the opteron-xeon topic......?

addresing yoiur question, exactly how.....? An itanium 2 box will beat an opteron is single and dual configurations with properly compiled software..... for 4 and 8 way machines, I wouldn't be that sure..... not to mention the huge difference in price.

because people keep playing "AMD will take the lead for 64-bit" card, and that doesnt apply in the server arena because thats where the Itanium is marketed. Intel specifically markets the chip for high end servers and even says on their own website that Itanium is better than Xeon for high end servers, offering a 30-40% performance increase for a 10% increase in cost.

Opertron 148 - 1405 Int 1393 FPU
Itanium 2 1.5ghz 6MB L2 - 1332 Int 2119 FPU

Are those figures for a 4 way machine or a single CPU one? As I posted, in a single CPU the itanium will be faster with very properly compiled software..... give me the score for a 4 way or 8 way machine versus a comparable opteron and I will believe it.

If you think that an architecture that NEEDS a chipset sharing a FSB with all the CPUs AND managing the memory can scale better than one with integrated memory controller per CPU AND dedicated paths to inteconnect the CPUs, then you need to revisit your concepts. For medium and high end servers, the bottleneck is data transfer.

Regarding the SPEc results, even if the FP number of the itanium is higher, most of the database/SQL or related server applications benefit more from INTEGER performance, right?

Do you have any numbers on 4 way or higher configurations for REAL server applications?

"For medium and high end servers, the bottleneck is data transfer" Do you think it is coincidence that what will be the most powerful supercomputer is based on Opteron?

Keep in mind that the compilers for AMD can only get better.....