Intel vs Amd Dual core

[clarkey01]

Originally posted by: Accord99

Originally posted by: clarkey01
Got any numbers to back that up ? reliability? why how many have you owned?, seems to me like your shooting in the dark.

It's the RAS features that are supported by the chipsets, stuff like hot-plug DIMMs and PCI(-e), memory mirroring, etc.. that make the Intel chipsets better.

There not numbers

 

[clarkey01]

Originally posted by: Accord99

Originally posted by: Markfw900
Guys the fanboy just wont' die or go away.... I don't care if the ahole doesn't believe any of us. AMD has the better solution at the moment, and he can't stand it. I am outta here.....

You're the ultimate AMD fanboy. You can't stand it because you realized you simply don't know what you're talking about.

lol check yourself before you wreck yourself :disgust:

 

[Acanthus]

Originally posted by: Jeff7181

Originally posted by: Acanthus

Originally posted by: Jeff7181

Originally posted by: Acanthus

Originally posted by: Markfw900

Originally posted by: Acanthus

Originally posted by: Jeff7181
Also remember that those two Prescott cores will be fighting for memory bandwidth. Halve a P4's memory bandwidth right now and you'll see a significant drop in performance. You're gonna be looking at the same thing if you have two processes running 100% with a dual core Prescott... each core being starved for memory bandwidth.
The difference between dual channel memory and single channel memory with the Athlon-64 isn't very significant... only a few percent with equal clock speeds and L2 cache sizes.

DDRII.

DDR2 won't help Intel. The Hypertransport idea and on-board memory controller of the A64 will wipe the floor with Intel. And dual dual-core machines will benefit even more This has been proven today with 4 and 8 way Opteron systems, the more cpu's the more they pull ahead of the Intel counterpart.

Independant FSBs on intel boards will correct this problem for enterprise. As for consumer level, im glad you know exactly how both are going to perform :roll:

I'm glad YOU know as well. ":roll:"

DDRII is NOT the solution to Intel's memory bandwidth problem. Do you understand the memory bandwidth problem? Right now the single core processors are using all the bandwidth they can get with DDR533. If you increase the speed of the RAM, say to DDR800, and have dual channels, two cores will STILL be starved for memory bandwidth moreso than a single core at DDR533 with dual channels.

To get back to the amount of available memory bandwidth per core that you're getting right now with a single core and dual channel DDR533, you will need dual channel DDR1066. DDR2 RAM that runs 533 MHz. I think it's safe to say that's not going to happen within the next year. So dual Prescott cores are going to be as starved for memory bandwidth as the 400 and 533 MHz FSB Pentium 4's.

You are simply wrong, 1 pentium 4 core, even at 3.9ghz (in my case) does not benefit from going from DDR400 to DDR520 shows no performance increase, bandwidth is not a problem on a single core.

DDRII will be utilized to its fullest on 2 cores, because 2 cores NEED MORE BANDWIDTH.

Intel still has another way out, a la rambus.

I guess Wesley Fink is simply wrong as well since his tests seemed to show quite a nice improvement by going from DDR533 to DDR667 even at 2.8 GHz.

No, you read the charts wrong, the CPU clockspeed is the reason for the perofmance increase, not the memory.

533DDR2
200x12 - 2.4GHz
800FSB

600DDR2
200x12 - 2.4GHz
800FSB

average of 1%-2% increase
-------------------------------------
533DDR2
266x12 - 3.2GHz
1066FSB

711DDR2
266x12 - 3.2GHz
1066FSB

average of 1-2% again.
-----------------------------

Other than those 2 comparisons, all the others are invalid because of enormous clockspeed differences on the cpu. OF COURSE YOU WILL SEE AN INCREASE WITH 50% MORE CLOCKSPEED ON THE CPU.

 

[Jeff7181]

Ahh, I see now.

*EDIT* Would be interesting to see some slower DDR2 tests to see at what point it starts to significantly effect performance.

 

[piroroadkill]

Some of the ignorance in this thread is truly, truly stunning.

 

[Gamingphreek]

Seems like we have a FelixDaKat wanna be.

Tell me, Accord, why would there be no such thing as 8-way Opteron systems, yet they have an 8xx series opteron?

Additionally how about some facts about this "RAS" . Right now you are comparing Intel chipsets to AMD chipsets. It is a lost battle. You should be comparing them to the Nforce Professional or something. How can you call us fanboys when you have sighted no sources to back up your claims.

SIght some sources im sure we all would like to read this information.

-Kevin

 

[Gamingphreek]

Originally posted by: Accord99

Originally posted by: Ged
8-way system

Verari Ships 8-way Opteron Systems
Apparently iwill and Verari think it's worthwhile.

These look like 2 4-way boards connected together. And it doesn't look like any of the major server OEMs who control the vast majority of the market think it's worthhile.

It doesn't matter what it looks like, it matters what it is.

-Kevin

 

[Jeff7181]

Originally posted by: piroroadkill
Some of the ignorance in this thread is truly, truly stunning.

Woohoo... post count +1... epenis +2!!!

 

[imported_Ged]

Originally posted by: Accord99

Originally posted by: Ged
The Problem with the major server OEMs is that most of them have contracts or deals with Intel currently and rather than risk that contract or deal they don't push the AMD platforms. That doesn't say that AMD isn't as good and Intel is better.

What about Sun? Sun has no relationship with Intel and their SPARC systems are not competitive with Power and Itanium.

Sun seems to be pushing AMD Opteron based servers. Fits in with how things seem to be going. No relationship with Intel, you choose AMD because it's cheaper and performs better.

As for OEMs and 8-way servers: HP Says Farewell to 8P Intel Xeon Servers

"Sun Microsystems is looking forward 8-way machines based on AMD Opteron processor."

 

[imported_Ged]

This thread seems dead, but I saw this today and thought of this thread.

Talks about SuperMicro making an AMD Opteron based board and has a link to a REPORT (PDF) that someone at Fermi Labs did comparing AMD Opteron to Intel Xeon.

It gives the thumbs up to the Opteron saying that the reliability of Opteron based systems is adequate for Fermilab applications "and the power consumption" is "significantly less than their Intel counterparts".

This is the sort of thing that will kill Intel Xeons: Power Consumption.

Fermi Labs only tested up to the AMD Opteron 250. I guess the AMD Opteron 252 would have really blown them away because it uses Less Power than the 250 because of the newer process.

 

[ribbon13]

Yeah, too bad the 252 is the only E4 stepping Opteron readily available.

 

[DAPUNISHER]

Good stuff Ged, thanks :beer: SuperMicro+Opteron=:heart:

 

[jbh129]

The preview of the 2.8D is very impressive for encoding, video creating and the like. I dont think AMD (even with dual cores) can compete at this pricepoint.

 

[Infernus]

With the A64 3500+ taking over nearly every single threaded tests, it's really going to be interesting with Dual Core A64s and Intel's answer to them.

 

[Gamingphreek]

Remember, AMD does not have HT and has a much shorter pipeline. While without HT they do much much better than Intel it is still only one and one thread. Moving to single core will bring a much more significant impact.

I would think that AMD will hold a sizable lead... especially since they have plenty of bandwidth and whatnot.

-Kevin

 

[Markfw]

Originally posted by: jbh129
The preview of the 2.8D is very impressive for encoding, video creating and the like. I dont think AMD (even with dual cores) can compete at this pricepoint.

What pricepoint ? they aren't even available for purchase yet ! And when the AMD dual-core does come out, thats when you can compare them, and the Hypertransport and integrated memory controller should kill the Intel dual-cores..