Sign in with
Sign up | Sign in
Your question

Integrated memory controller

Last response: in CPUs
Share
February 7, 2006 1:57:03 PM

Hey guys! Given the fact that the Athlon64 cpu generation features an integrated memory controller I as wondering how much of a performance gain these procs have over previous generation procs?

Let's say we compare an Athlon XP 3200+ against an Athlon64 3200+ running 32-bit applications, what would that much difference be?

Any comments appreciated...
a b à CPUs
February 7, 2006 2:33:06 PM

Compared to the A64 series, the socket "A" Xp3200+ should really be renamed a 2500+, as usually an A64/2800+ will spank an XP3200+ across the board...

The XP series at 3200+ are even usually slower then Sempron 2800 models...

(The whole Quantispeed Rating system was rather askew with the advent of the P4 "C" models, where a P4 2.8C was spanking an XP 3200+ quite handily...; but, of course, AMD never really claimed their XP ratings were compared to recent P4 models, but rather compared them to older T-bird cores or something along those lines)
February 7, 2006 4:18:13 PM

Quote:
Hey guys! Given the fact that the Athlon64 cpu generation features an integrated memory controller I as wondering how much of a performance gain these procs have over previous generation procs?


There's no comparison really, the Athlon 64 spanks the Athlon XP's primarily because of the integrated memory controller. If you want to compare processors head to head, check out the CPU charts.

http://www23.tomshardware.com/cpu.html

-mpjesse
Related resources
February 7, 2006 5:48:44 PM

They claim that the pr rating is the equivalent of an Athlon k6 processor running at the stated "speed."
February 8, 2006 12:28:15 AM

Quote:
P4 2.8C was spanking an XP 3200+ quite handily...;


I remember that 8) I remember my AMD loving friends buying 3200+'s but my 2.8c still runs faster AND cooler. I don't understand why people always go on about P4's being hot when Prescott was only a small part of the P4 story.
February 8, 2006 12:53:13 PM

Yeah. When I built my PC years ago, I was really on the line between a Barton and a NorthwoodC. In the end it was mostly Hyper Threading that sold me, as I've been known to multitask. :lol:  There were a few other reasons too, but that was the biggest. And I don't regret my choice for a second. But I have to say, I'd probably get an A64 X2 or a DC Opteron were I building a system at right this moment.

I'm actually hoping that by the time I need to build my next system Intel and AMD both will have some awesome low-power dualcores that I can use to build a nice quiet passively cooled rig. I got my 2.6NWC passively cooled, but I'm hoping that each of the cores in my next system can outperform that.

Actually, I'm also hoping that this whole virtualization technology catches on. What I'd really love to have is a quadcore x86-64 single CPU system with 8GB of RAM running two cores for a Windows instance and two cores for a Linux instance simultaneously, each with a seperate monitor and LAN, while sharing partitions of a RAID5 array with onboard hardware XOR. I wonder what year I can finally expect to pull something like that off for three grand. :?
February 8, 2006 1:24:22 PM

The Athlon XP was in deed a good processor, specially for its shorter pipeliens which makes it execute more instructions per clock cycle (IPC) and strong floating point unit than any P4 (call it NorthWodd or Preshott).

The problem with AMD was that they've hit a wall with the Athlon XP core.

The Athlon XP 2800+ (Barton core with 166MHz FSB) had a speed of 2.2GHz. It was obvious for AMD to release the Athlon XP 3000+ at a speed of 2.4 GHz and 200MHz FSB and the 3200+ at a speed of 2.6GHz (if they ever wanted to keep up with Intel). The problem AMD faced with these cores was that the 3000+ was slower than the 2800+ (33MHz less) and the 3200+ was only operating at 2.2GHz.

If AMD would known about SOI before, the Barton core would've been more scalable against any Intel offering (but that's how life is).
!