Opteron 144 slower than 3.2GHz P4

[wkulecz]

Looks like its gonna need 64-bit specific code to be cost effective.

http://www.extremetech.com/article2/0,3973,1217389,00.asp

--wally.

[Jkun]

This has been expected since the first benchmarks came out. I fear that the Athlon64 will become AMD's Pentium Pro...

J-kun

[Ssin]

... add to that the fact that AMD can't afford to lose this round.

[dZeus]

I've heard that the current 64-bit compilers from AMD don't do any wonders yet, producing code often slower than ones produced with Intel's 32-bit x86 compilers.

Hopefully they'll be able to optimise the compiler a LOT soon... AMD sure needs it.

[wkulecz]

IMHO if Opteron can't do as well as P4 on existing 32-bit code, there had better be some really fantastic 64-bit stuff on the horizon very soon or AMD will be toast on the desktop.

Who's going to put out special "re-compiled for Opteron" applications?

Most people bought the 386 because it ran 16-bit code faster than the 286 did (in addition to its DOS multitasking capabilities with Desqview along with a few 32-bit "DOS protected mode" apps).
--wally.

[Umfriend]

One thing to consider is that it may still run 32-bit applications (not recompiled) faster than intels 64-bit acrhitecture.

I assume developers will find it easier to work with a architecture they know already (x86), just extended than a full new one.

One the other hand, I have not kept track at all of IA-64, so I may be bulldropping.

[KRSESQ]

Looks like its gonna need 64-bit specific code to be cost effective.

Agreed. These tests seem somewhat lacking.

Are we to presume from these tests that an Opteron running at 1.8 GHz at 64-bit ought to be roughly equivalent in performance to a 32-bit P4 running at 3.2 GHz?

That being so, was the WinXP Pro version they used fully 64-bit aware? It doesn't say. If not, that alone might account for much of the disparity.

MORE TESTS, DAMMIT!

Kevin

[wkulecz]

No the review was straight up Opteron on existing 32-bit code and OS.

My point is, the Opteron cost more than the 3.2GHZ P4 and performed significantly worse. This makes for a very tough sell until some compelling 64-bit stuff is available.

The only 64-bit compelling thing for me is digital video editing/encoding. 32-bit compatability means very little if IA-64 gives much better bang/buck for my "killer app". IA-64 vs Opteron pricing and performance will be the key, its simply too early to pick a dog in this fight.

But this set of tests pretty much takes Opteron off the table for my next upgrade to run 32-bit digital video editing software.

--wally.

[Elie]

I too don;t think people should jump the gun until 64 bit applications and OS's start showing up with great benchmark results.

Cheers,
Elie

[Ray Austin]

yeah gize chill out a little...

the opteron is a server cpu... so these tests are bull!

the A64 is different from the opteron less pins plus other things. Although it too can perform 32bit operations.

I am sure you'd have the same problem comparing an Itanium to a P4 that is if Itanium could even do 32bit code... which it can't without emulation software.

[Chrono_Wanderer]

Ok, forget IA64, its a total failure

x86-64 seems promising, but only if AMD can get the speed up!

I am looking foredward to 64bit... just like 286-386

[bsdgeek]

IMO, Hammer just needs some clock scaling to be competitive. I have a feeling (whether it will ruin AMD or not is the question) that current Hammer will be the Williamette, and the .09s will be the Northwoods.

[Brian Ellis]

Although MS are working on Win64, it will probably be years before it is fully converted. Just think how long it took for a full 32-bit OS, without many 16 bit components, came out. W95 had a partial 32 bit kernel but most everything else was 16 bit. This improved somewhat with W98 but, unless I'm mistaken, W2k was the very first all-32-bit Win version, even though the first 32-bit capable Pentium was marketed in, I think, 1992 (please don't flame me if it was 93, I'm having a senior moment).

With MS, history has a tendency to repeat itself. Maybe this is why Intel do not plan to introduce a 64-bit "Pentium" for whatever for workstation use before 2010, even though they have a design already up their sleeve. This is probably a clever strategy to let AMD bite the dust!

[Fluff]

Also opteron has been designed to be multi processor! so i wouldn't be too fussed over the single processor benchmarks. I'll probably get one once pci express makes a showing. And they have moved to .09 process. Which will probably be 2nd quarter 04.

And when a dual processor ATX solution appears!

[Dr Mordrid]

Can't....violate....damned.....NDA.....CRAPPOLA!!

Dr. Mordrid