Nice Review, Damage despite the fact there was little to see, since it's quite obivious that the FX was going to beat the very rare P4EE 3.4Ghz.

As a sidenote I notice that, it looks like the K8 is begining to suffer from dismishing returns. The FX53 or 4000+ is almost as fast as the FX55. It's just very difficult to jusify the extra $100 to pony up the 5% performance increase that the FX55 has over the FX53. Moreso,with the 3500+ which is just half the price of the FX55 while it yields 80% of the FX55 performance.

and they both cost more money than a French diplomat can siphon off of a U.N. oil-for-food program on a slow afternoon

Ah yes, I am reminded once again why I love TR! :)

I think that AMD is possibly pushing the rating thing a bit far.....4000 for a bit of extra cache in my opinion is a bit dodgy - dont get me wrong i am an amd fan, but still, they better be careful. AMD needs to improve performance in multi-threaded apps in my opinion......read the hardocp review, and notice the comments at the end on multitasking apps - a lot of serious workstation users etc, will be using several apps at the same time a lot - i know i do, and i just use photoshop and web editing tools etc.

Still, good on AMD, my 64 3200+ oclocked is doing quite nicely......just keep those clock speeds climbing boys!! otherwise mr intel may just creep ahead.

Yeah, it's pretty much impossible for it to take 40min to get Archon up and running from disk under normal circumstances. The C64 only had 64KB memory to begin with, and not all of it could be used. A game wouldn't load a lot of data and throw it out the window (beyond say a loading screen and some music).

Say you could use 60K, in which case 40min of loading would mean a transfer speed of ~25 characters per second. The 1541 wasn't operating very fast by default (because the way it handshaked with the C64), but it wasn't _that_ slow.

Sure you weren't tape-loading? :-P

That's just broken. I loaded Archon on my old C=64 in a few seconds, not half an hour - and yes, it was on one of those 5-1/4" disks.

#71
But all that aside, congrats to AMD for establishing the Athlon 64 as THE gaming CPU. Now if they would just market that to their advantage somehow...

i think if AMD really wants to get ahead they need to start winning over more big OEM distributors, like Dell and HP. A lot of my friends who dont know anything about computers have only heard of intel cuz they only shop Dell or similiar. If AMD could get its fingers in that market they could make quite a reputation for themselves.
I think the only distributors that regularly sell AMD's are gaming manufacturers like Falcon, Alienware or Voodoo, and they dont reall sell systems in enough volume.

No SSE3 and no .90nm make Homer something something

Nice and thorough review, but given your all-out A64-recommendation to gamers, I would've liked to see a gaming reality-check of some sorts, running the games at a much higher res and maybe even with AA & AF applied.

#61, I agree, a harmonic mean seems inappropriate. I bet it's not a harmonic mean, but a geometric mean. A geometric mean is exactly what's useful for balancing diverse benchmarks. It tends to average RATIOS so for example it can detect that multiple benchmarks have, on the average, doubled in speed. You can even mix different kinds of data, like mixing an FPS measurement with something like database web transactions per second.

http://news.com.com/Photo+Barretts+mea+culpa/2100-1006_3-5417870-2....
I think he's praying, not apologizing.

What does AMD releasing a new high end CPU have to do with load times again?

Congrats on CPU review #40.

TR is the best place to get hardware reviews, bar none.

just comparing Anand;s and your graphs - it seems that you both used the same benchmark for Nero Expres 6.0.0.3 and Winzip.

However in Anands case the results are exactly opposite to TR's for Nero, so perhaps your AMD platform was not optimized enough?

I have no idea one way or anoher,

and the Winzip test, the P560 that wins your test sits in the middle of the pack @ Anandtech?

Any reasons for such big differences?

edit: just to put some numebrs with the comments

NERO:

Anand Intel: scores around 540
TR Intel: scores around 500

Anand AMD: scores around 490
TR AMD: scores around 650

In a bit different order
Winzip:

Anand FX55 - 385
TR FX55 - 407

Anand PIV560 - 412
TR PIV560 - 398

Anand 4000+ - 398
TR 4000+ - 412

The only one that scores fairly similary is

Anand XE3.4 - 411
TR XE3.4 - 417

Quite interesting differences, what could it be, drivers, different motherboards?

I think Scott is thinking about the load time for M.U.L.E....not Archon. We'd usually walk away from the machine to load that guy.

speaking of nice speed boosts, I just replaced the 4200 RPM Hitachi drive in this Emachine Athlon 64 notebook with a 7200 rpm drive and added a gig of ram. This thing screams now.

At the risk of painting myself as a statistics geek, it's important to point out that a harmonic mean is really *not* a fancy way of averaging.

Just to point it out, imagine if you have 100 and 1: the harmonic mean of the two numbers is ~2.0: notice how much closer that is to 1 than to 100. To illustrate what I'm saying, the harmonic mean of 1 and ten billion is 2.0! Another interesting point is that the harmonic mean of 1, 100, 100, 100, and 100 is 4.8. Harmonic means tend to be driven by the smallest number.

Harmonic means are really only good "averages" in very specific situations - resistors in parallel, decay rates, average speed (by distance, not by time).

WorldBench must be talking about taking the harmonic mean of the *inverse* of the times displayed. In other words - they're just taking the arithmetic mean of "tests per second" (or tests per hour, or whatever). I still consider this a little odd because that means they intend their tests to be the same level of difficulty. Given that they don't appear to run the same time on just about *any* platform, they don't really seem to be....

Hmm, must poke around a little more.

I find it interesting that a AMD's short pipeline proc is able to hit 2.8 on a 130nm process, and Intel's ridiculous pipeline 90nm Prescott will probably never hit 4ghz. OOPS! AMD's 90nm process also looks promising, Intel made the mistake of adding a few million too many transistors to the prescott, and that's the real issue with them. I suspected this and the 90nm Athlon seems to prove this.

Very impressive...

nice OC on that fx-55! fx-57 equivalent almost. AMD is definately the king

#53 LOL!

Nice CPUs, lousy prices, hope they drop fast.

I'll stick with my A64 3400+ for awhile it looks like :)

An account overdraft has occured.

Cause: Insufficient funds.

You have been issued a $30 overdraft fee.

Meh, nevermind. I just swollowed a bug...

Archon with no load times brought to you by:
Nintendo Entertainment System baby!

Dude, you should have tried loading programs off casette tape on the TRS-80 or Apple ][.... that was an exercise in patience. Thank god they didn't have more than a few K of memory.

xArchon 0.5 can be obtained here

http://xarchon.seul.org/download.html

Versions for Linux and Windows under Cygwin

enjoy

Great Review, but I would like to see more of a range of multimedia encoding, maybe with different programs. I hate to mention a "competitor", but I like how Anand has multiple codecs tested for video. I do a lot of video encoding at home and at work, and would like to see more multimedia comparisons. Maybe adding a vorbis encode to offset the mp3 as well. Just my 2 censt of course.
BTW, does anyone actually use Xmpeg? I personally don't know anyone that does, it is all GKnot or DVX.

great review as usual, but they both cost more money than an american diplomat can siphon from Iraq reconstriuction in a slow hour. ;)

My dad had one of those "portable" C64's that was like a large suitcase, with a keyboard the snapped onto the top, and about a 6" crt built in. I used to play Archon on it all the time, and I don't remember it taking more than a minute to load..

I can still remember some of the commands.. LOAD"*",8,1

crawdads? you didn't live in louisiana did you lol.

The Athlon 64 4000+ and FX-55 are easily the fastest x86-compatible PC processors available

In fact, once you take price into account, the Pentium 4s start looking quite a bit more attractive

AMD may be beating Intel in performance, but the value proposition isn't there

My, how things have changed...

Neither of these chips sells in any kind of volume, and they both cost more money than a French diplomat can siphon off of a U.N. oil-for-food program on a slow afternoon

LOL, that made my day. Nice! :)