Athlon64 Motherboards And Chips Compared

An anonymous reader writes "Just noticed that OverclockersClub has a new article (free, no reg, blah blah blah) that describes the AMD64 processors. The article talks about the differences in each processor and compares them as well as puts everything in a nice easy to read chart. Pretty nice article if you aren't familiar with all the new tech." Makes a good match for Johnny-boy's submission. He writes "HardwareZone has a 46 page article out that compares many of the Athlon64 motherboards out on the market now. If you are planning to get that Socket-754 motherboard, maybe this article is worth a look."

Comments

Don't get socket 754

[DarkHelmet]

Now really isn't the time to get an Athlon.

The 939 pin athlons are just around the corner, which is the migration path of most of the athlon sets.

754 series sets will still only have a single channel 128 bit pathway. It's not worth it.

Wait until the 939 pin, and get dual channel memory transfer in a non-FX Athlon64. Even if you're only getting half the cache (1 meg vs 512kb) on the 939 pin versions, chances are you will be able to overclock it more because it's a smaller die space.

46 pages... I wanted a motherboard review, not a dissertation :)

Re:Don't get socket 754

[lakeland]

And lemme guess, just around the corner from 939 pin athlons is ...

C'mon, we all know that the week we buy the latest gizmo it will be obsolete.

Re:Don't get socket 754

[eyempack]

Unfortunately it never seems the "right time" to buy any chip. You buy a specific chip then they change the die size and you have a coffee heating device as a chip, or they update the core, or they bump up the speeds. It all comes down to if it's mission critical at this point and if you need a machine right now. And as far as bang for your buck, depending on the applications you are running AMD still keeps the lead...well sometimes

Re:Don't get socket 754

[OverlordQ]

Humm I suppose I can't break off a pin to make my 940 Athon a 939 now can I? :)

Re:Don't get socket 754

[vollmerk]

Everything always gets outdated... I'm still holding out on getting a car. I figure I'll wait for one of those quantum teleporters...

Re:Don't get socket 754

[Anonymous Coward]

Unfortunately it never seems the "right time" to buy any chip.

Now is a great time to buy the Athlon 2000 XP (plus or minus a few hundred mghz depending on needs) as they are great performers for the price, and at roughly $50 for the cpu you cant go wrong.

Re:Don't get socket 754

[steveha]

I used to worry about sockets and the future.

Then I noticed that I never swap CPUs out anyway. Motherboards are cheap enough, I swap an entire motherboard with its CPU. In fact, usually I swap out entire computers.

Since we use all our computers, I usually build a complete new computer, get it working, swap it for the older one, and keep the older one handy for a while as a hot spare in case something goes wrong with the new one. Then later I find a good home for the older computer.

(Now that I'm buying Lian Li aluminum cases, I'll probably start swapping motherboards into cheaper steel cases, and putting new motherboards into the Lian Li case.)

But anyway, I might get a socket 754 motherboard and chip. It will outperform any computer I currently own, and it should have adequate horsepower to play Half-Life 2 and Doom 3.

steveha

Re:Don't get socket 754

[Fat Jedi Kid]

You youngin's with your pins and zif sockets back in my day we had to solder the chips onto breadboards. Using the heat from friction coming off of our fathers leather hitting our hands.

Re:Don't get socket 754

[Sivar]

Socket 745 Athlons have a single 64-bit memory bus, not a 128-bit memory bus. (probably just a typo)

In any case, it is important to remember: Athlons are not Pentium IVs. Athlons do not have the performance hit that P4s have with lower bandwidth. Currently, very few applications care whether you have single or dual channel memory--the performance difference is in the low single digits. After Athlon64s significantly ramp up in clock speed, we wil begin to see a greater advantage of having more bandwidth, but not before.

Also, I wanted to note that currect 512K Athlon64s DO NOT have a smaller die space. They are more or less 1MB chips with half the cache disabled. Future revisions will actually cut out the cache, but for the time being AMD needed to market a cheaper Athlon64, and didn't have the time or money to modify manufacturing equipment to manufacture a third completely different die. That said, die space doesn't directly have anything to do with how overclockable a chip is.

Re:Don't get socket 754

[ctr2sprt]

The difference is that, because my computer uses all the same standards as bleeding-edge 32-bit equipment, I can upgrade all the parts and make it un-obsolete. With Athlon64s, however, it's much less clear that you'll be able to do that. That's all we're saying. Even moreso than when buying a regular computer, be aware that your Athlon64-based one may become difficult, expensive, or simply impossible to upgrade. And it may happen very soon after you buy it.

If you're willing to live with the increased risk, fine, go for it. If I felt any need to upgrade my computer, I'd probably fall into the group that's willing to risk true obsolescence. But a lot of people want as much security as possible, and those people should probably stay in the Land of 32-bit Words.

Re:Don't get socket 754

[JBv]

I only buy hardware when I *really* need it. If I can wait for the next generation of components, then I really don't need an upgrade.

When I find that my CPU is underpowerd for my usage, I buy a new CPU, motherboard and memory and re-use as much of the older PC components as possible.

My 900Mhz Duron is doing q3a and ET just fine with it's 512 of PC133 and 3 hard drives (3+2+15 Gb) in lvm. The 3+2 Gb drive was recently replaced by an 80Gb drive (no lvm), because I *needed* the extra space for my data and I was no loger trusting the older drives.

Re:Don't get socket 754

[PReDiToR]

Have you never built a computer for someone, and put some of your known-good components into the machine, whilst upgrading at the same time?

I find this to be the single most efficient way to keep on top of technological evolution.
Someone wants a PC from me, they get a KT400 and AthlonXP, I get an Athlon64 and mobo to replace it. RAM, video and HDDs stay here until I need faster parts, or in the case of HDDs, they get dumped for being too small.

Re:Don't get socket 754

[Anonymous Coward]

> be aware that your Athlon64-based one may become difficult, expensive, or simply
> impossible to upgrade. And it may happen very soon after you buy it.
> If you're willing to live with the increased risk, fine, go for it.

I seem to be able to make each PC I build last 4 or 5 years, so I don't care what happens after I build it. Why upgrade when even a year later the same powered CPU or Motherboard is available for less than half price and "better" ones are available for the same price?

Re:Don't get socket 754

[Anonymous Coward]

Well.. There's PCI-X, for instance.

No matter how you view it, an Athlon-64 bought today just isn't future proof.

Unless you know for sure you'll make use of the 64-bit capabilites, you should just go with a cheaper Socket-A setup..

Re:Don't get socket 754

[Dan Ost]

Upgradability was a problem back when software demanded more than hardware
could provide, but now days, any computer you buy will have a processor
sufficiently powerful to be useful for the majority of needs. Also, computers
are so cheap that it rarely makes sense to put money into an older machine
when a newer, more powerful machine is available for about the same price
that it would take to upgrade the older machine.

Of course, special needs require special hardware considerations, but that
will never change.

Re:Don't get socket 754

[MarcQuadra]

Sorry, but after the initial shakedown, AMD is pretty stable in terms of their sockets and backwards compatability. I've had very recent Athlon Bartons running in boards several years old, KT-133 based boards. Hell, my main workstation is a KT-266a from 2000 with an Athlon-XP 2500+ in it. The FSB for the CPU is rated at 333 and I run it at 266, but it still WORKS. It's not like Intel, where every CPU upgrade has you shopping for a new mobo, power supply, and heatsink. I think Socket-939 will be around for a LONG time, like socket-7 and socket-a were.

Re:Don't get socket 754

[Anarke_Incarnate]

PCI X is available and has been for some time. I think you are confusing it with PCI Express, which is the new standard.

Re:Don't get socket 754

[Anonymous Coward]

+ The lifespan of Socket A was more of a historical accident rather than AMD's master plan.

+ Running a new Barton in a crippled old SDRAM board is silly.

+ 99% of new CPUs get sold to OEMs for use in new motherboards -- there really is no significant upgrade market. Neither AMD or Intel care about the tweaker crowd that upgrades CPUs every 6 months.

+ Intel has changed their sockets, but their current mobos are 1000x better than the original P4 models. Getting a 800Mhz bus with dual DDR is worth a new socket.

Re:Don't get socket 754

[Jeff DeMaagd]

Actually, Anand's has shown that the single memory channel isn't much of a hindrance on AMD's chips, or rather, dual channel isn't very beneficial yet, at least on the 64 bit chips.

I think the current biggest benefit of dual channel on AMD is being able to run more memory.

Re:Don't get socket 754

[GigsVT]

I'm not sure what you mean about upgradability. It uses standard RAM.. standard PCI cards, standard case/power supply.

The only thing that might not be upgradable is the CPU/Motherboard independantly. That's not really much loss.

I usually change motherboard when I change CPU anyway.

Re:Don't get socket 754

[orim]

"Upgradability was a problem back when software demanded more than hardware
could provide".

Yeah, no kidding. Used to be the software was written "for the next generation of hardware" (or the next CPU, that would make it run "just right"). Back then we didn't know because the machine we had ran program X the fastest we've seen up that point. We didn't think that it'll run in 10 seconds with the next processor, since the current one runs in in decent 20, compared to the last CPU (which ran it in 40).

These days, hardware has way outpaced the software's ability to use it... I mean, when's WinXP for 64-bit Athlons coming out? This summer/fall or something? Why would you even consider buying hardware before the software is even available? Either wait, and the hardware prices of what you want today will drop, or get something cheaper (and in this case 32-bit, and save your money!)
As it is, you'll just get angrier every day "your" software isn't shipping.

Re:Don't get socket 754

[Anonymous Coward]

Of course you can. Whether the CPU will still operate, however, is anyone's guess.

Re:Don't get socket 754

[Hoser McMoose]

What in the hell are you talking about?! The 64-bitness of these chips doesn't change upgradability in any way, shape or form!

It's not like you can take any old Pentium 4 motherboard and drop the latest and greatest processor into it. The first P4 boards were socket 423, then came the socket 478 boards that only supported 400MT/s bus speeds, than the 533MT/s bus speed boards. None of these are capable of supporting the current 800MT/s bus speed P4s, even if they share the same socket, and they certainly aren't going to support the LGA775 P4's that Intel will be introducing in the next 3-6 months.

If you want to buy a new processor you almost always need to buy a new motherboard and new memory to go along with it, particularly if you're talking about upgrading more than a year after the initial purchase. The Athlon64 doesn't change this one bit. Current Socket 754 boards will be supported for about another year with upcoming chips, socket 940 boards will be supported for at least that long with Opteron chips (which are the same price as the Athlon64 FX), and the first batch of Socket 939 boards probably won't support new chips produced more than a year into the future anyway.

Re:Don't get socket 754

[Mr. Piddle]

This is why a new home-built computer should use a middle-of-the-road CPU and have exactly half of its RAM slots filled. Then, that computer is not only cost-effective at the time of purchase, but it has a single guaranteed RAM and CPU upgrade down the line.

Remember that CPU pricing is non-linear, where the current top-of-the-line generally has a very stiff price premium. One thing I did a while ago was to chart the price/MHz of a particular line of CPUs, then I bought the CPU right at the top of the linear range before the curve upward began.

Re:Don't get socket 754

[rosewood]

That is why I went with an Athlon64 Notebook instead

http://www.elrosewood.com/archives/000014.html

Re:Don't get socket 754

[Bedouin X]

Why would you even consider buying hardware before the software is even available? Either wait, and the hardware prices of what you want today will drop, or get something cheaper (and in this case 32-bit, and save your money!)

Could it be the fact that the A64 is still a best of breed 32-bit system as well?

Re:Don't get socket 754

[steveha]

Have you never built a computer for someone, and put some of your known-good components into the machine, whilst upgrading at the same time?

Of course I've sometimes swapped some parts around while upgrading. But since we usually don't upgrade our computers that often, by the time we do want a new computer it's easiest just to swap the whole computer.

steveha

Re:Don't get socket 754

[shamino0]

These days, hardware has way outpaced the software's ability to use it... I mean, when's WinXP for 64-bit Athlons coming out? This summer/fall or something? Why would you even consider buying hardware before the software is even available? Either wait, and the hardware prices of what you want today will drop, or get something cheaper (and in this case 32-bit, and save your money!)

Gee, this sounds familiar.

How many years were 286 systems shipping before anybody shipped a mainstream OS to take advantage of it? When OS/2, Xenix and other 286-savvy operating systems shipped, how many of us chose to stick with MS-DOS and use the chip as a fast 8088?

How many years were 386 systems shipping before anybody shipped a mainstream OS to take advantage of that? How many years were 386's and 486's available before the market finally decided to move from DOS to Win95 and WinNT?

So I don't think it's at all strange if the market ends up using 64-bit processors as little more than fast 386's (the way most use Pentium-class systems right now) for several more years before finally deciding to use a 64-bit OS. And this is the way it should be - the average user has more than enough processing power to do everything he needs - e-mail, web surfing, MS Office, etc. Why should they switch to something new and relatively unproven just because it will take better advantage of their new hardware?

Us /.'ers who like to live on the bleeding edge may decided to move now (probably using some flavor of UNIX), but the rest of the world will take much longer to make that switch. This is the way it's always been, and I see no sign of anything being different this time around.

Re:Don't get socket 754

[Lost Race]

C'mon, we all know that the week we buy the latest gizmo it will be obsolete.

That's why I save time and money by only buying stuff that's already obsolete.

Re:Don't get socket 754

[Lost Race]

(Now that I'm buying Lian Li aluminum cases, I'll probably start swapping motherboards into cheaper steel cases, and putting new motherboards into the Lian Li case.)

Good luck squeezing your new BTX motherboard into that Lian Li ATX case!

Re:Don't get socket 754

[dellis78741]

Also, the AMD-compatible Intel 64-bit processors are just around the corner: http://news.com.com/2100-1006_3-5150336.html?tag=n efd_top They will demo a 64-bit Xeon next month at IDF. The new 64-bit instructions will be referred to as 'CT' (ala HT) but the important thing is that they are binary compatible with AMD's current x86-64 instructions. Now, as soon as they get past the 'demo' stage and release some chips, we can expect Microsoft to cough up its' 64-bit Windows for AMD64.

Well.....

[agent dero]

I was actually looking into Athlon 64's today; and i'm not seeing the price benefit compared to a PowerMac G5.

Right now, there's no GREAT 64 bit OS out there (linux, forget XP 64bit) I think we should treat Athlon64 like MacOS 10.0 (sorry, i'm a mac guy) for early adopters only

Give it another 6 months, then it'll be a great server/workstation solution

Re:Well.....

[10Ghz]

Well, Athlon64 does give you kick-ass performance. And it does so even if you run it in 32bit-mode. How is that different from G5? MacOS X is a 32bit OS as well. If you want to straight comparison of G5 on MacOS X (64bit CPU on 32bit OS), comparison to A64 on 32bit Linux of Windows would be suitable. Of course, you can run 100% 64bit system in Linux for example.

Re:Well.....

[steveha]

And I was actually looking into buying some oranges today, but I'm not seeing the price benefit compared to applesauce.

Or, to put it less obliquely, that's a strange comparison. A PowerMac G5 is for someone who wants a Mac. An Athlon64 motherboard is for... well, not someone who wants a mac.

Hope this helps.

P.S. The Athlon64 actually offers great price/performance in plain old 32-bit mode. It gets even better in 64-bit mode, but there's no reason to wait for ready availability of 64-bit software. Just as there's no reason to hold off on buying a G5 for a fully 64-bit MacOS.

steveha

Re:Well.....

[Moraelin]

I'll give you a price benefit all right. I've built an Athlon 64 system, with a Radeon 9800 XT and 512 MB DDR400 RAM. I did keep most of my old computer components, like hard drives, DVD drive, etc.

It was less than _half_ the price of an 1600 MHz G5 Mac with a Radeon 9800 Pro (i.e. previous generation), 512 DDR 333 RAM (yep, slower), a smaller hard drive, etc.

Even after changing the Mac's DVD writer to a DVD/CDR drive, it still stayed more than twice as expensive, and offering far less horse power. Go figure.

And if I'm to factor in the cost of buying all my software again, if I were to "switch"... well, you get the idea.

So there you go. Maybe you can't see it, but half the cost for _more_ power, sure looks like enough of a price advantage to me.

Re:Well.....

[markus_baertschi]

While there is no production quality 64bit OS for the AMD 64 ready today, there is a whole bunch coming up.

In three years, when it's time againt to buy something new, I prefer to have an AMD 64 as my second computer than a Pentium. The AMD 64 will allow me to run leading-edge stuff, while the Pentium will not. If, for whatever reason, the 64bit stuff is not important, it will make a fine 32bit machine also.

The AMD 64 CPU's give me a 64bit option today I don't have with any Pentium. The interesting part is that this option comes with no price or performance penalty, quite to the contrary, price/performance tends to be better in the AMD camp.

Markus

Re:Well.....

[Anonymous Coward]

Ahhh.. so now your earlier post to this story makes sense!

Penis envy. You have a crappy cheap-ass homebrewed computer instead of a nice shiny G5.. never mind, someday you'll be able to afford one instead of making do.

Re:Well.....

[Moraelin]

This "crappy cheap-ass homebrewed computer" can run more than an order of magnitude more games than that "shiny G5". So no, thanks. I don't need, nor want a G5.

But, hey... I know that some finer points, like "a computer is only useful if it runs the software _I_ want to run", are lost on most Mac fans.

Re:Well.....

[Junior J. Junior III]

Give it another 6 months, then it'll be a great server/workstation solution

Not if enough early adopters don't adopt RIGHT NOW. If you warn everyone off buying the technology today, who'll be your early adopters? Somebody's got to show up to prove there's a market, otherwise AMD64 will be just like Itanium or Alpha -- a good idea that never really caught on. Obviously, it's going to get better over time as the technology matures. And maybe the smart thing for some people is to wait for that maturity. But AMD needs customers today, and if you want to be able to be their customer today, SOME people are going to need to support them now.

Re:Well.....

[Anonymous Coward]

Right now, there's no GREAT 64 bit OS out there (linux, forget XP 64bit) I think we should treat Athlon64 like MacOS 10.0 (sorry, i'm a mac guy) for early adopters only

Well, the exact same 64 bit processor will be great as soon as the shiny 64 bit OS comes out. MacOS 10.0... you'd had to have gotten a different one, ne?

Mind you, it's a lot harder to patch a processor, but I think my point is still valid.

Re:Well.....

[billsf]

Well..... I think you are very wrong. Apple runs the G5 in 32bit mode and one amd64 easily runs circles arround two G5's. (of the garden variety, as those Apple uses)

If you need a real 64bit OS for amd64 try FreeBSD-5.2 if you are waiting for Microsoft, it will never happen. There are a number of really good 64bit OS's out there for other platforms like SPARC and Alpha. I've been using 64bit machines for years and they have total advantage over 32bit systems. If you use allot of commercial software then running a 64bit processor in 32bit mode is a waste. Please note that many commercial outfits want there applications ported and are quite willing to hand over their source with a "NDA". In the mean time there isn't that much missing in 64bit applications for Unix and Linux will certainly get it together for this processor. This is likely to be the 'PC' for the next several years, Microsoft or not.

If you believe in benchmarks, a single amd64 running at 2200MHz gives a performance rating of atleast 5400 and the best the i386 can ever do is 3400 as there some really nasty technical limits on how fast a i386 can run. I like the amd64 so much I'm going to get more. A dual or quad Opteron is most likely my next computer purchase. No more mousing arround with i386!

Re:Well.....

[Anonymous Coward]

Not if enough early adopters don't adopt RIGHT NOW. If you warn everyone off buying the technology today, who'll be your early adopters? Somebody's got to show up to prove there's a market, otherwise AMD64 will be just like Itanium or Alpha
Doesn't sound all that bad to me. Go RISC/FISC, go!

Re:Well.....

[Anonymous Coward]

The user friendly Mandrake Linux is out with a AMD64 version of Mandrake Linux 9.2.Mandrake 9.2 for AMD64

Fedora has a testversion for 64bit AMD. The future versions will support AMD 64bit. My dad runs Fedora. Fedora Core 1 for AMD64 test1

Gentoo has been supporting 64bit AMD for a while. But Gentoo is not yet for ordinary endusers. Gentoo Linux AMD64 Development

Re:Well.....

[hendridm]

> I was actually looking into Athlon 64's today; and i'm not seeing the price benefit compared to a PowerMac G5.

Seems to me that's like saying "I wanted to buy a dump truck, but am not seeing any price benefit compared to a back hoe". Why not buy the best tool for the job? Personally, price doesn't come into play for me when comparing Macs to x86 hardware because I personally hate Macs. However, to someone who does video editing or prefers the architecture, they probably wouldn't buy a Pentium even if it were much cheaper.

If I was a Mac afficionada, I wouldn't use a Windows machine because it was cheaper. I suppose if you're a Linux geek, which both platforms can run, the argument is a little different, but at least with x86 you're open to a lot more choices of distros.

Or are you just another Mac zealot spreading The Good Word wherever possible? If so, have you tried Gentoo? You might like it :)

Re:Well.....

[yarbo]

You can run Linux on a G5 or an Athlon 64.

Re:Well.....

[DarthTaco]

The Athlon64 actually offers great price/performance in plain old 32-bit mode. It gets even better in 64-bit mode

Most of us have seen the benchmarks showing the athlon 64 running very nicely in 32bit mode, but didn't we just have an article posted on slashdot saying 64 bit binaries are usually slower than the 32 bit counterpart?

Re:Well.....

[steveha]

That article was comparing SPARC 64-bit binaries to SPARC 32-bit binaries.

The Athlon64 (and all AMD 64-bit processors) have additional features that are only available in 64-bit mode. In particular, there are extra general-purpose registers available in 64-bit mode. Since x86 is starved for registers, the extra registers help speed up your programs.

This isn't just theory; the benchmarks show that the AMD 64-bit processors are faster in 64-bit mode.

steveha

Re:Well.....

[DarthTaco]

you sound knowledgable about the amd processors... I was talking with someone today that asked me what the difference was between the opterons and the athlon 64. I didn't know, my guess was it was a similar situation to the p4-xeons and regular p4s. Is that close, or am I way off base here?

Re:Well.....

[Nexx]

Ya know, that last sentence of yours is ironic when you toss it with your first :)

I don't run games, I run Adobe products. I don't need an A64, I need a mac. It's not about absolute performance, but also about how the UI *feels*.

Re:Well.....

[steveha]

Athlon64 chips and Opteron chips use the same core, so they will run the same software. Some versions of Opteron support SMP, but no Athlon64 chips support it. The various chips have various amounts of cache, various memory configurations they will work with, etc. For example, the socket 754 chip works with plain old DDR400 RAM, but only has one memory channel.

If you want the specifics, I suggest you use Google. Here's a good article to start with:

http://www.theinquirer.net/?article=10954

steveha

Re:Well.....

[Moraelin]

As I've said in countless other messages: the right tool for the right job. If the Mac is the right tool for you, hey, good for you. I don't have anything against Macs. It's just another computer, and just another OS.

What I _am_ against is:

1. Fanboy advocacy. The kind where if Joe Average asks for advice about _anything_, he gets religion instead of advice. Some fanboy's priorities are pegged at "must recommend XYZ to everyone." (Where XYZ can be AMD, Intel, Apple, NVidia, ATI, or whatever.) It doesn't matter what Joe wants or needs, the tail-wagging fanboy absolutely _must_ preach the One True Way of his corporate masters. Even if he just wants a car stereo or a lawn mower, someone just _has_ to figure out a way to recommend a Mac or Linux instead.

Point in case: we're discussing a comparison between A64 motherboards. That's it. _Not_ a comparison between the A64 and the G5. _Not_ a comparison between MacOS X and Windows. Not even one between AMD and Intel. That's it: which A64 mobo is better, for people who already decided they want an A64.

But no, you just _have_ to get people dropping by to say "noooo! buy a G5 instead!" Hello? What does it have to do with what we were talking about? All this fanboy stuff is just noise drowning the useful signal. No more.

2. The kind of troll which you can find a couple of messages up the thread, who even has to throw phrases like "penis envy" or "cheap-ass homebrewed box". (That cheap-ass box being an Athlon 64 3200+ with a Radeon 9800 XT. I.e., the fastest gaming PC money can buy.)

It's not just trolling, the "penis envy" thing is just disturbing. Not only it's not helping the discussion at hand, it's not even helping the other Mac advocates either. It's practically spelling it out that, in his sad little mind, it's the computer equivalent of the "if I put a big wing on my car it will compensate for this here small dick" syndrome.

Yeah, that ought to help Apple's cause ;)

So if you're going to pick on that last phrase, please do bear in mind the troll I was answering to in that message.

Re:Well.....

[Anonymous Coward]

MacOS X is a 32bit OS as well

What do you mean "as well"? There is a 64-bit Linux for Athlon64. I've used it.

Re:Well.....

[Nexx]

Well yes. I did try to mention that if you just take your comment out of context, it's pretty ironic. Sorry, I wasn't quite clear on what I meant.

Re:Comparison to a G5?

[Anonymous Coward]

Umm no.

G5 has OS X.

AMD64 has no real good OS choice. Unless you need it for scientific programming and data processing stick with 32 bit architecture or your G5.

They'll stop and think...

[Anonymous Coward]

...just long enough to come up with some fantastic 'benchmarks' that demonstrate the awesome power of the G5 ;)

Sometimes I wish I were stupid...

[jeeves99]

... then I'd have an excuse not to spend an hour reading this 46 page beast.

Am I the only one who is a little perplexed at the complexity of the AMD cpu roadmap? The constant barrage of codenames and pin settings is really becoming trying. A more solidified upgrade path with a set numbers of goals would be much appreciated.

Re:Sometimes I wish I were stupid...

[WIAKywbfatw]

AMD's roadmap is simple: faster processors as soon as possible.

All you have to do is worry about how much computational power you want and how much money you want to spend on a CPU and motherboard.

Let's face it, if you hope to see an appreciable speed bump when you upgrade, buying a first-generation chip and plugging it into a first-generation motherboard with the expectation that you'll get that big speed bump when you plug in a second- or third-generation chip a couple of year's down the line is the wrong way to go about it. Yes, the new CPU will have a faster clock speed but the rest of the motherboard will be two years out of date.

Take my AMD Athlon motherboard as an example. When I bought it a couple of years back, together with an 1200MHz CPU (then the second fastest chip in the range), it had all the latest bells and whistles. But today, its support for USB 1.1, DDR2700 RAM and even PATA RAID make in far inferior to the vast number of motherboards out there that support USB 2.0, DDR3200 and 3500 RAM and SATA RAID, not to mention IEEE 1394 (FireWire), Gigabit Ethernet, better POST reporting, etc.(I won't even start to debate the performance benefits of newer nForce2 Ultra chipsets over their older counterparts.)

To match the features of the latest AMD Athlon/Athlon XP motherboards with my older motherboard I would have to add in at least two, maybe three or four, PCI cards. This would work, but it would be an inelegant (taking up valuable PCI slots), costly (PCI cards aren't free) and inefficient (PCI cards require drivers, configuration, etc) solution. Far better and cheaper to upgrade the motherboard along with the CPU in one go, allowing me to put the older motherboard and CPU combination into another machine/my spares box/the charity bin.

Seriously, when buying a motherboard and CPU, look past the upgrade path. It's a serious red herring, even for PC enthusiasts such as ourselves.

Re:Sometimes I wish I were stupid...

[Sivar]

I find it entertaining that you sometimes wish you were stupid, then immediately proclaim that you don't understand something. ;-)

Seriously though, Anandtech has a decent explanation of AMDs rather creative roadmap here.

Re:Sometimes I wish I were stupid...

[John Hurliman]

At 46 pages x 2 ads per page that's 92 ads for the entire article. Figured at the standard rate of one cent per impression, the site makes almost a dollar every time someone reads the article. What are the chances review sites would release their reviews in pdf or a similar format for offline (ad-free) reading? Zero.

Speed for speed's sake

[ObviousGuy]

Back a few years ago, these speed increases really meant something. It meant the difference between waiting for the OS to finish some task and being able to use the computer without much noticeable latency. These days, the difference just isn't as staggering.

I will admit, though, that if you use KDE/Linux there are some things that could definitely use a speed-up like switching between apps and loading the GUI shell. However, beyond that, modern operating systems work just fine with today's processors.

The argument to this is always "what if you're doing serious number crunching or graphical rendering?", but the answer to that is that there are dedicated DSPs out there that can perform those computations much more efficiently than the CPU. Relying on the CPU to give good Quake framerates is like relying on your auto-body shop to soup up your ricer. Yes, there are some increases in performance, but the real horsepower behind these things lies in the video card and engine, not in the CPU and rice spoiler.

I'm all for improvements in chip technology, but software lags so far behind the capabilities of modern CPUs that it's preposterous to climb on the upgrade cycle, regardless of the circumstances.

Re:Speed for speed's sake

[Sivar]

I have to disagree.
First of all, games are not necessarily limited only by the video card. Certainly if you run the latest games in the highest resolution with 8x AA, your video card will be the bottleneck, but often times only these extreme situations make that true.

Morrowind, for example, doesn't really care much about your video card. If you have a Geforce 3, it is happy. It does, however, care about your CPU. If your CPU is not god incarnate, your frame rate will be limited, particularly in some of the more dynamic scenes. The fastest CPU at the time of release, the P4 2.53GHz, could not muster much of a frame rate regardless of video card.

Any 2D game will be CPU limited as well. Baldur's Gate 2 still chugs on some of the extremely large fights even on my AthlonXP 2500+.

In Starcraft, I assure you that my carrier attack will slow your frame rate regardless of your CPU. ;-)

Other than in video games, I am currently transcoding a Babylon 5 video from MPEG-2 to DivX (using Xvid) on my laptop. It is an Athlon64 3200+--the fastest laptop processor money can buy (well, strictly for video transcoding, the highest end Pentium IVs are actually slightly faster) and it takes about 6 hours for a 2hr movie, 3 hours for an episode. If I had a 20GHz Athlon64 it would still take forever.

To come to a point, yes, modern operating systems do tend to run fine on modern fast processors (with the possible exception of WindowsXP and anything running KDE or Gnome2 ::ducks::), but there exists quite a bit more software than old games and operating systems.

A few other examples:

- There isn't a computer on the planet fast enough to install Gentoo Linux quickly.

- FreeBSD's make world will be noticeable non-instantaneous for many GHz to come.

- Waiting for Visual C++ in Windows to compile... Well, anything at all, is not instantaneous even on an 8-way Xeon.

- Waiting for Regedit in Windows to search for a certain key or value will NEVER be fast on ANY computer. I don't know what search algorithm Microsoft chose for that thing, but it's damn slow for searching through just 10 or so megabytes of data.

- Anything ever written with SWING in Java. It was slow in 1996 and it's slow now. To avoid flames, I love Java as a language, but SWING is slower than a dead slug stuck in frozen molasses.

The opinions expressed in this post do not necessarily reflect the opinions of the poster.

Re:Speed for speed's sake

[bluewee]

Just you wait till MS releases longhorn and your processor of today screatches to a halt...

Re:Speed for speed's sake

[jmv]

Actually, current DSPs aren't *that* fast. With x86 CPUs that have a *theoretical* performance in the order of 10 gflops, the DSPs have lost ground. Not only that, but they're much more complicated to program. Believe me, I'm doing all kinds of audio processing and if you give me a CPU that's 10x faster, I'll make use of it in a minute. There's still so many things you can't do right now with audio (even more true with video) because it would be too slow.

Re:Speed for speed's sake

[Powercntrl]

Other than in video games, I am currently transcoding a Babylon 5 video from MPEG-2 to DivX (using Xvid) on my laptop. It is an Athlon64 3200+--the fastest laptop processor money can buy (well, strictly for video transcoding, the highest end Pentium IVs are actually slightly faster) and it takes about 6 hours for a 2hr movie, 3 hours for an episode. If I had a 20GHz Athlon64 it would still take forever.

One question, why does it take so long for your rig to convert MPEG2 to Xvid? I have an Athlon XP2000+ system running WinXP and using FlaskMPEG 0.6preview2 and the latest Xvid binaries from Nic, I get an average of 22FPS encoding rate. Considering film is 23.976FPS, I am getting almost realtime encoding speeds. Even doing a two-pass encode and adding the time it takes to do the audio afterwards (usually about 20mins per hour) would only make a 2 hour movie take a little over 5 hours or so.

Not trying to flame, just curious why your encoding speeds are so slow...

Re:Speed for speed's sake

[Moraelin]

"Anything ever written with SWING in Java. It was slow in 1996 and it's slow now. To avoid flames, I love Java as a language, but SWING is slower than a dead slug stuck in frozen molasses."

That's funny.

I wrote a very complex Swing GUI in 1999, complete with highly customized look and feel, font anti-aliasing, and overkill use of graphics. Guess what? It ran perfectly ok on a 400 MHz K6-II with a TNT graphics card. Go figure.

Yes, Swing is _not_ newbie friendly. If you're clueless, Swing gives you enough rope to hang yourself, _and_ the guns to shoot yourself in both feet.

However, any half-competent Swing programmer should be able to get perfectly adequate performance out of it. Anyone who can't get it to work fast enough on an Athlon 3200+, no offense, but is one of those clueless burger-flippers who shouldn't have got hired as a programmer to start with.

Re:Speed for speed's sake

[Anonymous Coward]

Gee, you're a real prize asshole aren't you!

Re:Speed for speed's sake

[root:DavidOgg]

he's right. Java is heavily criticised not because it deserves it, but it's just the in thing to do.

Re:Speed for speed's sake

[secondsun]

No, this isn't interesting it is a troll, but I will bite.

I have a Radeon AIW 7500, I want to play Savage, but the 7500 makes it look like shit. So I want a new video card and can buy a 8x AGP one, but my motherboard is only 4x. So I am already upgrading two components, why note spend a little more and get a new proc and mobo and make everything faster?

I am sure many who upgrade are in this circle. Their computer works fine, but when they need to upgrade one part they may need to upgrade others to get it to work fully.

(Yes I know there isn't much diferenc between 4x and 8x in performance but it was the most easily identifiable example on hand).

Re:Speed for speed's sake

[TheSunborn]

It is just a shame that nobody ever published how to make swing fast. And the fact that many developers still think that swing is threadsafe don't help.

Re:Speed for speed's sake

[sweede]

I'd hate to say it causs i'll get +flamebait, but it is probably because of less than optimum linux based application performance. I have a XP 2500 w/dual channel memory and i've encoded the entire Star Trek TNG to DiVX 5 using the Gordian Knot suite. each video took around an hour from DVD rip to completed AVI. that all included two pass encoding and a seperate audio encoding.

Encoding a Full DVD takes considerably less than 5 hours, probably more like 2 using GordianKnot.

Re:Speed for speed's sake

[fruey]

Other than in video games, I am currently transcoding a Babylon 5 video from MPEG-2 to DivX (using Xvid) on my laptop. It is an Athlon64 3200+--the fastest laptop processor money can buy (well, strictly for video transcoding, the highest end Pentium IVs are actually slightly faster) and it takes about 6 hours for a 2hr movie, 3 hours for an episode.

What resolution are you transcoding to? The same as the original DVD? I can't believe it's that slow. What are you using to do it? How much RAM do you have?

I have an Athlon 2400+ and two pass XviD wouldn't take that long, and I only have 128MB of RAM. Unless, of course, you're leaving it at the full original DVD resolution and not chopping out any of the black bars.

Having said that, I reckon my PC could transcode at full resolution faster than in 6 hours. Heck, I less than halve the resolution and I get a full 2 hours transcoded (including audio) in 3 hours or so.

Re:Speed for speed's sake

[bhtooefr]

Certainly if you run the latest games in the highest resolution with 8x AA, your video card will be the bottleneck, but often times only these extreme situations make that true.

Then again, even a high end GeForce (because the Radeon wouldn't work with the AGP bus) won't make a P100 with 512MB of RAM and Windows XP Pro SP1 get over 14.5FPS on Q3A (if THG is to be believed - they usually aren't).

Re:Speed for speed's sake

[Anonymous Coward]

There isn't a computer on the planet fast enough to install Gentoo Linux quickly

ooooh you're looking for a fight, aren't you :)

Re:Speed for speed's sake

[Sivar]

I'm not sure. I am encoding the videos with very high quality settings, and they end up pretty large (1.5GB for a 2 hr movie--I am looking to archive). If I encode in lower quality, it can go faster, but there are other factors such as the encoder used, settings, 2 pass vs. 1 pass, etc.
I am still pretty new to DivX transcoding, so I may very well be doing something wrong. :)

Re:Speed for speed's sake

[Sivar]

I am not an advanced Java programmer, as most of my coding is C or C++, sometimes asm. I was mostly referring to commercial Java applications I have used, such as JBuilder.
SWING is so high level, I am really not sure how one could screw it up. Well, actually I take that back--Seibel Systems sells a trouble ticket package that gets GUI widget data from a remote server every time--when you click on a drop menu, when you open a list, etc. It does no caching, even if the values don't change for months. THAT is a great way to screw up a GUI.
Another app I have used (and am going to write a replacement for because it just sucks _so much_) updates it menu data from _FTP_ on every use of a widget. I would sure like to find the idiot that dreamed up that design.
Some of the other Java work I've done, which may have been bad Java, was terribly slow as well. Loading a series of GIFs, for example, simply takes forever. This was using relatively straight forward Java APIs, though there are probably faster replacements.
In any case, I don't claim to be a Java expert or professional Java developer, but I am hardly someone that should be burger flipping instead. :)

Re:Speed for speed's sake

[Sivar]

I am using Xvid 2-pass encoding in the full resolution of the original source (less the black spacing bars of widescreen video) with max quality settings, using the original audio, sometimes doing noise filtering. I think the primary reason may be that I tend to encode at extremely high bitrates--as high as the encoder thinks it can make use of, usually about 800MB/hr depending on the complexity and motion in the video clip.
In my case, the Babylon 5 DVDs are a bit noisy too, so may be more difficult than some to encode.

Re:Speed for speed's sake

[fruey]

I don't know that you need to go full res for acceptable results. If you reduce a little bit then you'll still have better than SVHS quality and will be able to encode more quickly. This will most certainly help out for noise, which will benefit from a noise filter + resampling rather than the other way around.

However if you're a bit of a perfectionist then by all mains re-encode at full res less any black bars, and pay the time penalty for that of course. The extra noise won't make encoding any slower as such, it will just effect the general output quality. Resizing (small reduction, say to 560 wide) + filtering noise just may give you more aesthetically pleasing results. Still I guess you've already experimented and you just want top quality backups so ... take my rambling with a pinch of salt.

Re:Speed for speed's sake

[Sivar]

Thankyou for the advice. I am still in the experimental stage, so will try your suggestions and see how it turns out.

Re:Speed for speed's sake

[Mr Smidge]

- Waiting for Regedit in Windows to search for a certain key or value.. I don't know what search algorithm Microsoft chose for that thing, but it's damn slow.

I think they use Bogosort.

Re:Comparison to a G5?

[Anonymous Coward]

thats a bit unfair... windows 2000/xp are great and linux is pretty cool nowadays also... (writing this on MacOS 10.3 tho)

i have a winxp box at work and my mac at home both kick arse

my 2 cents

Another article

[ValourX]

I wrote an AMD64 article a while ago... something a little simpler, for those not so technically-minded:

AMD 64 Explained

Someone said above that there are no good AMD64 OSes... bullshit... SuSE 9.0 AMD64 is more than usable, and FreeBSD 5.2 AMD64 is almost perfect; in fact I'm typing this from Mozilla Firebird on FreeBSD 5.2-RELEASE AMD64 right now.

-Jem

Re:Another article

[Anonymous Coward]

Any good user apps for 64 bit linux? Didn't think so. I ran some DEC 64 bit systems running RedHat for a while the only thing they were good for was scientific code.

Re:Another article

[ValourX]

Almost the entire FreeBSD ports tree works just fine on AMD64, although some programs have to be compiled with -fPIC.

OpenOffice doesn't work yet because Java doesn't compile yet, but this will be fixed very soon as Sun is working on porting Solaris and Java to AMD64 right now. KDE, GNOME, and all associated programs work just fine in FreeBSD/AMD64. Grip, XMMS, Mozilla, Evolution, Bluefish... they all work perfectly.

-Jem

Re:Another article

[Anonymous Coward]

Whether they work or whether they work faster or better are two different things.

Re:Another article

[ValourX]

I don't have any data on if they work any faster, but each program certainly compiles faster in 64-bit mode than in 32-bit mode. Noticeably faster. I have numbers to prove it, but I'm waiting to publish all my data in an article sometime soon.

-Jem

Re:Another article

[ctid]

That's a very nice report. I was just looking for a quick definition of HyperTransport and you've cleared it up nicely for me! One thing you might want to change is where it says that the AMD64 range always has 1024kb cache. That's no longer true with the Socket 754 AMD64 3000+, which has the same clock speed as the 3200+ but only half the cache.

Re:Another article

[Anonymous Coward]

I've got the Windows64 beta, but no 64 to run it on. Woe is me.

Re:Another article

[Anonymous Coward]

Someone said above that there are no good AMD64 OSes... bullshit... SuSE 9.0 AMD64 is more than usable, and FreeBSD 5.2 AMD64 is almost perfect;

They mean there's no Windows XP 64-bit version yet. It'd be kind of silly to by the newest fastest CPU and run Linux or FreeBSD on it when a 2 year old processor/motherboard combo will be more stable and have more driver support. I will pick up an AMD64 rig when Windows XP 64-bit comes out.. not before.

Would someone mind telling me the difference...

[double-oh three]

Would someone mind telling me the difference between the 939 pin and the 940 pin? What difference can that one pin make?

Re:Would someone mind telling me the difference...

[Anonymous Coward]

It means that you can effectively double the number of functions that the CPU is capable of.

If pin 939 (they number from 0-939) is off, the rest of the CPU pins act like a normal CPU, but when it is turned on, all the pins behave in 64-bit Ath64_SNM mode which is basically a 64 bit enhanced mode with extra featuresets designed for streaming data (not necessarily from the Internet).

Re:Would someone mind telling me the difference...

[10Ghz]

939 = Single-CPU only. 512KB of L2-cache, 128bit mem-controller

940 = 1-8 CPU's. 1MB of L2-cache, 128bit mem-controller.

Re:Would someone mind telling me the difference...

[The One KEA]

The 939pin processors, which will be sold in single and dual-channel variants, will not require registered memory like the Opteron does. This means that they will be able to operate much faster and be much more overclockable.

Re:Would someone mind telling me the difference...

[Sivar]

Socket 939 will allow motherboard manufacturers to easily make 4-layer designs.
In English: Cheaper motherboards for the dual channel Athlon64s.

Athlons are efficient with their use of memory bandwidth, so current Athlon64s don't really care about the second memory channel much at the moment. It has a minimal effect on performance. However, since processor technology moves more quickly than memory technology, future 3+GHz processors will start to see a significant benefit from the added bandwidth. Of course, by then, DDR2 will be readily available so we'll just have to see how it all turns out.

Re:Would someone mind telling me the difference...

[Imperator]

Wow, the 941-pin socket must be really something!

Cheers, :)

Re:Would someone mind telling me the difference...

[Anonymous Coward]

The 939pin processors, which will be sold in single and dual-channel variants, will not require registered memory like the Opteron does. This means that they will be able to operate much faster and be much more overclockable.

Yeah, maybe the unregisted memory will let you do some overclocking, but you'll be much more memory limited, too. As memory gets bigger and bigger, Registered Memory will be a must in order to overcome the memory buss' signal slew and ringing caused by loading the bus with so many memory chips. See, that little "thang" referred to as a "Register" when describing registered memory is not just some pai-in-th-arse memory buffer: it's a place where electrical signals get Re-mastered in order to insure they make it down the line in good shape. And the more memory you have, the more load you have, and the more likely the signal requires "Re-mastering".

So, in the future when we're all running 4GB plus, we'll also be running Registered Memory.

Registered Memory - It's the Future.

Re:Would someone mind telling me the difference...

[CrosseyedPainless]

In my world, you have been modded (+5, Funny).

For all the good it does you here.

Re:Would someone mind telling me the difference...

[Tarqwak]

Correction:
939 won't be 512 kB L2 cache only, the 939 pin Athlon 64 FX series will have 1 MB L2 and cheaper Athlon 64 will have 512 kB L2 (both will have dual channel memory controller), also 939 pin Athlon 64 will get 1 GHz HyperTransport bus, soon enough die shrink to 90 nm (currently 130 nm) and will not require registered/ECC DDR memory. It's been covered in The Reg/The Inq for many times.

940 pin Athlon 64 FX-51 was just rebranded Opteron, Socket 940 will remain Opterons domain, single channel DDR Socket 754 will be probably targeted for mobile and other low power devices.

Intel 90 nm Prescott core wont run on every Socket 478 mobo (because of the sky high current requirements) and next year you'll have to get a Socket T (775 pin LGA) mobo for newer Prescott or its successor Tejas core.

AMD went from: ... -> Socket 7 -> Slot A -> Socket A -> Socket 940 -> Socket 754 -> Socket 939

940 & 939 have a bright future.

Intel went from: ... -> Socket 7 -> Slot 1 -> Socket 370 -> Socket 370 (FC-PGA) -> Socket 423 -> Socket 478 -> Socket 478 (Banias) -> Socket 478 (Prescott) -> Socket T

Socket 478 (Banias) & Socket T have some sort of future.

IIRC

Athlon64 Motherboards And Chips Compared

[Anonymous Coward]

Let's see the Motherboards are about 8 inches square but the chips are much smaller ...

Re:Comparison to a G5?

[jhunsake]

OS X is 32-bit. Nice try.

!opteron == no dual proc

[rsborg]

As much as I love AMD, I would recommend against the Athlon64 chipsets, unless you *must* have a 64 bit chip. What is interesting, however, are the Opteron chips, where you can easily buy a nice dual proc mobo that has some nice features. Of course, this will cost you ...and the price hasn't dropped in the past couple of months, too much :-(

Of course, 754 is being deprecated and all that, but I thought I'd put a word in for what I'd buy... if it weren't so damn expensive. *sigh* Will we ever have dual athlon64 goodness?

Re:!opteron == no dual proc

[runderwo]

As much as I love AMD, I would recommend against the Athlon64 chipsets, unless you *must* have a 64 bit chip.

Why? They are much faster at running even 32-bit code than Athlons. They dissipate less power. They have safety features built in to prevent overheating, and power throttling built in to prevent less wasted energy when idle.

Perhaps the only reason not to move to the AMD 64 platform is the entry price, currently. The early adopters will take care of knocking that down for the rest of us.

Re:!opteron == no dual proc

[Anonymous Coward]

Will we ever have dual athlon64 goodness?
It's only been a year. Give it time & let this pin issue settle. I for one am still waiting for the 8 way processor setups promised back in 2003. I'm betting the release for the 8 processor Athlon 64 wil coincide with the release of Duke Nukem Forever, seeing as the Athlon 64 8 way will be bare minimum. Along with a Geforce Dust Buster Edition or a Radeon 1xE^1000.

Re:!opteron == no dual proc

[The One KEA]

Not even the entry price is that much of an issue now; the release of the Clawhammer-based Athlon 64 3000+ for the sale price of only $215 significantly lowers the barrier to entry. Sure, all the necessary parts for a completely new system may still run all the way up to approx. $850, but having a low-spec processor like the 3000+ available only helps AMD"s product adoption.

Re:!opteron == no dual proc

[Anonymous Coward]

I have an 8-way Xeon from 2000. It *kills* my P4 desktop.

Re:!opteron == no dual proc

[p3d0]

That's great advice, since clearly there couldn't possibly be anyone who would want a uniprocessor machine.

Re:!opteron == no dual proc

[Klaruz]

Entry price?

Give or take an athlon 64 3000 is about the same speed as a p4 3.0. According to pricewatch:

Athlon 64 3000: $211
P4 3.0 Ghz: $259

The motherboards are about the same price. I got a really nice one with 2 raid chips, gigabit ethernet, firewire, lots of usb ports, pci slots, etc. For $130.

Granted, it's no $150 board/chip combo. But it's a near top of the line system for just over $300 + ram. I would have killed for these prices a few years ago. Not to mention with cpu speeds these days and it being 64 bit I shouldn't need to upgrade for a while.

Re:!opteron == no dual proc

[Hoser McMoose]

Will we ever have dual athlon64 goodness?

Probably only if/when Intel brings out dual processor P4 systems. The market for dual-processor Athlon64 chips, when dual-Opteron's already exist, is pretty much zero. Sure, I'd like one and you would like one, but honestly 99% of the people buying computers won't even consider a dual-processor setup and those that do are mostly looking at the high-end (ie Opteron or Xeon).

So, unless Intel makes dual-capable P4's a checklist item, AMD isn't likely to support it. Of course, the light at the end of the tunnel here is that AMD is seriously considering dual-core processors for the successor to the Opteron/Athlon64. Still 2 years away (and that's assuming AMD sticks to their schedule, which isn't too likely), but could make for a nice buy when that time comes.

Are the apps there?

[leftie_hater]

Are there any apps that are 64 bits? Is there any reason at all to go 64bit?

Re:Are the apps there?

[Anonymous Coward]

Of course.
Starfox 64
Mario Cart 64
Random Nintendo Franchise 64
Oh wait... for the computer then no.

Re:Are the apps there?

[ValourX]

Most open-source projects are now in the process of, or have completed AMD64 compatibility. I'm typing this from Mozilla Firebird on AMD64 FreeBSD 5.2-RELEASE. I have a whole bunch of programs from the Ports system that work perfectly... the ones that aren't ported yet are the proprietary clones, like the Flash plugin, GAIM, and Java. Opera doesn't work in 64-bit mode yet either, neither does TextMaker.

-Jem

Re:Are the apps there?

[viralburn]

Are there any apps that are 64 bits? Is there any reason at all to go 64bit?

The Far Cry game engine is being optimized for athlon 64.

Re:Are the apps there?

[Anonymous Coward]

Firebird 64 rocks.

Seriously Firebird on any platform is superior.

Re:Are the apps there?

[SD-VI]

Even if you ignore 64-bitosity (64-bittiness?)-- others have already pointed out valid reasons not to do so, so I won't go there-- the A64 is quite fast even in 32-bit operation. I see it as more of a neat feature than a hook, but then again I've never had to use more than 2GB of RAM in a 32-bit system.

Re:Are the apps there?

[Anonymous Coward]

I have had to use more than 2 GB RAM in a 32 bit system. Since the Alphas were discontinued we've been using 32 bit architecture to run some scientific models and we've had to tweak the kernels quite a bit and still had limited success. The AMD64 is going to be a boost to all of us in the low cost scientific computing world.

Re:Are the apps there?

[leftie_hater]

So the next version from Microsoft will be Windows 64?

Re:Are the apps there?

[dave-tx]

Synopsys is in the process of releasing 64-bit Opteron executables for the Linux platform. So for those of us engineers who do very large synthesis and place-n-route work, yes, there's a very good reason to start thinking about 64-bit Linux systems.

Re:Are the apps there?

[Anonymous Coward]

Not everything on AMD64 HAS TO RUN AT 64bits (user land).

The important part is getting your OS and (most importantly) KERNEL DRIVERS all 64 bit. SLES is fine & stable, but drivers are lacking, especially in the new "SATA" area (hardware SATA raid being virtually non-existent). Even some older SCSI RAID card drivers aren't "64 bit clean".

Once you clear that hurdle (or slap a PATA drive in!), plain old i386-i686 32bit binaries will run about 20% faster, due to increased performance in the OS. (Not to mention that if you have more than 4GB of memory, each 32bit process gets a FLAT 4GB to play with- no more 2gb/2gb or stupid PAE garbage).

Re:Are the apps there?

[Anonymous Coward]

The first apps 64-bit apps to come out that will be major improvements over their 32-bit counterparts will be encoding tools. If you rip a lot of DVD's or CD's you will see huge performance gains.

Re:Are the apps there?

[Anonymous Coward]

Actually 99% of the code in those games are 32-bit. UltraHLE didn't bother with emulating the entire 64-bit word size of the N64's processor, so it HLE:d the small bits that DID use the 64-bit word size.

Re:Are the apps there?

[Anonymous Coward]

I would be very surprised if Gaim doesn't work on 64-bit ...

Most of these programs already happily run on 64-bit processors like Alpha, Sparc, etc... In most cases, there is no "porting" to be done.

It's different, though, when you're waiting on a proprietary vendor (like Sun) to provide you with a binary.

Re:Are the apps there?

[Jondgonzales]

They are realeasing the 64 bit versions of several games (Unreal Tornament 2003-04, Dues EX invisable war, Call of Duty, and C&C generals). The 64 bit version of WinXP is coming out in the middle or late 2004. More games are converting to 64 bit because they will work better.

No 64bit Linux OS???

[_Pinky_]

If I do recall there is a gentoo live CD out right now.. In fact the gentoo page has a Athlon 64 faq out here:
http://dev.gentoo.org/~brad_mssw/amd64-tech-note s. html
Now, like all new technologies, there maybe certains apps that don't work, compilations errors, and other problems... But how will they be fixed unless people try it, and send back bug reports?

Tired old flamebait argument

[Powercntrl]

There are still many tasks for which there isn't enough computing power for. Factoring large prime numbers, encoding/editing video, rendering 3D graphics, applying audio filters, etc...

Every time a newer/faster/better CPU comes out, someone says it is not needed for the majority of computing users. While that may be true currently, who would want to tolerate using a 386SX/16 today just because current 32-bit X86 proccessors are really just souped up 386s?

If you're happy with your old processor, keep using it. No one is going to take it away from you. Chances are, you'll start to see the benefit from more powerful processors and applications that take advantage of what they can do and you'll upgrade just as you probably have in the past. You're not still using an abacus are you?

Re:Tired old flamebait argument

[inburito]

There are still many tasks for which there isn't enough computing power for. Factoring large prime numbers, encoding/editing video, rendering 3D graphics, applying audio filters, etc...

That is probably something you do not need any computing power for as it can be done in constant time.

Re:Tired old flamebait argument

[Saberwind]

Most people don't need a computer to factor large prime numbers, since they can do it in their heads. Instantaneously.

Re:BOW DOWN TO YOUR CORPORATE MASTERS

[Anonymous Coward]

> How these corporate leaders sleep at night without the urge to commit suicide is beyond me.

See 6 zeros after my bank balance would be a pretty effective suicide deterrent for me. ;-)

1 MKB of L2 cache!

[Anonymous Coward]

According to the side by side comparasion chart there's 1 megakilobyte of L2 cache on the 64-FX! With a gig of memory on die, no wonder it's so expensive.

Applications for more speed

[Spy Hunter]

Better real-time realistic physics for games (including on-the-fly synthesis of human-like motions for characters), better AI for game characters, real-time video processing/editing, even better real-time sound synthesis, faster compiling, higher-level languages for more productive programmers, voice synthesis and recognition, robot motion control (walking), natural language processing, next-generation smarter desktop interfaces, useful computer vision, AI (as in, a computer that passes the turing test).

There are still plenty of cool applications for general-purpose processors that require more speed. Then once these things can be done by themselves, you'll want to do them all at once. And there are always the crazy applications that nobody's come up with yet.

Re:Applications for more speed

[Junior J. Junior III]

A few of those problems you mention are not the sort of problems that will be solved simply by throwing more speed at them (though more speed won't hurt and may help). AI, for example, is hampered because the problem still hasn't been properly understood or defined.

Re:Applications for more speed

[Spy Hunter]

Speed isn't the only issue in AI, of course, but I think more speed will be needed when the solutions are finally found.

CPU speed still matters

[poszi]

Back a few years ago, these speed increases really meant something.

There are still a lot of situations where faster CPU is great. I do scientific calculations for my work and, surprise, the faster the CPU, the quicker you get the results. Actually, cheap commodity PCs made a revolution in my field, where you no longer need an access to a terribly expensive supercomputer to do reasonable simulations.

I've got also a digital camera and image manipulation is very CPU intensive. Unsharp mask on a 6Mpixel file takes several seconds and if you need to aply it to hundreds of images, you can do the math. CPU is also important in ogg encoding, program compilation and just anything that takes 100% CPU if you check top.

Re:Comparison to a G5?

[Moraelin]

So the G5 has a 32 bit OS, which is obviously far better than the Athlon 64 having a 32 bit OS ;)

Not to mention that an Athlon 64, even in 32 bit mode, runs circles around a G5. But wait, at some point in the undefined future, there'll be some miracle IBM compiler and 64 bit OS for the G5, which makes it all faster. Just y'all wait and see. Unlike the Athlon 64, which, uh, is also waiting for a 64 bit compiler and OS to make it all faster.

Sometimes the logic of Mac fans is a bit too strange for me to follow.

Here's another idea: if a Mac is all you need, good for you. By all means, stick to your Mac. I'm genuinely glad that you found your dream computer.

But for some of us a Mac just doesn't fit the needs. E.g.,:

- Games. Yes, I know that you can buy a whole 20 games for the Mac, some of them almost 10 years old (e.g., Fallout), and some of them Solitaire clones that you can download for free in the Windows world. But some of us, you know, need more games than that.

- Price. Yes, the dual G5 is a nice computer, but the price I've paid to build my Athlon 64 3200+ computer, including a shiny new ATI Radeon 9800 _XT_, was a _third_ of that. Or half the price of a single processor 1600 MHz G5 with 9800 _Pro_. On account of keeping my old case, hard drives, RAM, PSU, etc.

And if I'm to add the price of buying all my old software again for a "switch", the price comparison is getting even more disastrous for the Mac.

So basically all I'm saying is: the right tool for the right job. For some of us the Mac is just _not_ the right tool. Our choice is simply "Pentium 4 or Athlon 64".

Re:BOW DOWN TO YOUR CORPORATE MASTERS

[Anonymous Coward]

Actually, it would be homicide bait - please kill me if I ever put millions of dollars into a bank account.

Athlon64 Coming Along Slowly

[gbulmash]

I wish I'd done more research on hardware compatibility, particularly motherboards, because installing 64-bit Linux has been a bitch. I'm only now getting to the point where I can have a fully-working installation without having to add in redundant devices to compensate for onboard chipsets that AMD64 Linux distros couldn't work with.

Nvidia Nforce drivers only got released in the last month so my onboard LAN on my ASUS SK8N works. Mandrake 9.2 RC1 recognizes my Promise onboard SATA RAID controller, but SuSE doesn't, and even then the driver in Mandrake is an 0.83 release.

I haven't played with the Fedora Core release candidate test version for Athlon 64 yet.

IMO, If you want to run 64-bit Native Linux on AMD64 without a lot of headaches and weeping, wait another 6 months until the distros and drivers have solidified more. In 6 months, you'll probably be able to get a CPU a generation or two higher than you can today, but for the same money, and you'll be able to install AMD64 native Linux much more easily... It's win-win.

- Greg

Re:Athlon64 Coming Along Slowly

[The One KEA]

The Ethernet/LAN driver issue is no longer a major problem - if you can find a distro which bundles a 2.4.23/2.6.0 or later kernel, it will include the new forcedeth driver, which is a clean-room reverse-engineered driver for NVIDIA Ethernet devices. It works very well, and I've seen lots of positive feedback.

Right now, though, you're probably right about the immaturity of 64bit Linux distros - IMO Gentoo is the one distro that is most likely to mature soonest on the AMD64 platform.

Re:Athlon64 Coming Along Slowly

[niittyniemi]

I wish I'd done more research on hardware compatibility, particularly motherboards, because installing 64-bit Linux has been a bitch.

I think you'll find that FreeBSD has moved ahead quite nicely on AMD64 and you could well have better results than Linux. By all accounts 5.2 is pretty solid on all tier 1 platforms.

Re:Athlon64 Coming Along Slowly

[greg_barton]

I just installed Mandrake 9.2 for the AMD64 on my eMachines M6805 laptop. Works like a champ, no hitches. (Though I'm still running the WXGA screen at 1024x768. HAven't tried that yet...)

Re:Athlon64 Coming Along Slowly

[gbulmash]

I just installed Mandrake 9.2 for the AMD64 on my eMachines M6805 laptop.

Is this the boxed version or the RC1 available on their FTP and mirrors?

Re:Athlon64 Coming Along Slowly

[greg_barton]

Yep. Start here

Re:Comparison to a G5?

[Cesare Ferrari]

Athlon 64 runs rings around a G5? Really? Have you got both to demonstrate this, or is this from reading reviews on the web?

I write DSP code, and i've got some very impressive results from a G5 when running code which previously gave less than exciting results on a G4. The G5 really is a class act.

I've not tested the code on an Athlon 64, but only on an Athlon XP 2500. DSP code tends to be FPU or memory bound, sometimes both in different parts of the algorithm so it is pretty good at giving a machine a proper workout.

My XP 2500 is running roughly at 2Ghz, and compared to a G5 at 2Ghz the Athlon takes around 50 to 100% longer to run the same tests. That's comparing a G5/gcc 3.3 build against a x86/VC7 build. Neither is the best compiler for the platform, but both are pretty useful, and possibly typical for currently released software.

I'd be very interested in running this build on an Athlon 64 - that'll still be a 32 bit test, but it would be interesting for me to see the benefits of the on chip memory interface. Rebuilding for 64 bits might take a while since the code is large (and ugly). Anyone with a spare 5 minutes willing to run a binary for me?

Re:Comparison to a G5?

[BenjyD]

Except that the Athlon XP-2500 isn't an Athlon64. It's a budget chip - it's around 65GBP here in the UK (probably $65 in the US, grumble grumble). I don't think you'd get very much of a G5 for that.
Just because they run at the same frequency doesn't mean you can compare their performance.

Re:Comparison to a G5?

[BenjyD]

Which was pretty much what you were trying to say. Just ignore me, I'll go back to dozing while I wait for the IT department to install my bloody software.

Re:Comparison to a G5?

[Cesare Ferrari]

I'm quite aware that the XP isn't an Athlon 64. I'm also quite aware of the prices of processors (and how much I paid for them when they were released). I'm not entirely sure how the price of a processor affects it's performance though - maybe you can help me out with that one.

Also, I can compare the performance of processors at the same frequency. I can even compare the performance of processors at completely different frequencies as well. That's what i'm doing. It's called benchmarking.

My question remains, anyone willing to run a test harness on an Athlon 64 for me? I'd be interested to see how my code runs on it.

Re:Comparison to a G5?

[Cesare Ferrari]

Ok :-) Still looking for an Athlon 64 to try. None of my friends have (yet) taken the plunge.

In that case, you are very lucky because:

[Phekko]

I have been requested by the Nigerian National Petroleum Company to contact you for assistance in resolving a matter... ;)

Feeling lucky?

[A nonymous Coward]

Which one?

Re:Feeling lucky?

[Anonymous Coward]

the one on your head

Why the big fuss over 64?

[386spart]

Commodore already did it in the 80s!

Actually, there are a number of them already...

[sangfroid]

For example:

• SuSE 9.0 Pro for amd64.
• SuSE Enterprise Server 8 for amd64.
• Mandrake 9.2rc1 for amd64.
• Fedora Core1 test1 for amd64.
• Gentoo's amd64 info
• Red Hat Enterprise Linux AS & WS are available for amd64.

I've been using Gentoo's amd64 stuff for a little while on my new Shuttle Box. Things are generally good although there are still a lot of packages that are masked. KDE is also problematic which may be a turn-off for some people.

A colleague just got a new dual-opteron Workstation from Pogo and is running SuSE 9.0 pro for amd64 and is rather happy -- just about everything plays nicely.

Multimedia has significant problems on both systems. No flash player for 64-bit, mplayer and related multimedia requiring 32-bit codecs. Nvidia amd 64 drivers require some patching if they work at all, at least as of last wednesday.

Otherwise quite happy with all of these. Mandrake claims to have multimedia stuff working properly (see above link for info) but wants to eat my partition table so I haven't checked it out yet.

--
"Now you'll see why they call me the Velour Fog" --Zapp Brannigan, 25-star General & Cpt.

Re:BOW DOWN TO YOUR CORPORATE MASTERS

[Vihai]

Actually, putting 6 zeroes after my negative bank account would be anything except a deterrent for suicide :)

Off-topic, about Swing

[Moraelin]

Indeed. The full name is Moraelin F Asshole. F stands for "Flaming" ;)

But now seriously, it's not even about _my_ GUI. I know of other teams which have programmed Swing GUIs too. E.g., there's one big Swing-based enterprise front-end being built two floors up from my office.

I can't recall any of them having _Swing_ related performance problems. Performance problems with the database or the EJB back end, yes. "Swing is too slow" problems, no.

A Swing GUI may take milliseconds for the whole form to be painted, instead of micro-seconds for a native Windows GUI. But that's still orders of magnitude below what the user even starts to notice. And even further below what the user will call "slow".

Don't get me wrong. I'm _not_ a fan of Swing. It does have issues. As I've said, it is _not_ newbie friendly.

E.g., for a language (Java) whose claim to glory included automatic-dealocation via a garbage collector... Swing sure brings back precisely memory leaks and the need to de-allocate stuff manually. (Yes, those listeners.)

It also does require some expertise and some work to get that performance out there. E.g., if you add items one by one to a combo box, and they're lots of items, be prepared to spend _minutes_ before that loop completes. On the other hand, adding them all together, finishes in milliseconds. Better yet, write your own Model class for that combo box(sein' as Swing _is_ MVC based.) That'll work even faster.

So, to wrap it up, yes, Swing needs you to _work_ and _read_ to get a good program done. But then that's what programming is all about. And if you do your homework, yes, you don't need an Athlon 3200+ (nor a G5) to get adequate performance with Swing.

I'd expect anyone who's paid to code to a framework -- regardless of whether it's Swing, EJB, Struts, MFC, .Net or whatever -- to actually spend some time _learning_ what they're supposed to do. Learn the patterns (a.k.a. best practices) _and_ the anti-patterns (a.k.a. worst practices) _and_ spend some time thinking how and why and which apply to your actuall problem (a.k.a. design.) _Then_ jump into coding.

Programming is _not_ about randomly banging on a keyboard, and hoping that it'll eventually work.

It's not _that_ unreasonable a wish, is it?

Re:Off-topic, about Swing

[Anonymous Coward]

_I_ _wasn't_ aware that _underlinig_ stuff _was_ a syntax _requrement_ of _Swing_ _Java_!

_Very_ enlightening.

Re:Comparison to a G5?

[Anonymous Coward]

Add some 20 percent for clock speed, 15 for better L1/L2/memory bandwidth and then some 10 if you're using vector instructions. 10 more for not being register starved if you happen to have 64-bit OS.

Linux friendly boards

[Anonymous Coward]

The article classifies several boards as definitely Linux-unfriendly, since they uses components (Realtek Gigabit LAN, SATA RAID) that it claims are unsupported. Has anyone actually built a Linux box based on one of those boards with all the on-board components working?

Get a cheap, powerfull 64bit platform today

[Pegasus]

Find a used, secondhand alpha system.

Still most powerfull cpu for the clock, excellent support by all free unices, excellent hardware (DEC rules) ...

Nice and hot (like most of today cpus), power hungry as well, usually comes in big boxes with enough room for all the case modding you want.

Re:Get a cheap, powerfull 64bit platform today

[Anonymous Coward]

Just make sure its got an EV6 or faster processor in it. I used to have an EV5 (i think it ran at 500mhz) at work running redhat and it was a lot slower than my athlon at home.

Although alpha has this image of being super fast, its sadly not as true as it once was. A good athlon or p4 will probably outperform your second hand alpha. However I do prefer the alpha SRM (bios equivalent)over what your pc offers. You get a nice command line where you can query hardware, select boot devices etc.

alternately...

[RMH101]

...he's someone who has the skillz to build his machine from the Little Rubber Feet upwards as opposed to spending twice as much on a G5 just because some TV ad said it was "the fastest computer".
I'm not knocking Macs at all, I like them, but you can't criticise someone or call them cheap because they have the ability to build from scratch what you have to buy.

what the hell

[SQLz]

All they did was build a product spec matrix. Thats worthy of Slashdot? There is nothing in there we haven't known since the launch of Athlon 64.

Re:Comparison to a G5?

[bach37]

But for some of us a Mac just doesn't fit the needs. E.g.,:

Right on man. My girlfriend just bought a 12" iBook- it's perfect for her computer needs. But for me, I like building my own machine, messing with Linux, PC games, etc. So Mac is just not for me. Panther is quite nice, though.

Scott

Re:Comparison to a G5?

[bhtooefr]

Actually, it's got GCC, and it's also got SuSE, Mandrake, Fedora, and FreeBSD.

Re:Comparison to a G5?

[Moraelin]

As you probably know, the Athlon 64 has 3 floating point units, versus the Athlon XP's 2. That's 50% more FP power, in and by itself, even in 32 bit code. Which, among other things, is a reason why the Athlon 64 does better in games.

On the other hand, the on-chip memory interface does lower memory latency, but not (directly) the raw bandwidth. It's still a 64 bit wide, 400 MHz memory bus. Stuff which mostly reads sequentially through memory will still run faster on a Pentium 4 with a 533 MHz memory bus. For the obvious reson.

So you can't really extrapolate Athlon XP results to an Athlon 64.

That said, as I've said, the right tool for the right job. If for your DSP code the G5 runs faster (and it could happen), then the G5 is the right tool for you. For me, my home PC being basically a glorified game console in a big tower case, well, it _has_ to be a PC just for the games part.

I'd offer to run your test for you. I'm curious myself. But, you know, in this time and age I'm too paranoid to run executables from perfect strangers on my computer.

(Yes, before someone points it out, I'll admit that the virus and trojan situation did make me consider a Mac before. Then again, installing a firewall and Opera was cheaper.)

Tell you what. If you can write a Java test that I can look at and compile personally, I'll be more than happy to run it for you. (Not going to buy MSVC for that.)

Sheesh

[p3d0]

Some dingbat makes this argument every time computer performance is mentioned. I swear I'm going to start making all you dummies as Foes.

If you're going to get a new MB to run Linux

[miracle69]

specifically, if you want Serial ATA, stay away from boards with the Silicon Image 3x12 SATA controller. IT IS NOT LINUX COMPATIBLE under modern distros. Silicon Image advertises it as LINUX COMPATIBLE, as they have binary only drivers for Redhat 8.

I was dissapointed that by Gigabyte K8A Pro motherboard had this chip on it and it DOES NOT WORK under Linux.

But otherwise, the platform is nice.

Re:If you're going to get a new MB to run Linux

[avenj]

Actually there are Silicon Image drivers that support that chipset emerging now for the 2.6 kernel

Re:If you're going to get a new MB to run Linux

[evilviper]

if you want Serial ATA, stay away from boards with the Silicon Image 3x12 SATA controller. IT IS NOT LINUX COMPATIBLE

Call me crazy, but wasn't the main advantage of SATA (over SCSI) that it's software-compatible with ATA?

Not that I really care that much... SCSI kicks ass.

AMD64 vs. Opteron

[Listen Up]

What is the difference between an AMD64 and an AMD Opteron? I see advertisements for both, but can't find anywhere that tells me what the difference is. Thanks.

Re:AMD64 vs. Opteron

[avenj]

AMD64 refers to the architecture (formerly x86-64). The two chips are the Athlon64 (desktops and notebooks) and the Opteron (workstations and servers -- mostly SMP-land). The Opteron (besides doing SMP, at least if you have a 2xx, 4xx, or 8xx) has more memory bandwidth. The current Athlon64 FX-51 is pretty much an Opteron 1xx.

Re:AMD64 vs. Opteron

[Hoser McMoose]

AMD64 is an instruction set, or more specifially, it is a 64-bit extension to the IA-32 instruction set (which, in itself, was an extension of the 16-bit x86 instruction set, and so on). AMD64 often goes by the name x86-64, which is the original name for the instruction set early on in the development cycle.

The AMD Opteron is a processor that uses the AMD64 instruction set. It is designed for workstations and servers and can be used in a glueless SMP setup for up to 8 processors (>8 processors is possible but requires extra core logic chips to connect them together). It runs at clock speeds of 1.6GHz up to 2.2GHz (current top speed), has 1MB of L2 cache and 128-bit wide memory controller integrated onto the die, as well as 3 hypertransport links for interprocessor communication and I/O. It is marketed under model numbers such as 140, 246, 848, etc, with the first number indicating the maximum number of processors usuable in an SMP system (1xx chips for uniprocessor systems, 2xx for duals and 8xx for up to 8-way systems) and the second two numbers showing relative performance. Personally I am quite fond of this particular numbering scheme for the processors.

The AMD Athlon64 is another processor that supports the AMD64 instruction set. It is designed for desktops and mobile systems, so it will not work in multiprocessor configurations. Currently it runs at 2.0 or 2.2GHz with 2.4GHz chips on the horizon. They have either 1MB or 512KB of L2 cache, depending on the model, either a 64-bit or 128-bit memory controller (again, depending on the model), and are sold using two main model numbe schemes. The first is for the stock-Athlon64, which are sold as 3000+, 3200+, 3400+, etc. These numbers show a rough approximation of their performance as compared to an Intel P4 running at the 3.0GHz, 3.2GHz and 3.4GHz (AMD may not say this officially, but it's fairly obvious that this is the intention of the model numbers). I don't like this model number scheme too much, but on the other hand I don't find it any better or worse than the totally useless clock speed (MHz or GHz) rating that is traditionally used to sell chips. The second model scheme is for the Athlon64 FX line of chips, a chip targeted at the high-end "enthusiast" market (read: bratty kid gamers with too much of their parents money on their hands). These chips are sold as the Athlon64 FX 51 and the upcoming Athlon64 FX 53, with the numbers merely referencing the relative performance of the chips.

Hope that clears a thing or two up. For more information, RTFA!

It's also fun to build from parts.

[Ayanami Rei]

It feels more like "your own computer" then a commodity box. You know everything that went into it. You even sat through an OS install. :-)

What, is SuSE some kinda taboo term?

[Ayanami Rei]

I've had 64-bit (end to end) SLES 8 for awhile now! It kicks serious ass. No 3D accel, but I'm not using it locally for much anyway. (I could fix it but I'm lazy like that, it also has crappy built-in GFX so I'm not missing much)

PCI cards require drivers...

[Ayanami Rei]

but so do all those built-in bells and whistles on the motherboard.

Re:PCI cards require drivers...

[WIAKywbfatw]

Yes, but the drivers (and any updates for them) for all those bells and whistles are sourced in one easy place: your motherboard manufacturer's website. Getting drivers for one device (a motherboard) is easier and less time consuming than getting drivers for several devices (a motherboard, a USB 2.0 card, an IEEE 1394 card, a SATA RAID card, etc).

Also, you know if all your devices are on the motherboard that they'll work together smoothly and that there won't be any conflicts between them. Conflicts of this nature aren't that common nowadays (eg, IRQs aren't the nightmare that they once were) but they sometimes do crop up with add-in devices.

blackdown is available for x86_64.

[Anonymous Coward]

Get it now.

Tom's Hardware articles

[glinden]

Tom's Hardware has been running a great series of articles reviewing motherboards for the Athlon64. ExtremeTech also has a good review of Athlon64 motherboards. And AnandTech recently wrote up a useful AMD 2004 CPU roadmap.

I've been looking at this a lot lately since I was just about to build a new box. Ultimately, I decided not to go with a Athlon64 (too expensive for the limited benefit), but I did find reading all these articles useful in making that decision.

Re:Tom's Hardware articles

[Noren]

Be aware that Tom's Hardware reports results in which Intel consistently performs better with respect to AMD in comparison to most other people's results.

This appears to be deliberately skewed- for example, in their recent Athlon64/P4 comparison they ran they used slower memory sticks with 2-4-4-8 memory timings on the Athlon64 versus 2-2-2-5 timings on the P4.

Re:Mozilla on amd64

[billsf]

Firebird will compile and work. It has some serious bugs and I eagerly await the next 'drop'. The latest Mozilla is a winner. Be sure to use mozilla-gtk or you will have frustrating problems like not being able to resize your browser window. This new Mozilla works well in 64bit mode. It is faster than anything you've seen and that is the browser I'm using now. Still a browser is for browsing, a mailer for mailing a newsreader for news, an IRC client or better, Unix talk or ytalk for chat, etc. The Firebird concept accepts this rather obvious situation.

Debian install

[fdicostanzo]

Dos anyone have any experience installing Debian (unstable) on one of these?

REPLY to this if you are a C/C++ programmer

[zaqattack911]

Ok guys I have one CPU question that is yet to be answered. Aside from increase memory access and integer/float width. What could the possible advantage of a 64-bit 3D game have.

I doubt any of the calculation in a modern 3D game would need variables as accurate or as large as 64bits. Thus how could there be any speed increase?

Register size/Bus speed/hypertransport all can be added to current 32bit platforms. The introduction of 64bit instructions as far as I can tell will not offer any benefit to a gamer.

It's not like you can "pair up" instructions now... those instructions that used to be 32bit, when recompiled simply take up 64bits now, right? If your video games don't require hugely accurate numbers... The 64bitness of an instruction set adds nothing!

What am I missing? anything?
And don't give me that crap about 64bits means more width for memory transfer. That's bullshit. Your nice little 32bit instructions are taking up that bandwidth already just by the recompile for 64bits.

Re:REPLY to this if you are a C/C++ programmer

[Hoser McMoose]

Ok guys I have one CPU question that is yet to be answered. Aside from increase memory access and integer/float width. What could the possible advantage of a 64-bit 3D game have.

General 64-bit vs. 32-bit? Not much. In fact, the float width doesn't even change (it's always been 64 or 80-bit on PCs and most other architectures), just the integer width and larger memory access. Of course, AMD64 also doubles the number of integer registers (and makes them true general purpose registers, as opposed to the semi-general purpose registers of IA32) and doubles the number of XMM registers (for MMX/SSE/SSE2). It also cleans up a little bit of cruft here and there.

Register size/Bus speed/hypertransport all can be added to current 32bit platforms.

Bus speed and hypertransport could easily be added to 32-bit platforms. The P4 already has a fast bus and Transmeta's upcoming Efficeon processor will use hypertransport as it's I/O connection to the outside world.

Register size though, well that's another story altogether. If you're changing register size (and the number of registers) you're breaking compatibility. If you're breaking compatibility anyway, why not fix THE main limiting factor for 32-bit systems, ie the fact that they can only address 4GB of memory.

It's not like you can "pair up" instructions now... those instructions that used to be 32bit, when recompiled simply take up 64bits now, right?

Not really. The instruction size stays the same with AMD64. The standard word size (int or long int, which are the same on most architectures) stays the same, still 32-bits. The only thing that doubles in size are pointers. You also have the option of using 64-bit integers (long long int) natively instead of having the compiler hack them into two 32-bit ints like with 32-bit processors.

If your video games don't require hugely accurate numbers... The 64bitness of an instruction set adds nothing!

If your game (or any application) needs hugely accurate numbers you're using floating point stuff anyway, and as mentioned above, that's always been 64 bit or even 80-bit in some cases (all x87 FPUs support 80-bit floating point numbers, though it's rarely used by compilers). The larger integers are just that, a larger range for integer numbers. It's fairly rare to use long long ints, so this isn't a huge deal. Of course, when you need long long ints you need to do at lest 3 times as many instructions on a 32-bit machine as compared to a 64-bit one.

What am I missing? anything?

You're missing a few things. First, games right now don't use more than ~2GB of memory, the maximum that 32-bit chips can support with things getting ugly, but they will. In fact, we're probably only about 2 or 3 years away from it being common to use 2GB+ of memory for games. Also, games are definitely NOT the only market out there. Using more than ~2GB of memory is already quite common for servers and workstations, and doing so on a 32-bit chip is not ideal.

The next thing that you're missing is doubling the number of registers. Sure, you could do that with a 32-bit chip, but if you are going to break compatibility you might as well fix as many potential problems as possible in one fell swoop. This also nicely offsets any potential performance loss you might see by doubling the size of your pointers (ie more memory bandwidth requirements and more strain on cache). In fact, by doubling the number of registers you often end up with FASTER code when compiled for AMD64 as compared to IA-32, while with most other processors (eg the G5 or an UltraSparc) you typically end up with slower 64-bit code vs. 32-bit code. The extra registers not only mean more space to store data (rename registers can handle that on 32-bit arhcitectures), but fewer load/store instructions, thereby lowering the size of your code and reducing cache and bandwidth use.

In the end most AMD64 code runs about the same speed as IA-32 code. Th

Re:REPLY to this if you are a C/C++ programmer

[p3d0]

I can't think of any reason a game would use 64-bit integers. The advantage of AMD64 for games would be from the extra registers.

The ULTIMATE cranky oltimer rant:

[Thud457]

You had fathers?!!!

45 Pages?!?

[mj2k]

45 pages? This is /. Look at posts on other articles, you're lucky to get us to read a 1 page article much less 45.

Re:Comparison to a G5?

[Nazmun]

Dang i just lost my comment... time to rush... if you really need to know about the performance differences. Athlon64's have a far lower memory latency and the fx series also has twice the latency. So if your dsp processing (if that includes audio/video) requires lots of bandwidth there then you'll have a great increase there. A athlonxp has at best a 400mhz fsb while the g5 has from 800-1000mhz fsb. A64 if you use the maximum channels on the thing is advertised with 1600mhz. On the chip itself is 2x the cache (depending on which processor you buy though, some lower end models have 512k)... more importantly you have one more fpu unit which gives about a 50% increase in fpu capability (used to have only two units).

Re:Comparison to a G5?

[Nazmun]

Also no... i don't think it runs rings on a g5 either... It's just a lot faster then the axp and if your using the fx line it's also a lot faster then the p4's in pretty much every area except the athlon's weakpoints where it's tying at the moment.

Re:Comparison to a G5?

[ameoba]

DSP code, eh?

So you probably want something like... umm...

Vector Processing?

XP2500+ does MMX, SSE & 3DNow!
G5 has Altivec

No comparison; Altivec is going to smoke the XP2500+'s vector capabilities.

Athlon64 not only doubles the number of registers, allows you to easily move 64bit values around and gives you more memory bandwidth but it also adds the SSE2 vector extensions. Considering that the Athlon has a pretty nice FPU unit already, sounds like most of the points where your G5 surpassses the AMD chip have been addressed.

I cann't give any definate benchmarks; short of having equally good vectorizing compilers or equally good hand-tuned vector optimizations, we really can't compare the 2 archs on this kind of work.

Re:Smaller Die Space

[Atmchicago]

If I remember correctly, some of the processors, due to glitches in the manufacturing process, have only half usable caches. So AMD disables the half that doesn't work, and sells it for cheaper, instead of discarding an otherwise good processor. I could be wrong about this.

Re:BOW DOWN TO YOUR CORPORATE MASTERS

[shamino0]

Actually, putting 6 zeroes after my negative bank account would be anything except a deterrent for suicide :)

Depends on which side of the decimal point you put them on :-)

HP ! delivering XP64

[PSL]

"Windows XP 64-Bit Edition Version 2003 supports the AMD64 instruction set, however it is currently only available if you are a MSDN (Microsoft Developer's Network) subscriber, or if you buy a system from HP with the OS pre-installed."

I called HP 2 weeks ago and they told me that the XP that was shipping with the AMD 64 was not the 64 bit OS rather the 32 bit OS. And I asked him what the point of buying the 64 bit CPU. He was a bit confused and told me that the 64 3000+ would run slightly faster than a 32 3000+ but it wouldn't be that much difference.

One thing to note is MS does have 64 bit OS's for the 64 bit Intel. But intel doesn't have a 64 bit cpu out yet. Go Figure.

Re:BOW DOWN TO YOUR CORPORATE MASTERS

[Vihai]

Yup... I should have known... damned Euro... I want my Liras back!

Windows Longhorn

[MonkeyCookie]

Don't worry about excessive processor speed! Windows Longhorn will come along and put all that processor time to good work, drawing fancy-schmancy graphical alpha-blended 3D animated animals that annoy the hell out of you!

When compiling with gcc for 64-bit host,

[Anonymous Coward]

instead of for a 32-bit host, how do you write the following lines:

typedef int signed_32;
typedef short signed_16;

My Experience...

[tacokill]

Well, I just did this AMD64 upgrade so I speak with a little experience.

Past config:
AMD 1.4ghz CPU (Not XP or MP)
Abit Mobo - VIA KT133a
256meg PC133 SDRAM (Corsair - good stuff)
few drives, dvd-burner, etc
Nvidia Geforce 4200ti vid card

Current config:
AMD64 3000 (2 ghz, I think)
Gigabyte K8NPro Mobo: with many more features than I previously had (RAID, sounds, giga-lan)
256meg PC3500 DDR (Hyper-X - good stuff)
...and all the other stuff

Total Cost of Upgrade: $440.00 --- and yes, it is MUCH faster than my old machine. By any measure.