AMD Bumps Up Socket AM2 Launch Date

Thrill-Ki1l writes "According to DailyTech AMD has moved up the launch date for their new socket AM2 processors. The manufacturers of the new AM2 chipsets and motherboards have their hardware ready to ship early so AMD decided to launch the chips 2 weeks early. The new launch date is May 23rd."

Comments

condolences

[caffeinemessiah]

to everyone who just dropped a ton of cash building a brand new athlon x2 socket 939 systems...my condolences. at least this will bring the price of the higher-end athlon x2s down for the rest of us...yay!

Re:condolences

[mrchaotica]

You would hope anyone building a system would have done their research and would have known that new chips were coming.

Re:condolences

[Bin Naden]

I didn't get screwed: I bought the cheapest CPU with a socket 939 motherboard and will buy a better CPU once the prices go down much like 754 cpu went down once 939 came out.

Re:condolences

[ePhil_One]

You would hope anyone building a system would have done their research and would have known that new chips were coming.

Really, the jokes on the people who by this new socket. I hear that eventually it well be replaced with something even newer! I'm waiting until 2019, when there will be no more computer upgrades (society will collapse July 17th, 2019, ending all new product development short of the flint arrowhead)

Re:condolences

[Bob_Robertson]

Actually, the release of new hardware should drop the price of dual and quad 939 motherboards to buyable levels, as well as the dual core 939 CPUs.

It's not like the 939 performance has dropped, only that the bleeding edge has cut another swath. Let them go, and enjoy the savings!

Bob-

Re:condolences

[bersl2]

How about those of us who dropped not so much cash on such a system?

As long as I catch DDR1 prices at its lowest point (to reach 2GiB RAM total), my Opteron 165 @ 2.49 GHz system should cut it for a while.

Re:condolences

[mrchaotica]

Well, when you're this close to the launch of the new socket you might as well wait in the hopes that you'll be able to upgrade the CPU or motherboard separately once over its lifetime, if you so choose.

Re:condolences

[toddestan]

Actually, I feel sorry for the people who built 754 systems back in the day. It seemed that AMD moved onto Socket 939 pretty quickly after that one.

Though I really feel for anyone who has a Socket 423 Pentium IV system. Very short lived standard (1.3-2.0Ghz), expensive rambus memory, 100Mhz bus ("quad pumped" to 400Mhz), and really odd CPU coolers that screw into the motherboard that are virtually impossible to get replacements for.

Re:condolences

[TubeSteak]

I'm waiting until 2019, when there will be no more computer upgrades (society will collapse July 17th, 2019, ending all new product development short of the flint arrowhead)

You know that very few people use the Mayan calendar anymore?

It was the flint arrowheads that gave you away.

Re:condolences

[PayPaI]

According to the article S754 is going to actually outlast S939.
AMD's Socket 754 to Outlast Socket 939

Re:condolences

[buswolley]

I just built an AMD 64 3800+ X2. I researched for quite a while, and was completly aware of the new socket. This does not bother me.

1)I can't really imagine anytime soon when I will need more power than I have with this system.

2) The price was good.

3)You can overclock the shit out of them.

4)I can scan for viruses, rip a cd, and play a top end game concurrently without a glitch or slow down.

5)It is a stable mature technology(socket 939). I previously made the mistake of purchasing the first edition of a new intel socket, and regretted it for years.

Let the new sockets, chips and boards work out their chinks before you buy, I say.

Re:condolences

[Nutria]

So Im happy as hell with my X2 4800+ and 3GB of DDr1 with Nforce4 chipset and 2TB of disk space and Geforce 7800 GTX and will be for quite some time to come.

Geez, how could you not be happy with that system? Must be kinda noisy, though.

Re:condolences

[tlynch001]

I can scan for viruses, rip a cd, and play a top end game concurrently without a glitch or slow down

If you ran linux you wouldn't have to worry about the virus scanner.

I guess you would'nt have to worry about the game either...

Re:condolences

[drinkypoo]

I have a hard time feeling bad for people who went to systems supporting RAMBUS. People who don't know shit about computers wouldn't know - but they should be asking someone who does know about 'em what they ought to buy, and anyone who makes an unresearched purchase can just fuck off. Including me, when I do something that dumb.

summary

[The_GURU_Stud]

what should also be on the summary is that it was bumped up to coincide with Intel's paper launch.

Is it really worth it?

[Bin Naden]

Is it really worth it to be an AMD processor with a DDR2 memory controller? Sure this meant a lot for the intel architecture but from what I understood the AMD architecture will not gain a lot from this memory speed increase. Also, the latency on the memory will likely also increase which might cancel all gains made from the increased speed. Therefore, I'm waiting and seeing before I get me one of those.

Re:Is it really worth it?

[MBCook]

There was an article somewhere recently (Anandtech? Tom's Hardware?) checking performance of DDR2 versus DDR on the Opteron. They determined that DDR speeds below 533 (IIRC) would hurt the performance. At 533, it was about even. As the processor and or memory speeds up then you will see the benefit. This isn't strictly necessary right now. I think it was actually due to the latency issue that you mentioned that this was the case.

I'd rather see FB-DIMMs, personally. But the move to DDR2 was going to happen at some point. Better now (when it's not necessary and people can still choose a great processor and DDR combo) then later (when DDR is more expensive and they were hurting for the change).

I seem to remember that was going to be something else with this socket upgrade (in the form of processor features) that was more interesting or offered better performance increases than the memory change. I don't remember if it was SSE4 (is that out yet?), a better branch predictor, AMD's Vanderpool (can't remember the name), or what.

Re:Is it really worth it?

[PsychicX]

This article examines the question in depth. Hope that helps.

Re:Is it really worth it?

[Bin Naden]

I'd rather see FB-DIMMs

A maximum of 192 GB of RAM... Is that enough for windows Vista?
http://www.anandtech.com/cpuchipsets/showdoc.aspx? i=2741&p=9/

Re:Is it really worth it?

[ArbitraryConstant]

"Sure this meant a lot for the intel architecture but from what I understood the AMD architecture will not gain a lot from this memory speed increase."

Early indications are that Intel's architechtural improvements with Conroe will give them a significant edge over AM2 chips, even though AMD still has an on-die memory controller and Intel doesn't.

"Also, the latency on the memory will likely also increase which might cancel all gains made from the increased speed."

They'll be using 667 and 800 mhz clock speeds, so it won't be too bad, but the gains from the extra memory bandwidth aren't as big as one might expect. At least for AMD.

"Therefore, I'm waiting and seeing before I get me one of those."

Sound advice for any pre-launch situation. :)

I'm going to take the wait-and-see approach as well, but so far looks like I'll be continuing my pattern. My last few computers have been Pentium II, Athlon, Pentium 4, and Athlon 64.

Though I rarely get top-of-the-line chips, so if AMD cuts the prices for their high-end stuff when Intel has faster parts, that might make them preferable even if Intel holds the performance crown at the time.

I mostly want dual cores and hardware-supported virtrualization so I can run OpenBSD under Xen. Both AM2 and Conroe will have that, so it's pretty much a question of who gives me the best performance.

Re:Is it really worth it?

[interiot]

Yeah, Pacifica (AMD's hardware virtualization support) is going to be available for the first time with AM2.

Re:Is it really worth it?

[Lonewolf666]

AFAIK DDR2 brings better throughput but at the expense of more latency. For the current AMD chips the overall result seems to be equal, but once you go to more cores I guess the throughput will become more important.
A few months ago I've read a review of AMD Opteron vs. Intel Xeon dual cores. For one socket, both systems had similar memory bandwidth and while the AMD was faster, the difference was not dramatic. For two sockets (4 cores total), the Xeons still had to make do with the same memory bandwidth because they shared the memory interface, while the Opterons gained a second memory interface because each socket has its own memory controller. In the test results, the two-socket Opterons completely smoked the Xeons. Conclusion:
When four processor cores have to share a memory interface, it better be a damned fast one.

Now AMD has announced that they want to introduce four-core processors eventually. Hence the need for plenty of bandwidth.

Re:Is it really worth it?

[hattig]

Just a FYI as you might be interested, Anandtech today did another Opteron vs. Xeon test, this time with 8 cores per machine, 4 sockets.

The 2.6GHz Opterons stomped on the 3.0GHz Xeon MPs by around 30-40%.

Article

Woodcrest (Conroe based server part) will have the same bandwidth restrictions that hurt Xeon in the above test. Expect to see the usual suspects test four socket Woodcrest with cache intensive benchmarks.

However they're not a large part of the market really.

Too many sockets!!!

[Freaky Spook]

Although I find AMD to be better chips, I still stay with intel because their chips are all on one socket.
AMD at the moment offers no upgrade path because their Semprons & Athlons are different sockets, you can't turn a budget AMD box into something more powerfull without replacing the motherboard.

Intel allow people to start with a Celeron and easily upgrade to a P4 or Pentium D if more CPU power is required, I find the upgrade option far better value.

Re:Too many sockets!!!

[Bin Naden]

The reason that AMD had to go this route is that the AMD architecture has the memory controller embedded in the CPU chip while the intel machines have the controller contained in the motherboard. This means that AMD have better overall memory performance, however, a change in memory technology forces them to redesign their CPU and use another socket for fear that consumers might put those new CPU's in older 939 motherboards and either fry them up or call for tech support too much.

Re:Too many sockets!!!

[Z0mb1eman]

Y'know... I used to reason along the same lines. After 10 years of building/upgrading my own computer, however, I realized one thing:

Not once has it made sense to upgrade the processor. Every time it would've provided only a small gain compared to the expense and hassle. I've upgraded the RAM, hard drives, and video card several times - but by the time I felt I needed a better processor, it made more sense to replace the entire motherboard and go with a new generation (or build a new box entirely).

To each his own, of course. But in the future I'm going to worry about the upgrade paths a motherboard offers for the RAM, drives, or video - and not the CPU, because I know I'll never bother.

Then again, so far I've mostly had AMD boxes, so maybe there is some truth to what you say. :)

Re:Too many sockets!!!

[forkazoo]

Personally, I feel AMD has been a bit more reasonable with sockets. Pentium 4's have at least 423, 478, 775, and Pentium M has 478 and 480. Socket A lasted a long time, overlapping with Socket 370, IIRC. Since then, we've had 754, 939, and now AM2. (I'm ignoring 940, but also ignoring the Xeon sockets...) Since AM2 isn't actually out yet, that's two sockets that have been released in the same time frame that Intel released five.

Re:Too many sockets!!!

[Spokehedz]

Err... the P4 Comes in two socket styles. The old mPGA and the new 755-socket-whatcha-ma-callit. So right away, your argument is null and void.

The 'old' 468-pin is just that--the OLD socket. Meaning, it's a completely different chip altogether.

AMD has had some issues with sockets lately, I will grant you that. But Intel has been behind the 8-ball for a while now. AMD is now the innovator, and they are just playing catchup.

Re:Too many sockets!!!

[mrchaotica]

Socket A lasted a long time, overlapping with Socket 370, IIRC.

Nope -- you forgot Slot A. (I got bit by that one...)

Re:Too many sockets!!!

[Gwwfps]

Not to mention all the different chipsets for 775 that are not compatible with all 775 CPU's.

Re:Too many sockets!!!

[billcopc]

Amen! As a shop tech I hate Intel for NOT diversifying their sockets. For the most part, each chipset is tied to a specific series of processors. It doesn't matter that it fits in a socket 478, if the chipset doesn't like the CPU it's not gonna boot PERIOD. Instead we have to worry about families like Prescott, Northwood, Willamette.. or even worse their friggin product numbers that you have to break down to figure out the speed and features of the cpu. Try telling a relative on the phone to buy a Pentium 672.. much easier to say "Get an Athlon 4000!".

Excluding old sketchy motherboards (ECS, Gigabyte), you can pretty much stick any Socket-A AMD on any board. Same thing with the 939, even the early boards will run fine with a screaming new dual-core.

As for the upgrading issues, it's very rare that you'll swap just a CPU, unless you bought a crappy CPU to begin with. Most people who want to future-proof their PC will buy the best kit they can afford and make it last. They won't get a budget CPU that they know will be obsolete within a year or two. We're still in a young computing age where everything is in great flux. Perhaps in another decade or two, computers will have reached a point where things have stabilized and one set of mainboard, ram and power supply can last through several generations of processors. For now, that's just a penny pincher's absurd dream.

Re:Too many sockets!!!

[Saeger]

Not once has it made sense to upgrade the processor.

Usually you'd be right that it makes no sense to upgrade the CPU in leu of building a completely new system around it, HOWEVER one CPU upgrade that does make a significant difference for a lot of people these days is going from single to dual core -- if your motherboard supports it (via bios upgrade or not).

I just upgraded from a uni AMD64 3000+ -- that I built with my MSI Neo4 Platinum system a little over a year ago -- to a dualcore AMD64 X2 3800+ (best bang/buck) and the difference in FEEL is night and day no thanks to the small GHz bump. This was the first time I've bothered to replace a CPU on the same motherboard in over 10years of building my own systems. socket 939 might be end of life, but a dualcore upgrade is a worthy exception.

Re:Too many sockets!!!

[tomstdenis]

I fell into this trap when I learned by new 915G ASUS board wouldn't handle an i820 smithfield core.

Intel may use 775 for everything nowadays (except for Core Solo/Duo btw) but the required chipset changes. To be safe on the Intel camp today you need >=i945 northbridge. But that didn't help all the 915G users like me back in the day.

Might be safe to point out that unlike Intel at least AMD has multiple vendors of chipsets. So while AMD is going through transitions (hint: So will Intel be when they realize that an FSB for memory is a dumb idea) you're not stuck on one vendor for getting the requisit parts.

Tom

For those who don't know about it...

[Spy der Mann]

Here's good info on the AM2.

fake?

[muszek]

The story is fake. Nothing in our industry is released early.

Re:fake?

[Bin Naden]

This is what we call a paper release as pioneered by IBM in the good old days. When the competitors threaten to release a better product then yours, you claim that you are developing an even better product than them in order to prevent consumers from switching to the competitor. The actual release date doesn't matter, all that matters is that consumers anticipate that you will have an awesome product coming along that they will want to save money for. And as IBM proved, you don't even have to release the product. You might just want to do this in order to cause your competitor financial hardships.

939...

[Hyter]

and 939 was suppose to be "future proof."

Re:939...

[Bin Naden]

and 939 was suppose to be "future proof."

And 64k was supposed to be more than anyone would ever need.

Re:Why bother?

[IlliniECE]

Its not trolling. Its true. Intel proved this at IDF. Furthermore, AM2 really is all promise without delivery http://www.anandtech.com/cpuchipsets/showdoc.aspx? i=2741

But how much more money will it cost?

[layer3switch]

I am not talking about superficial retail madeup price. By past pricing scheme with early release claim, the tendency normally bend toward higher price otherwise cheaper with later release.

I was looking forward to get it when it came out, but if higher price tag came along with it because of earlier release date, I would just wait.

AMD's AM2 processor seems to be DRM-free

[UseFree.org]

Hastening the launch date by a couple of weeks isn't too significant, but AMD certainly deserves to be congratulated for (apparently) leaving DRM out of their AM2 microprocessors. In contrast, Intel has succumbed to RIAA/MPAA pressure and betrayed their customers by stuffing Treacherous Computing down their throats.

I'm also happy to see that AMD has not put DRM into its AMD Live! technology, which competes with Intel's DRM-ridden Viiv. I'm sure AMD is taking a lot of heat from the entertainment cartel for not handcuffing users, and I hope they'll continue to keep their products DRM-free.

And let's not forget that AMD has been supportive of LinuxBIOS by actively ensuring that their motherboards can run it.

Re:AMD's AM2 processor seems to be DRM-free

[heinousjay]

So tantalizingly close to on-topic. Excellent zealotry, A+.

What about power use?

[Entropius]

I've heard power consumption cited as one advantage of DDR-2 over DDR. On laptops that idle around a dozen watts, even a few hundred milliwatts in the memory subsystem can be a signficant change.

Re:Wow!

[Ajehals]

Yeah, cause Vistas coming early... No.. wait.. Sorry

That old link?

[John Jamieson]

Fine, and I can show you an article that says the 65nm Athlons will clock 40% faster, and the conroe is actually slower if you don't fit in the 4Meg L1 cache. (anyone can add cache)

Why don't we wait 6 months and then start trash talking, when we have actual products.

One of two things has happened
1. AMD has become complacent and has no strategy of really updating a now old product this year. In this scenario they were lulled asleep.
2. Intel has stunk so bad that AMD has been holding some cards close to the chest because it did not need to play them.

No staged demo by either AMD or Intel will give us the answer, but we will know in 6-8 months.

Why, exactly?

[Weaselmancer]

You would hope anyone building a system would have done their research and would have known that new chips were coming.

Putting aside that new chips are always coming up, why would this matter?

Only reason I can think of is that you're suggesting that you might want to put a faster cpu in later on. Is that it? If so...is that a really common thing to do? Because it seems kinda....well, dumb to me. You unplug your existing cpu and stick it in a box. Then buy another one that's only a few percent faster. Then unplug that 3 months later...and stick it in a box. Seems like a waste of money to me.

Every time I've done an upgrade, it's been a whole system upgrade. And then, I only do it every 5 years or so. Is there really a need to stay on the bleeding edge all the time that I'm missing?

Re:Why, exactly?

[Znork]

"Then buy another one that's only a few percent faster."

As far as I can tell, AM2 will be a merge of the desktop sockets, which means you'll eventually have a fairly large range of performances.

If AM2 reclaims the same kind of staying power that socket A had, this also means you get the capacity to do cascade upgrades once you have a few systems. Stick a new CPU in your desktop? You dont stick the old one in a box, you move it to the server, which gets faster... and your old server CPU can be moved to the media frontend, which also gets faster... etc.

"Every time I've done an upgrade, it's been a whole system upgrade. And then, I only do it every 5 years or so. Is there really a need to stay on the bleeding edge all the time that I'm missing?"

Mmm, if that's how you do it, you probably wont benefit, no. In fact, it's far better to _never_ stay on the bleeding edge; bleeding edgers always get hosed by the price/performance ratio, and someone spending half of what you spend every 2.5 years will probably have a higher average performance over time on his systems, and twice as many systems. And again, once you reach the number of systems and component standardization needed for rolling upgrades, that's when you really start getting the payoff...

Re:Why, exactly?

[imboboage0]

Is there really a need to stay on the bleeding edge all the time that I'm missing?

[sarcasm]But what about your SETI scores?!?!? And PCMark! Those are my life! You're such a n00b![sarcasm]

I kinda like my 939. if i wanna upgrade from 3000+ to 4800+, i can. that's a huge improvement still available.

Re:Way to go Apple!

[ClosedSource]

"It's pretty funny to see the UNIX-hating, Intel-hating Mac zealots of the world to have converted within a couple of years into pro-UNIX, pro-Intel zealots that talk like Apple invented UNIX and the x86."

Are you sure its not just UNIX fans' imagination that Mac zealots have become pro-UNIX? It seems to me that Mac zealots like whatever Apple makes, they don't care much about the underlying technology as long as the user experience is good. I seriously doubt that long-time Mac users are getting excited about the OSX command line interface.

Memory Capacity?

[Bill Dimm]

I may buy/build a workstation soon that will need the capability of holding a large amount of RAM (say 32GB - calculations requiring a lot of memory but not a huge amount of CPU) at some point down the road (I can probably put off fully populating the memory for a while). I came across this post, which seems to say that motherboards for DDR2 will allow more DIMMs (16 2GB sticks is a lot cheaper than 8 4GB sticks right now, at least for DDR). It is talking about DDR2 with Opterons. Is there a launch date for DDR2 on Opteron? Is the capacity actually greater with DDR2? Is DDR likely to become scarce down the road, causing DDR2 to be a cheaper option for future expansion? Any opinions are appreciated (I haven't had an excuse to buy hardware in a long time, so I haven't kept up on such things).

Sidenote: Yes, I am aware of the iWill DK88 (16 DIMMs DDR) - anybody have any experience with it (especially with Linux)?

AMD socking it to Intel

[mikesd81]

It sure looks like AMD is socking it to Intel with the lawsuit and now this release.

It doesn't help either that HP/Intel's Itaniums aren't taking off and the AMD Opterons are doing well. Plus, as others have mentioned, AMD seems to backing Linux pretty tough. And I was looking @ alot of embedded devices and I see more AMD based ones than Intel.

It's interesting how far AMD has come in the last 10+ years.

Re:Way to go Apple!

[tomstdenis]

No, Intel is releasing a dual-dual-core not a quad-core. It's going to be two separate dies on one chip.

Intel still hasn't figured out the whole "let's go MP" part yet either. While they are doing things like L2 sharing to speed up proccesses in the same dual-core part (which is a mixed blessing though) they are still using the single-FSB to do MP.

Tom

Re:Upgrade Horror

[tomstdenis]

I don't get this argument.

First off, the only way DDR2 will suck is if they never hit 800Mhz speeds. Since there is a lot of pressure on this I'm sure the memory manufacturers will keep at it.

Second, PCI-E replaces AGP [which was around for a long time]. PCI-E is the new standard for a lot of high bandwidth parts including things like network controllers. So while I can imagine in 5 years time we will start hearing about a new bus coming out, PCI-E will still be around. Christ we still have PCI on motherboards and that was introduced in the early 90s [I recall 486 boards with ISA/VLB/PCI slots!!!].

Third, don't be a fool. Buy what you *need* not what you think will be patchable.

If you really need two cores with a decent cache get a 4400+, 4800+ or FX-60. Those parts are pretty darn fast. I mean I know people who STILL use P3-MP setups running at 550Mhz today... I'm sure a dual-core 2.2Ghz part will be fast enough to compile or run desktop apps for years to come.

Just buying AM2 because "it's new" is stupid, unless that is, you're a developer who wants to support all new platforms.

The trick though is to figure out what you actually need in terms of computing power and topology and buy to that. Otherwise, you'll just waste your money buying stuff you can't use wishing you had something else.

Tom