Slashdot Mirror
A First Look at AMD's M2 Platform
Knight Thrasher writes to tell us that Tom's Hardware has an interesting first look at AMD's AM2 platform. From the article: "While Intel will be answering later this year with its Merom/Conroe processors, AMD officially says that the introduction of its AM2 platform and DDR2 memory support in the second quarter of this year will be able to maintain its current lead. Unofficially, we know that AMD will launch six dual-core and two single-core AM2 processors on June 6 - later than initially expected but well in time for Intel's Conroe, which will be introduced in September. Tom's Hardware got its hands on a stable engineering sample of an Athlon 64 X2 4800+ for Socket AM2 and will publish benchmark results as first as a first impression of the new Socket and processors tomorrow."
and will publish benchmark results as first as a first impression of the new Socket and processors tomorrow.
Nice to see the Editors are living up to their name.
Your hair look like poop, Bob! - Wanker.
So, Slashdot is now referring to articles which will be up tomorrow?
Last M2 I remember was a Cyrix. q-:
-Ben Vander Jagt
1) Publish story about fancypants new platform being delivered to reviewers
2) Reveal that the benchmarks won't be available until later
3)
4)
5)
6)
7) Dupe!
(Actually 4-6 are also duplicates)
Are the boards the proc sit in designed to dump BIOS yet? Or are we still stuck?
09 F9 11 02 9D 74 E3 5B - D8 41 56 C5 63 56 88 C0 45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
FB-DIMMs should be available by now. If I would go out and buy a socket M2 processor, I'd have to buy a new socket and processor when FB-DIMMs came out (or the switch to DDR3 or whatever). If we had FB-DIMMs then one processor would work with DDR/DDR2/DDR3/SD/whatever just by switching out the memory since the interface is serial and built onto the memory chips. It would allow the life of boards to be extended much longer. Look how long PCI lasted. If you bought a new motherboard in the PCI era and you could keep using it all the way up to now because the socket stayed the same and the memory modules just changed (even though the physical pin out stayed the same) you could do it. Now that PCI-Express is here, we could do that easily for the future.
FB-DIMM is supposed to simplify the board layout too since you don't have to run all those parallel data/address lines to each DIMM. This is supposed to make layout much less complicated. Imagine how many pins would be needed on an Opteron if they wanted to put 4 memory banks on the processor instead of the 2 they have now. That would be a few hundred extra pins. With FB-DIMM that might be one hundred extra pins.
The only need to update the socket would be to provide additional power pins (you could future proof this a bit by putting extra power pins on) or other features (I've heard of someone, Sun perhaps, trying to put Ethernet on the processor die).
I like AMD, but isn't it time we get past these custom memory interfaces for each standard?
Comment forecast: Bits of genius surrounded by a sea of mediocrity.
Why can't they just make a socket with 200 pins and stick with it for a few years? I'm very tolerant when it comes to planed-obsolescence, but this is just getting stupid.
I thought DDR3 was the future?
I read that it is expected late 2006/early 2007 and Samsung claims it'll be 2x the speed of DDR2 and it'll operate at 1.5v (less power consumption).
I know NVidia is already using it on video cards...
[Fuck Beta]
o0t!
Well, they are German...
Not a moment too soon...any earlier, and I might not have just splashed out for a top-of-the-line socket 939 system. Hooray for the expensive future!
i've just bought an 939 for my bro. he will kill me when he knows it will be obsolete in a no-time. platforms should last longer...
Yes, with FB you can put all your DIMMs on a single set of signal lines. But you can without FB, too (within capacitance limits). The reason you don't in either case, is regardless of FB or no, if you have more parallel lines, you can transmit more data at once.
If you put more memory on one set of signal lines, you can reduce latency, but you cannot increase bandwidth.
http://lkml.org/lkml/2005/8/20/95
I'd rather read the AnandTech article on AM2
When Stephen Job announced his "I-Minor" McIntosh last year, it really caught my eye. Wanting to buy or build a small computer for my already cramped breakfast bar, I started pricing out similar hardware. The results startled me. Most of the configurations I found cost more than the humble US$499 of the "I-Minor", often much more. To match price with MAC I had to configure with a much bigger shuttle-style case.
So here's my question. What computers are currently on the market to compete with this? When my wife asks for the "cute little I-Minor McIntosh with dotMax Tigger OS® that MAC just invented", what PC can I buy instead?
Being sort of a slow news day, I tried something different and actually read TF "article", as there were no comments featuring "in Soviet Russia" yet posted. First off, SPONSORED LINKS are evil and annoying.
I see the blurb mentions introduction of 6 dual core & 2 single core chips, and I wonder if this will be the new product tier differentiating mechanism: dual and single. Traditionally, we'd see the low end, which was the crippled version of the mid-range, then the high end typically added more cache and un-crippled SMP abilities. Perhaps the low end will be single core, mid-range dual, and high end w/larger caches & 4/8-way ability.
Now that the MHz "wars" seem to be behind us, it's a race to pack multiple cores onto chips, which I see as a good thing. I've always had a thing for SMP rigs (my current & previous boxes are duals), and dual-core going mainstream means several good things for us SMP freaks, the least of which is more affordable 4-way boxen!
In closing, I'd like to mention that this whole blurb about a story (which is in fact an ad vehicle) which references a yet-to-be published story, is rather silly and bizarre. And poorly written. Like my post.
Nice to see the Editors are living up to their name.
I think you mean the 'Editros.' Yeesh, get it right. ("Don't call me 'Yeesh.'")
I'm more interested in announcements from PMA, anyway.
When you think of it, apart from dual core there hasn't been much movement in the processor industry since ... 2003. All we've really seen is small tweaks ... and an emphasis on lower power usage.
Lower power usage is precisely what makes residential dual core PCs and blade servers feasible. The same attitudes toward lower power usage will be necessary to control heat flow once Moore's law runs out (not soon, but eventually) and we end up having to build into the third dimension with stacked dies.
If they actually release it on June 6th 2006, (6/6/6) they should work out some kind of deal with ID to include a free copy of Doom3 with each processor for the rest of the year. If not, they should do SOME kind of fun thing with it. Although, they will probably chicken out, just like when they clock doubled the 333mhz processors, somehow they ended up with 665mhz.
Nobody believed them when they said that they won't make you buy a new mobo to upgrade to dual-core processors. Amazingly, AMD kept their promise! They even migrated some Opterons to 939 so you can upgrade your home computer with a real server chip. Now compare this to Intel and you'll see how disciplined and customer-friendly AMD have been.
Of course, they want to make use of DDR2, and since your old motherboard doesn't have DDR2 slots, you'll need to buy a new motherboard to use DDR2. That's the end of the story! You'd have to be high to think you could keep your board and just upgrade to DDR2. AMD switched the pinout a tiny bit so that you don't make the mistake of plugging in an incompatible processor into the board. There's nothing more to it than that.
So maybe people are complaining about being forced to go to DDR2, but I don't think that will happen. I'm quite sure there will be several new AMD processors for Socket 939, probably priced at the same level as their AM2 counterparts. The only difference will be the memory controller. Of course, it won't make much sense to buy 939, with DDR2 being almost as cheap as DDR.
Maybe people were complaining about the extra burden on mobo manufacturers to retool, but this is absolutely minimal, as the Anand article makes clear. We will see many cheap AM2 boards almost right away, because they are so similar to Socket 939 and 940.
Really, this is a great illustration of how a socket change should look.
I thought FB-DIMMs were merely buffered (as the name implies).
Nope, they're not. They're arranged in a big serial shift register.
The problem with this? Latency goes up as you put in more DIMMs. Why? Because data from the 4th DIMM has to pass through (not just by) the 3rd DIMM, 2nd DIMM and 1st DIMM to get to the CPU.
Sound familiar? It's just a retread of RDRAM.
No thanks. Intel boned themselves with this before. If they want to push this, they better get ready to take a backseat to AMD again. DDR outdistanced RDRAM handily on performance and price/performance, I'll be surprised if things are any different this time.
I hope you enjoy your higher clock speeds, you'll need them to try to get your latency down to managable and your bandwidth up to normal. I mean, with 1/4 as many data pins (I assume the 28 data pins carry only 16 bits of data at once, vs 64 of DDR/DDR2), you're going to have to go 4X as fast just to match the bandwidth and latency of a regular system. And as soon as that 2nd DIMM is put in, you're behind on latency and you're going to have to play catch up.
http://lkml.org/lkml/2005/8/20/95
"Are the boards the proc sit in designed to dump BIOS yet? Or are we still stuck?"
UEFI 2.0 isn't fully baked yet, but it will be soon. EFI 1.10 is an 'Intel' spec, and besides, it won't do 64-bit extensions. Moral, the suk.
Look for there to be UEFI 2.0 boards later this year, most likely at the Vista launch, but not before.
-Charlie
Since 940 was ECC and 939 was not ECC, will there need to be two different sockets for DDR2 ECC and DDR2 standard?
Yes! AMD has DDR2, I wonder how long it will take Intel to catch up with them on that one....oh wait, nevermind.
So where do we get AMD boards with something other than BIOS? Apple needed features which EFI could provide while BIOS could not.
Jesus was a compassionate social conservative who called individuals to sin no more.
(and don't even get me started on why Cell wouldn't make a goood desktop processor)
James P. Barrett
A video-card? Allow me to clarify: Video-cards routinely have 256bit mem-buses right now, and they have RAM that runs at around 1GHZ, giving them metric assload of memory-bandwidth. In order to achieve that, they use RAM-chips that are soldered right on to the board, and they have hefty heatsinks. What if processors had something similar?
Processors would be sold in cards not that different from vid-cards these days. They would connect to a slot, and they would fhave the CPU, and attached to that CPU would be about 512MB (maybe more, maybe less) of very, very fast RAM on 256bit bus. Of course, it would cost a bit, but not more than video-cards do today (I bet that GPU's are more expensive to make than CPU's are). Yes, there are issues of memory-expansion, but what if there were regural DDR2 mem-banks attached to the northbridge on a "normal" 128bit bus that could be used for additional memory?
If we had a SMP system with this kind of setup, it would offer A LOT of bandwidth. Each CPU would have very fast RAM attached directly to it. And they could access the RAM attached to the other CPUs. AND they could also access the RAM attached to the northbridge.
Or maybe if they used the locally attached RAM as L3-cache? 512+MB of cache, anyone?
Is this idea completely stupid, or does it have some sense to it?
Lesbian Nazi Hookers Abducted by UFOs and Forced Into Weight Loss Programs - -all next week on Town Talk.
Guess what, the submission was rejected, and yet in the same day they let through a story about a self perpetuating miracle motor. Why do I even bother!
AMD just licensed something called zram which will allow them to bundle oodles of L3 cache in the near future.0 216034806.htm
http://www.geek.com/news/geeknews/2006Jan/bch2006
ZRAM takes 1/5 the space typical SRAM does. So they could have 5MB of cache for the same amount of space 1MB of "normal" cache would consume. Still nowhere near 512MB ;). Although the latency/bandwidth would be higher.
Lesbian Nazi Hookers Abducted by UFOs and Forced Into Weight Loss Programs - -all next week on Town Talk.
when you could snap a x486 chip on top of your existing x386 chip?
My father when that route to upgrade an IBM 65SX. I was amazed something so simple actually worked. It gave new life to an old machine.
I think what prevents most people from ever replacing the chip in their system is that by the time they want to they are so frustrated with their old system they just want to replace it entirely. By frustrated I mean they have loaded it down with so much junk it is just a bear to use.
Also throw in the fact that many systems are really tight fits inside. Then top it off with heat sinks that can be a royal pain to remove or put on, all the while making you feel as if your going to break something and people won't even try.
Back in the 386 days when the whole machine cost big money a processor upgrade was a viable alternative for mom and dad. Today when you can get a new machine for less than $200 after rebates why would they bother?
* Winners compare their achievements to their goals, losers compare theirs to that of others.
I have a P4 2400 with 512 mb DDR1, and an AGP 4x slot. And I have no need to upgrade, except maybe buy some cheap 1 gb RAM to play with. I mean, what's the point? I will hardly notice any difference. When AMD release some 4x CPUs and their dualcores get cheaper... Or Intel's for that matter... Then, maybe, I will think about upgrading, but right now, its just completely pointless for me. The only thing that's troubling me is that I will have to buy an AGP card (maybe it will be a 6600gt) if I want to play some modern games...
First off, you left out several currently used Intel sockets, including the mobile one(s). The AMD side has been very simple: Budget (754), Mainstream and Workstation (939), Servers (940). That's it. It's not hard. The mainstream S939 offering spans everything from 1.8GHz super-cheap CPUs to dual-core opterons and monster (2.8GHz) FX62 CPUs. If you can't get an upgrade out of your S939 board you're not planning it right. Mobile? Shit, you can put your low-wattage Turion64 in (almost) any old 754 board! How's that for an upgrade?
Heck, I've got a S939 board which also supports AM2 via an add-on daughterboard. For me, that's a really pointless solution considering the overall price of switching CPU and memory, but hey, it's there if you want it. If this works out, the maximum CPU that I can upgrade to on this S939 board hasn't even been announced yet, maybe not even planned.
What socket will the new Intel Monroe stuff use by the way?
You mean like Intel's Slot 1 and AMD's Slot A designs from 5 years ago? Which put the processor on a card, which also contained high speed memory and was slotted into the motherboard?
Of course the reason for it then was because they couldn't get the enough L2 cache on the chip itself so they built an external cache at 1/2 CPU speed. As soon as they could fit the L2 cache they wanted onto the chip they moved away from the slot design because it's far more expensive to make.
Damn im after buying a 200euro socket 939 motherboard and 4400+ X2 processor for 500euro...
my wallet hates how fast the IT insdustry moves!
FB-DIMM is really great if you want to make an extremely stable server and don't care as much about latency or just happen to not already have a memory controller on die.
Oh, and yeah, Intel owns the patents so you would be paying your competitor and be beholden to them to chose your memory type.
And all the memory manufacturers would have to pony up to Intel and also be beholden to them.
I won't go into the technical details about how AMD's design is geared towards low latencies and suffers greatly when latencies get too high while Intel's (old Netburst) design was geared towards high speeds and could tolerate terrible latencies but suffered greatly when speeds choked it. We probably won't see FB-DIMM as Intel is shifting over to the Centrino family of CPU's that are closer to AMD's design idea.
[RIAA] says its concern is artists. That's true, in just the sense that a cattle rancher is concerned about its cattle.
I know the effect without knowing the meaning bitbrains "lol".
Not to be confused with Cyrix's ill-fated Pentium clone "M-II" (M2) from yesteryear.
Jeeze, what a poor naming choice from AMD's standpoint.
http://www.tomshardware.com/2006/02/21/a_look_at_a mds_socket_am2_platform/
Why is there such an editorial delay at slashdot all the time? I'm not new here, but Jeez.
What features were those? Difficulty in booting Windows XP? Surely if MS can make do booting using a BIOS then the great Apple can manage it. Didn't Apple use a plain old PC for their Intel developer program? Soon we'll be hearing that Apple "invented" EFI.
Hmmm and less heat dissippation is good how?
It's probably just me!
See link:
http://www.theinquirer.net/?article=15189
What happens is on each clock, the 1st DIMM transfers its data to the CPU. The 2nd DIMM (if there) transfers its data to the 1st DIMM, the 3rd DIMM (if there) transfers its data to the 2nd DIMM, etc. Thus each DIMM gets the data from the next DIMM, puts it in a buffer, thus regenerating the signal electrically and emits it upstream on the next clock. You could call this "daisy chained". This limits how far each DIMM has to drive its data, which is why FB-DIMMs can claim better electrical characteristics.
What this means is to get data from the 3rd DIMM, it takes 3 clocks just to get the first bit of data (presumably 16 bits) to the CPU. The "good" news is that since this system is (semi-) serial, the clock has to be very high already, so this latency is somewhat mitigated.
FB-DIMMs are point to point, but the point-to-point doesn't mean each DIMM connects to the memory controller, it simply means each DIMM is only on a bus with one other DIMM (well, one upstream bus and one downstream bus). Each DIMM forwards data along these busses in both directions. But again, there is no way for the 3rd DIMM to get to the memory controller without going through (and not just by) the 2nd and 1st DIMM first.
Your extension of my argument to PATA vs SATA just underscores your misunderstanding. My concern is with the latency of intermediate forwarding of data. SATA (well, the version in regular use) doesn't even allow you to attach multiple devices to a single bus, let alone have the devices forward the data to the head. Note that PATA allows multiple devices per bus, but it is a true bus, in that the data from the far device just goes by the near device, not into it and back out.
SATA is taking off due to connector costs and cable routing in the case. RAM doesn't face cable routing difficulties. It does face signal routing difficulties, but these only need to be solved once per motherboard design at worst, not once per installation as in cable routing. In addition, the signal routing complexity is much higher for ultra-high speed busses and thus the problem of signal routing will be solved the same way for FB-DIMMs as for DDR or DDR2, which is one company (Intel) will make a reference design and the other motherboard designers will just leave those signal lines alone and add other signals in the I/O area where they want to put on additional SATA RAID controllers. And in regards to connector costs, FB-DIMMs don't change the DIMM connector and thus don't reduce the cost of the DIMM connector. So I don't see a parallel here at all.
Finally, as to DIMMs and busses being forward compatible forever, it's just not going to happen. You'll have the same problem you did with SDRAM (or DDR or RDRAM). All SDRAM was compatible with each other, just the speeds changed. So you can use your old slow DIMMs as long as you don't mind that slowing down all your memory accesses.
Finally, the reason RDRAM failed isn't as simple as your comments that the RAM people screwed RAMBUS. The problem was the RAM people didn't feel like being screwed by RAMBUS. RAMBUS wanted license fees on all RAM made (see their grab at applying their patents to DDR) and so they tried to make RDRAM the standard. Intel also wanted more money per motherboard sold (not just happy selling the CPU). Intel's first attempt at making this happen was Slot 1, where they force-bundled the 2nd level cache memory in with the CPU (2nd level cache SRAM revenue could be $30-$50 per mobo back in the Socket 7 days). Note that 2nd level cache moved to the main CPU chip later. Intel additionally decided to license slot 1, claiming patents on it. Regular front side busses could not be patented, as they were purely functional, considered the most basic way to do something. Slot 1 was positioned so as to patent the physical connector and form factor so they could enforce their fees.
Intel decided to threaten VIA (a very popular Socket 7
http://lkml.org/lkml/2005/8/20/95
I doubt the RAM industry is going to move to a standard (DDR3) that is unworkable. Capacitance can be beat by stronger drivers. Maybe they'll just have to go to that. It does seem we'll end up going to buffered memory either way. Although I'd prefer it be just be registered (like ECC) not this FB-DIMM stuff.
http://lkml.org/lkml/2005/8/20/95
I did real all 3 pages. However I also saw on the page of the article I sent you that the picture showing how much better the signal routing on FB-DIMMs is a "dual-channel" setup, despite having 4 DIMMs.
1 812&page=2
I consider this a typical setup, and the article does too.
The real crux of this thing is you confused "point-to-point" with each DIMM necessarily connecting straight to the memory controller and stated such. That's not the case and I wanted to clear that up. For ultimate performance, you would do such a thing, but the positioning of FB-DIMMs seems to be for servers (and lotsa DIMMs) instead of for home machines at this time.
See pic here:
http://www.pcstats.com/articleview.cfm?articleid=
and other places. FB-DIMM is heralded as the way to attach more RAM more than to attach the same RAM faster. And I think given cost structure (it'll cost more), that's wise.
We'll see what happens with this. Like I said, I'm nervous about it, given the history behind the companies involved.
http://lkml.org/lkml/2005/8/20/95
There are a number of very blatant failings in this article. The biggest issue is this is obviously silicon that is not ready for primetime and is over 3 months from release. There is no reason to put up benchmark numbers yet...in fact I'd say it's surprising how well the silicon did, equal to released Venice processors almost.
The other very blatant thing is that it's stated that they are using DDR2-667 memory, however it is 100% obvious that the memory is running at DDR2-400 speeds. (There is no way the Sisoft memory benchmark should be equal to DDR1-400, it should fully reflect the EXTRA bandwidth from 667) As we've seen in P4 reviews, DDR2-400 is MUCH worse than DDR1-400. The latencies are horrible in comparison. This is a completely unfair benchmark based on incorrect memeory speeds. A fair comparison would be showing DDR2-667 running at the actual speed rather than the early silicon DDR2-400 speed. IGNORE ALL NUMBERS FROM THIS REVIEW AS THEY HAVE ABSOLUTELY NO VALUE.
I just bought an AsusA8n32 motherboard ((I know its over-hyped)) and it is compatible with the X2, which I was planning to get. Will my a8n32 be compatible with the new X2s coming out in June? And if I wait until june to buy my processor for my new rig, will I shoot myself in the foot over paying $200 for that motherboard? It's not too late for me to return my mobo if its not compatible, or if its going to drop dramatically in price. I would LOVE some suggestions and imput. Thanks.