AMD Announces 65-nm Chips, Touts Power Savings

Several readers wrote in about AMD's entry into the 65-nm manufacturing generation. The company introduced four chips to be manufactured with 65-nm process in the first quarter of 2007 to replace existing 90-nm chips in their lineup. AMD is playing up the power economy of its line, claiming that even its existing 90-nm parts consume less than 50% the power of Intel's Core 2 Duo, averaged over a typical day's usage, while the new 65-nm chips will be even stingier with power. Next stop, 45-nm. The article says that AMD has a goal of catching up within 18 months to Intel's lead on the way to 45-nm technology.

this is great news from AMD

[matt328]

It will probably drive down the costs of the Core 2 so many of us can justify buying one.

18 months is, like, a generation

[Junior J. Junior III]

So if AMD doesn't catch up with Intel by that timeframe, they'd be pretty much screwed, wouldn't they?

Hopefully we can see some Socket AM2 65nm stuff in the retail channel soon.

Nice chip but... will we get to see the benefit?

[theshowmecanuck]

"With the Vista rollout, it's more and more important to multitask and multicore without a super loud box -- that's the end goal," Huynh said.

Is anyone as tired of software companies eating up the gain in hardware performance as me? And for what? How about someone writing better software, not just new software. I got sick of buying new hardware just to open the same document because the O/S or new Office suite was bloated/full of shit/required way more horsepower just to do the same task. No Vista for me. I'll stay will XP and Linux on my (older) machines. And if MS forces people to go Vista, I'll go Linux or BSD. If I get new hardware, it will be to make these systems faster, not make new software, doing the same job, run the same speed.

Re:Depends on the Architecture

[Tumbleweed]

Quite frankly, on a level playing field I think we're going to find out that Intel's Cores are not quite as good as a lot of people have been raving. AMD's architectures are just a whole lot better.

You know, that's what a lot of people said before the Core Duo came out, and they were proven quite wrong. I was as surprised as anyone, but I learned the lesson: wait for the actual chips to be tested.

Once Intel puts the memory controller on-die like AMD has, it's going to *really* hurt AMD. HyperTransport doesn't seem to have any advantage at all on the desktop, so AMD's only real tech advantage right now is that on-die mem controller. Perhaps once we all have 8+ core chips on our desktops, you might see some HT advantages, but I believe I read somewhere that Intel has plans for on-die memory controllers and an answer to HT in the wings for 2008, though obviously that's just rumour at this point.

Re:Depends on the Architecture

[fitten]

but I'd expect Core Duo's apparent lead to be pretty short lived. There's nothing inherently brilliant in their architecture, other than them stealing a march on AMD in terms of 65 nm and adding cores like there's no tomorrow

Obviously you haven't read anything about Core2. Core2 is, on average, 20% faster at the same clockspeed as AMD's Athlon64. If you liked the fact that AMD's Athlon64 was comparable to the Pentium4 at 2/3 the clock speed, then you simply must like Core2's being faster at ~80% of the clock speed of the Athlon64... while still being on the obsolete, much maligned FSB architecture that AMD spends so much time and money poo-pooing over their obviously-so-much-better IMC+HT technology.

AMD still leads Intel by a country mile on budget processors as well.

Well... when you can't lead in performance, you try to lead somewhere else. Yes, the launch of Core2 parts drove AMD to cut the prices of their processors by 50% or more in order to stay competitive. Had they not done that, they would be selling nothing right now because even a fanboi couldn't justify buying AMD at the complete destruction that equal priced CPUs from AMD would get compared to the Intel parts. So, AMD dropped back to attempt to remain king of the bargain market until they could release something that would put them back into the performance game, which Intel currently owns (not counting the obviously boutique, one-off 4x4 deadend attempt to save face that AMD marketing released).

I have three Athlon XP and four Athlon64 machines at home (and no Intel other than two laptops). This migration of socket requiring a new CPU and all has bitten me already. My S939 parts barely lived a year (two years given the entire lifetime) and already AMD is requiring me to buy a new motherboard *and* CPU if I wish to upgrade because they're already killing off S939. Many of the already existing S775 boards for Intel will upgrade the CPU to Core2 (perhaps with a BIOS flash). It makes me kind of wonder how long Socket AM2 is going to last given that they're already talking about S1207 and some future move to DDR3 (yet another socket change?) Currently, for me to upgrade my AMD machines, I have to buy a CPU + motherboard because even the S939 X2s are EOL'd. I've thought about buying some upgrade AMD CPUs but I'm not going to do it. Core2 offers too much performance increase with the promise of socket compatible quad core very (very) soon. I believe AMD is requiring S1207 for quad core, so if you bought AM2, you're already EOL'd because of the change for DDR3 and quad core. AM2 was dead before it was released, it seems.

Re:Idles at 3.8W?

[rrhal]

Intel seems to get a bit of a pass on the whole memory controller thing. When you see people compare power usage you rarely see the memory controller on the Northbridge added onto the Intel ledger.

Bottom line is we get very fast efficient chips for cheap.

Re:Technology, progress.

[eudean]

When it stops being profitable to continue scaling transistors. The reason we scale is because it makes chips cheaper, faster, just plain better, meaning Intel and AMD (and others) can sell it for lots of money. If scaling no longer achieves that goal, you'll see it end in a heartbeat (just like you saw clock speed ramping come to a halt once it wasn't selling like it used to).

Re:Technology, progress.

[Bender_]

I think while being smartassy and all you completely forgot that the atomic radius is not equivalent to lattice spacing in crystals.
Your numbers are way off.

Re:Idles at 3.8W?

[Lonewolf666]

I remember a test in the German C't magazine where complete computers were tested. Everything being equal except mainboard and CPU. The CPUs were AMD Athlon 64 vs. Intel Core2Duo

Under load, the Core2Duo machines used a bit less power. Idling, the AMDs were better. The overall differences were pretty small compared to the total power consumption, so I'd disregard them for a typical desktop that does NOT run 24/7.

And BTW, avoid the old Pentium 4/Pentium D. Those are really inferior.

They went to CPU ratings for the opposite reason

[brokeninside]

With the Athlon series, AMD went to CPU ratings because, clock for clock, an AMD part was significnatly faster than an Intel part at the same clock speed. They had a hard time getting people to reject the ``megahertz myth'' that a faster clock was always a fast chip. And, in fact, this very myth came back to bite Intel in the ass and is part of the reason (but certainly not the only reason) that Merced died a much deserved death.

But other than that, I pretty much agree. As has been said for many eyars, brand loyalty is a socially acceptable version of battered spouse syndrome. Products should always be objectively evaluated for suitability to a task. Sometimes one chip is better for a certain task. Sometimes another chip is better. Sometimes the CPU doesn't really matter at all.

Not a contradiction

[Lonewolf666]

I take it those 2.8GHz Xeons were based on the old Netburst Architecture (P4).
AMD did better with the Opteron, but the new Xeon 5100 are Conroe-based.
Conroe vs. Netburst = massive improvement ;-)

THE FUTURE IS COMPLETELY SOLID STATE

[Anonymous Coward]

I'd rather see less dense microprocessors. The problem is this big chip manufactuerers make these super dense mammoth race car processors, it concentrates heat in one place, requiring a CPU fan, which is a mechanical component that can and will eventually fail.

I would rather have a motherboard with a CPU design distributed, where the surface area is spread out in such a way, that is completely solid state without any moving parts to fail, namely, a CPU does no require active cooling.

The future lies in completly 100% solid state devices. What does this philosophy mean?

No CPU fans, no power supply fans, no harddrives (flash memory instead), no noise, no moving parts whatsoever. Distributed or minimized waste heat.

Reliability and shock resistance skyrocket. You can seal them inside waterproof and dust proof and fire proof boxes.

I've implemented these kind of computers, at critical network points, and literally you can plug them in FOR YEARS and forget about them.

They are absolutely silent. And absolutely low powered. And totally reliable. All of which, is a very beautiful thing from an engineering standpoint.

These guys need to hire me on their team, because the definition of what is going to be expected in computers is going to radically change in the near future. Computers are going to go under, they are going to become *embedded* in everything with no expectaions of mainenance to them at all and installed in harsh environments. They either work or they fail after decades of use (or preferably, never), and then you replace them.

Reliability is CORE, and to achieve it, computers are going to have to abandon all cruches and become purely solid state devices entombed in indestructible plastic modules.

How often have you ever reformated the drive on your wrist watch? Had to reinstall an OS on your cellphone? Had to replace the CPU fan on your calculator?

I often wondered, why in heck weren't motherboards encased in protective plastic casing, and the same for ISA / PCI / AGP / PCI-x cards. Putting a comptuer together today has become like putting together lego blocks. And each component should be as equally durable and interchangable based on standards as a lego block.

Einstein
http://anarchy-tv.com/ [anarchy-tv.com]

AMD resorts to marketing hype

[Sebastopol]

I love when AMD has to invent specific test conditions to win via marketing.

"Testing all day."

"Typical usage model."

*yawn*

Get back to us when you have a better design, and not a better marketing department.

Re:Depends on the Architecture

[AcidPenguin9873]

I'm not going to be stupid and claim that all of Intel's performance gains in C2D come from this, but consider:

1. Intel puts twice as much L2 cache on their dies as does AMD for their high-end parts. The high-end Intel part has 4MB, the high-end AMD part as 1MB per core for a total of 2M.
2. Intel's L2 cache architecture is "shared", meaning that when you're running a single-thread benchmark, the single core on which the benchmark is running basically has access to the entire 4M L2. This is significant because it now means the single core running on an Intel part has 4M of L2, whereas on AMD it has 1M. This is quite a nice feature for Intel's single-thread performance.
3. In 32-bit x86 mode, there are 8 general purpose registers (GPRs). That means spill/fill code (which certainly hits in L2 and almost certainly hits in L1) is a lot more common. In other words, memory ops are more common. Intel, with the C2D, introduced a more aggressive out-of-order memory architecture, basically allowing any memory op, even with an unresolved address, to execute out of order, fixing it up later if there was a problem. This really, really helps with memory ops, especially the common spill/fill ones going to the (cached) stack. In 64-bit AMD64/EM64T code, there are more GPRs for the compiler to play with, so you have less spill/fill code and fewer memory ops, which mitigates C2D's advantage here. That's one of my theories on why 64-bit performance of AMD vs. Intel chips is closer (the other, which isn't a theory, is Intel's lack of a 64-bit-capable IOMMU, causing the OS to use bounce buffers for DMA to high-addressed main memory).

What I'm basically saying here is, C2D's larger cache and more-aggressive load/store architecture are really helping it for certain apps. My guess is that the libtomcrypt benchmarks are run in 64-bit mode (mitigating Intel advantage point #3 above) and either a) fit entirely in the L1 or L2 of both processors, nullifying any cache advantage C2D has, or b) fit in *neither* of the L2 caches, and is significantly larger than 4MB, which lets AMD's faster on-die memory controller make up for its lack of L2.