AMD's New Venice Core Shows Overclocking Potential

Vigile writes "It looks like the new Venice core processors from AMD are going to offer more than just 90nm technology through the entire line up. According to this article on PC Perspective, it is going to offer a lot of headroom for future processors as the author was able to overclock their 2.0 GHz sample to 2.8 GHz! I think I hear an FX-61 calling my name!"

Comments

unlocking?

[thundercatslair]

Will it be easy to unlock these though, because if there is potentially to destory it I would not risk it.

Re:unlocking?

[bersl2]

Multipliers on AMD processors are unlocked in the downward direction.

Re:unlocking?

[eRacer1]

Multipliers on AMD processors are unlocked in the downward direction.

Athlon 64 processors are unlocked in the downward direction. Athlon 64 FX processors are unlocked in both directions.

water cooling

[essreenim]

..with water brought to you directly from the highly polluted canals of Venice. Sniff, ahhhhh, I love the smell of sewage in my PC..

nt

[Anonymous Coward]

I think I hear an FX-61 calling my name!

Sorry, actually, that's my Intel chip. Noisy bugger.

uh

[eobanb]

What real good does overclocking 2 to 2.8 really do? These cores keep getting faster and faster, yet the increase in number of floating-point operations per second achieved isn't really that spectacular. How about a more intelligent (parallel) architecture to begin with?

Re:uh

[Anonymous Coward]

Warning Independent Thought Detected.

The white van has been dispatched.

You will be taken to the Marketing 101 Re-education center.

Hmm.

[iostream_dot_h]

An 800MHz overclock on stock cooling is absolutely incredible... But it kind of makes me wonder why AMD doesn't make the default core speed on the proc higher.

Re:Hmm.

[eobanb]

Because then we'd complain about how we can't overclock it. It's not about technology anymore, it's about psychology.

Re:Hmm.

[Anonymous Coward]

This is partially a manufacturing issue.

Since all the chips in a given line use the same core, they all have the same speed paths, ie some signals take longer to get from A->B than others because of more logic, longer distance, etc. The difference comes in during manufacturing. These companies are good at making transistors, but they don't get them perfect every time. When a chip is designed, they look for a theoretical maximum/minimum speed. If a chip doesn't meet the minimum speed at production is is scrapped, this is relatively rare considering the complexity.

On the other end you have chips striving to make maximum speeds. If every chip off a die could be rated at the maximum speed, that would be quite a feat, but it doesn't work that way. After the chips are made, they perform speed tests on them and "bin" the chips.
Chips get placed in lower bins for one of two reasons.
(1)some of the transistors weren't quite up to par during testing/"binning" and ran a little slower and would become unstable in the higher speed ranges
(2)they have to drop a chip into a lower bin for market segments, ie this speed is popular and we're out of them... take the next speed up and drop them into this slot.

That's why sometimes overclocking works, and sometimes it doesn't. It's more likely to work on second gen chips, as they work out glitches in the manufacturing process and more chips are "artificially" lowered in clock speed. That's also why there's a risk in overclocking, if you have a chip that made it into the lower bins because of a manufacturing inconsistency, the chip will be unstable at higher speeds, generally only reasulting in calculation glitches, but possibly physical damage, depending on problem.

-Anonymous Computer Engineer

If its ANYTHING like their XP mobiles, Sign me up

[BiggestPOS]

I love the overclock I've got on my 2600+ XP-M running at 12.5 * 200 with nothing but a nice heatsink and fan.

The Barton core is awesome, and AMD is just refining their game here, working with the same basic silicon for the A64 and the XP. Intel's brains are divided up among way too many incompatible irrelevent architectures.

Just my 2 cents.

Re:Intel-Rating?

[ergo98]

So what does 2.8 Ghz in AMD mean in terms of Intel performance?

Duh...

2.8Ghz -> 9081 AMD Cybermarks -> 84.7 ISO 9011:2005 quartets -> 1.7E10 Intel TruePerfs.

I think that was fairly obvious.

Duh!

[bstadil]

Within the same architecture the clockspeed is almost directly linear with performance. IE 2.8 is 40% faster than 2.0

Or were you just trolling for Intel?

Re:Duh!

[maraist]

That is assuming you're compute-bound, instead of memory-bandwidth, harddrive-bandwidth, or some other kind of IO bound

Hard-disk bound is hardly ever a factor for system-upgrades. If you're HD bound, it's unmistakable, and you usually are doing something worth the money of upgrading the disk-system. 3D grahpics-card bottlenecks, on the other hand are real and subtle.

As for memory bound, I'm not aware of any benchmark (other than synthetic memory-testers) that didn't improve semi-linearly merely because of being memory bound. Increasing CPU speed these days generally means increasing the cache-speed which implies speeding up critical memory paths.

Non-von Neumann Memory Architecture

[MOBE2001]

How about a more intelligent (parallel) architecture to begin with?

Unless you have a way around the von Neumann bottleneck, what intelligent architecture are you thinking about? Adding multiple cores will eventually hit a wall because of memory bus contention. The only solution I see is for someone to create a memory architecture that permits unlimited simultaneous memory access. At that point, fast processors will not matter much. Just have a bunch of cheap processors share a single huge memory space.

Re:Non-von Neumann Memory Architecture

[hyc]

re: unlimited simultaneous memory access - it's called a crossbar switch, and a lot of parallel supercomputers use them. They are fairly expensive, in real $$ and in terms of board space, etc...

The HyperTransport that AMD uses is not a bad interconnect in the meantime, for people on smaller budgets...

Re:Non-von Neumann Memory Architecture

[maraist]

something revolutionary and cheap, maybe a new optical memory

Revolutionary and cheap.. You don't ask for much do you? Optical is coming slowly, but I'm not convinced it's ever going to replace electric current/voltage-based computing. At least not for general computing.. The problem is shrinking optical paths; you need a wave-guide for optical paths; for electric current, all you need is a string of closely spaced ionized atoms. Theoretically you could get down to a couple-atoms thick of wire with electric current.

Moreover, photons are only slightly faster than electric-current. Electrons move between 0.6 and 0.9 times the speed of light. What photons are really good at is traveling long distances without dispersing as heat. Electrons move only a couple atoms before bouncing into something. But you can do lots of really useful things with electrons that you can't do with photons... Having photons mimic the functionality of electrons might not be doable on the same scale (meaning by the time you get 30 million photonic transistors on a die, you could probalby get a billion electric transistors).

Quantum computing has the same density dilemma as photonic computing. But at least quantum computing does more than electric or photonic switching, so it doesn't need as many functional units. Don't expect to see an Intel Q4 any time soon.

As for a more practical architecture. If practical and economic are what you want then the Pentium 3's with a flat BUS multi-CPU architecture is where it's at. Lots of cheap cores on as simple an architecture as you can get.

The problem of course is in the mathmatical algorithms that we use to do real work. Most steps of computational algorithms are inherently dependent on the results of previous steps, and are thus not parallelizable. single-threaded CPU's have gotten VERY good at parallelizing individual instructions. The compilers aren't well suited for helping the CPU out, so things like the Itanium were supposed to exploit such parallelism. But the loss of backward compatability (and the Itanium's focus on floating point) spelled the death nell for that architecture.

IBM, Intel, AMD are all pushing multi-threaded execution.. Basically giving up on figuring out how to make a particular algorithm work. They're pretending that a CPU which works well on a high-end server with lots of independent jobs (web pages, database transactions, IO requests, etc) can be sold to a market which is trying to scroll the mouse wheel on an excel spreadsheet with a thousand rows. The spread-sheet navigation is extremely sequential. A dual core CPU will be noticeable since there are periodic background tasks which often "get in the way" of your foreground task. But a 3'rd/4rth CPU is not likely to be useful at all to non-workstation end-users. (My workstation generally has 8 visible applications, all actively running).

Personally, I think the answer is taking a step back from MHZ and pipelining. Go back to a 3, 4 or 5 stage pipeline with MASSIVE read-ahead decompilation of instructions (similar to transmeta). Get lots of high-speed cache on board with as little latency as possible (current large-cache architectures have HUGE latancies). By lowering the CPU MHZ, you reduce the latency to the all-important main-memory. Advance the state-of-the-art in power-consumption (I've read of several very novel approaches, including decreasing power to the point of statistically acceptible and correctable errors occuring in the computation). Perhaps put a second core on the CPU, but don't just put two identical masks.. Make use of the fact that a CPU has hot and cold regions.. Rewire both devices so they're really one big device with two functional CPUs..

Develop better heat-dessipation techniques.. THey've been very creative over the years.. Flipping the chip so the silicon directly presses against the heat-sink, for example. They've introduced lower-resistence copper as the main wire interconnect, which was a major material-science challenge. Newer exotic materials may provide for better heat conductivity and voltage regulation. The cooler you run a CPU, the higher the power it can dessipate, the more power you can shove into it, the more work you can ask it to do.

-Cheers

Re:Non-von Neumann Memory Architecture

[Anonymous Coward]

Electrons move between 0.6 and 0.9 times the speed of light.

That's a pretty fundamental error for someone acting like an expert to make, don't you think? At 0.9c, we don't call them "electrons," we call them "seriously badass beta rays."

It's not the electrons that propagate the signal, it's the potential difference the electrons are at. I have no idea what voltage you'd need to get electrons to be travelling at 0.9c, but I'd put it well into the MeV range.

Re:Non-von Neumann Memory Architecture

[lostchicken]

E=(gamma)mc^2
gamma = (1 - v^2/c^2) ^ (-1/2) =2.3
E = 2.3 * .511MeV/c^2 * c^2 = 1.17 MeV

Yes, that is, in fact, one bad ass beta particle.

only downwards

[doormat]

AMD chips have multipliers unlocked downwards. That means if its got a 10x or 12x multiplier, you can chose 8, 9, 10, up to the default number. It works well, even if you dont want to OC, you can turn down the multiplier and crank up the FSB for better performance.

Re:only downwards

[WNight]

Funny, but it misses the point. You can buy faster ram, and the bus is capable of those speeds - it would do them with a faster CPU installed, by default - so nothing is being run over spec.

Overclocking just like the northwoods...

[Man in Spandex]

This is so dejavu.

Now it's AMD's turn to pull an Even Steven on Intel with cool running cpus that also O/C high. That SOI sure does wonders ever since they started using it on the first A64's.

Most people don't run around overclocking their cpus but it is a great market to target (oh I'm da rappa!) because Intel has had great cores to O/C ever since the first Northwoods until the first Prescott, the bacon-cooker.

Re:What's the point of higher core clock if ...

[Anonymous Coward]

Parent doesn't really seem to know what he's talking about (perhaps he glanced at an architecture book once). The memory hierarchy of almost all modern processors ensures that only a very tiny portion of instructions generate real disk accesses. Relatively few apps are really effected by storage speed... look at some gaming/application benchmarks for a 10k rpm disk vs. a 7.2k rpm disk with the same buffer size.

Re:2.8GHz? I've got that now

[JoeShmoe950]

gigahertz are a fairly useless comparison between different chip types. A 2.0 ghz AMD64 might run circles around your 2.8ghz P4, while a 1.5Ghz Pentium-M could go faster than an AMD XP 1800 without worries. Architectures make this happen. If a 2.0ghz AMD64 can go the same speed as a 2.8ghz P4, obviously the 2.0ghz AMD64 is running more instructions per megahert. This means, that each one counts for more. Thus, a .8ghz increase is a huge increase in speed. Imagine running a 2.0ghz P4. Not very fun, eh? Now, the difference between a 2.0ghz P4 and a 2.8Ghz P4 is smaller than the difference between a 2.0Ghz AMD64, and a 2.8Ghz version of the same exact chip. That is a huge speed increase!

what up with the clock speed nowadays

[Enrique1218]

I remember when there was an actual megahertz race between amd and intel. Now it appears as though everyone is out breath. I can't believe we are still talking about 2.0 ghz AMD processors. Are they ever going to break 3 GHz? Intel seems to be no better off. How long was it since the first 3 Ghz was release and there is no 4 Ghz chips in sight? As a mac user, I can only revelled that physics has caught up with everyone and I no longer have to spout out about the megahertz myth in defence of my platform.

Re:what up with the clock speed nowadays

[mp3phish]

They are hitting the limits of the physical world with their current known solutions. Until there are more breakthroughs and improvements in chip fabrication, you won't see many 4GHz chips any time soon.

Just as an example, for intel to be able to get to 3.8GHz they had to decrease their chip performance significatnly. So now their IPC (instructions per clock) are lower on the 3.8GHz chips than previous P4 chips. Every time they bump up the GHz they have to extend the pipeline. This lowers IPC.

So you have this race between the physicists who are in charge of coming up with innovative ways to overcome physical limitations in chip fabrication, and you have the engineers redesigning their chip to work around these limitations. It is an uphill battle both ways and they have finally hit the ceiling where it is significantly detrimental to cost/performance at anything higher (for now).

Re:what up with the clock speed nowadays

[Perdo]

Physics has caught up with no one. Transistors are still getting smaller, but heat is on the rise, as any 2.5 Ghz water cooled G5 owner knows.

Think of it this way: work costs watts.

No matter if you do a given amount of work using a narrow speedracer architecture like P4 or PPC970, Or a wide architecture like G4, Athlon64/Opteron, Itanium or Pentium M, the work done costs watts, and generally the speedracer designs start paying more in work per watt.

The real current limitation is architecture complexity, where no one has a big enough brain to fit more than 150 million or so useful, non-cache transistors into their heads to debug the chip when there is a problem. Bob Colwell, former chief architect for Intel for the Pentium Pro/II/III/4, has spoken and written at length about this.

When he left Intel, there were perhaps 2 people left that could debug the Pentium 4.

Tejas was cancelled for this reason, as it was an even more complex version of the P4, certainly with AMD64 instructions included, possibly some EPIC (Itanium) compatibility, and a sort of SSE4 called at the time TNI or Tejas New Instructions, that were supposed to be the last straw in bringing complete vector processing to the x86 world, which Apple of course calls Altivec.

This complexity limit has caused architecture advancement to virtualy stagnate, while Moore's law marches on. 200 million transistors last year. 400 million in 2005 a billion in 2007. What to do with the transistors? Add more cores, since individual cores can not get any more complex and cache has a limited effect after 1mb, as Itanium and the G4 show. Cache is a poor substitute for a good memory bus, and after 2mb it's all crutches to keep poor architectures competitive with the better architecures out there.

Why the stagnation at 3 Ghz, or more specificly 3.06? Because that is all the northwood architecture could do, and Prescott, its replacement, was starting to hit that complexity limit and was delayed 8 months.

When Prescott arrived, it was hot, almost 175w per cm^2. This was not the process, 90nm, that caused the heat, because the Dothan (Pentium M centrino) was only 27 watts on the same process, and no one could figure out why it was so hot, so Intel got stuck, ramping clockspeed only 533 mhz in two an a half years, after doubling clockspeed to 3.06 from 1.5 in the previous 2 years.

AMD changed horses from 2.0 Ghz Athlon to 2.0 Ghz Athlon 64 and jumped 25% to 100% better perfromance, depending on the benchmark, mostly due to the integrated memory controller, not it's 64 bitness. It would take a 3.2 to 4 Ghz Athlon to match a 2.8 Ghz Athlon 64, and a 4.2 to 5.4 Pentium 4 to match it.

There is a performance race on, and a marketing bullshit race for clockspeed which may or may not mean a processor performs better..

Sounds like you have only been following the marketing bullshit race..

But then, you are an Apple owner.

Pretty simple..

[cbreaker]

There's plenty of explinations.

Here's some:

A) The chip is designed to run very cool. Overclocking it makes it hot, but it still runs fine. Just very hot.

B) The chip is designed to be run at higher speeds, and the initial offering is clocked-down. This gives AMD a few steps before more core/retooling work.

C) The cooler that comes with the CPU is very good.

Karaman

[Karaman]

I think of AMD64 more as a consumer, then a flame=seeker. Is it the most powerful - NO CLUE Is it stable - YES Is it cooler - YES Is it affordable - YES Is it for a PC - YES Why should I buy anything that is more advertised, but actually too expensive. I dont buy it. Others buy it. But not me! I like my AMD :) IRTFA and I am going to say it once: Overclocking capabilities does not mean just speed, they mean stability under extreme circumstances, therefore granted stability under normal circumstances!

Re:Overheating issues?

[YU Nicks NE Way]

The parent is currently moderated "Insightful" -- but it isn't. It's wrong.

P = I^2 R. For a processor, the current applied to each transistor is proportional to the clock frequency and the resistance is constant, so the power consumption per transistor (ceteris paribus) rises as the square of the clock rate. For modern processors, the power consumption of the chip is basically due to the total switching power of the transistors, and thus the power consumption rises roughly as the square of clock speed.

Re:Intel-Rating?

[jm92956n]

Link

Sytem Includes:

• 3ghz Intel Pentium 4
• 512mb RAM
• 80gb 7200 RPM HD
• CD-RW
• 19 inch Ultrasharp digital LCD

Total Price: $658, free shipping included. Add in an extra $200 for a PCI-Express video card and, at $858, it's comparatively inexpensive. It's not an excellent machine, I understand that; and, while I'm willing to pay a premium for a better machine, I don't expect the premium to be more than 10-20 percent more.

Re:2.8GHz? I've got that now

[ArbitraryConstant]

Intel and AMD chips have completely different designs. In general, Intel chips are designed to blast through simple code very quickly (as Intel thought that's all chips would be doing by now), and AMD chips are designed to be able to handle branches and conditional code better. Also, current AMD chips have a memory controller on the chip itself rather than on a helper chip on the motherboard, which makes their memory access faster.

Before Intel hit the gHz wall, the strategy was actually working out pretty well. They were at a bit of a disandvantage in some areas, but for the most part the clock speeds were so high it didn't matter.

With the new Prescott core in Intel chips, they increased the penalty for branching in anticipation of still higher clock speeds. Those speeds never came, so they're at a disadvantage now.

At more or less the same time, AMD upgraded the memory interface of their chips, which improves performance in most areas in addition to helping them catch up with media stuff. At the same time they kept and in some cases improved their performance on branchy code. They avoided the gHz wall by improving performance without pumping clock speed.

I think Intel assumed Itanium would take over in areas that needed branchy code back when they comitted to the Pentium 4 design in the 90s. It arrived very late, and it turns out regular desktop users still need to deal with branchy code.

Better review

[uodeltasig]

Here is a better overview of the changes and feature additions

Re:Intel-Rating?

[Suzuran]

Because, since the amount of work done by an instruction on one processor differs from the amount done by the same instruction on another processor, it was a rather _Meaningless _Indication of _Processor _Speed.

Not on an AMD machine

[Perdo]

With AMDs hypertransport and integrated northbridge, every processor you add adds another memory bus. It's call NUMA, for non uniform memory architecture, supported in Server 2003, XP Pro since sp2 and Linux since 2.4, perhaps earlier.

NUMA was first used by SGI with their late 90s MIPS machines.

Intel uses a shared bus, with the exactly the limitations you describe, except with their Itanium in 8 way+ configuration.

Re:So what you are saying is..

[obeythefist]

Not necessarily. A lot of CPU's fail testing at very high speeds but run with perfect stability at lower speeds. The CPU companies are profit driven, so they're happy to get some money for the CPU instead of throwing it.

Now, you can get yourself a cheaper CPU and overclock it, knowing it's probably capable of higher speeds, but there's a big risk of stability issues.

The current generations of CPU manufacturing process make very good error free batches compared to what it used to be like. So CPU's tend to work quite well at high speeds but still get badged down. That makes sense from a corporate perspective - if there is demand for a slower, cheaper CPU, you can sell into that market with higher specced CPU's. That just happens to be the way the market works.

The alternatives are untenable. It makes no sense for AMD to deliberately make a batch of CPU's specifically intended to be 2.0GHz when it costs the same as making a batch of 2.8GHz CPUs. AMD then has the *choice* of selling these CPUs at whatever speeds and prices the market demands.

Would the parent prefer than AMD make special 2GHz only CPU's to sell? Or perhaps AMD should instead only sell > $600 high end CPUs and not sell budget range CPUs at all?

This is the way the industry works. If you don't like it, feel free to go back to using transistors instead of IC's.

Re:why go further when we should make better

[t_allardyce]

What a load of BS, Try any recent kde or gnome based distro on anything slightly old and it you'll see its unusable, while linux itself runs fine i've yet to see a decent GUI, they're all obsessed with letting you configure your windows to look absolutely any way you want, when all anyone really wants is fast response time and a few reasonable config options. Boot windows 2000 (the only decent creation out of redmond) on the same PC and you'll see some what i mean. CPU speed and memory aren't so important if you've been running a simple word/web system for for the last few years, but when it comes to games, 3d-animation, video editing, image editing, audio composing/mixing, or even just compiling, then the difference is high. Also people often forget their hard-drive witch can be slow as shit.