AMD Demos "4 x 4"

[Specialist]

From PC WORLD:

http://blogs.pcworld.com/staffblog/archives/002475.html

Jerry Jones
http://www.jonesgroup.net

[ROF]

Never a dull moment when it comes to AMD. Where Intel integrates AMD innovates.

[waheed]

EDITED

[ViRaL1]

They were able to put two AMD dual-core chips together and outperform a single Intel dual-core chip? :O

WW, I'm impressed.

[ROF]

They were able to put two AMD dual-core chips together and outperform a single Intel dual-core chip? :O

WW, I'm impressed.

Where did you read it was an Intel Dual Core? I thought the article said "identically configured one dual core chip"? Identically configured would require it to be an AMD branded core not the inferior intel thingie. The impressive part is separation of memory channels for the Quad. Intel still thinks 4 cores can be fed through the same pipeline and achieve maximum efficency. A high school freshmen taking physics should explain to intels technology chief why this innovation from AMD is important to quad and deca-core chips.

[lordsmurf]

AMD and Intel remind me of the tortoise and the hare.

AMD is the hare, running like hell, screaming "look at me!" and doing all it can to show everybody how fast and great it is.

Intel is the tortoise. It does not run around, it is not an attention whore, it generally does not care about all that nonsense. It is sensible, does what it needs, and always comes out ahead in the end.

[ROF]

Intel is the tortoise. It does not run around, it is not an attention whore, it generally does not care about all that nonsense. It is sensible, does what it needs, and always comes out ahead in the end.

When was that? I have never seen Intel as a leader. They are generally a follower of what others innovate. I will agree on the attention whore. AMD usually has something new to talk about while Intel just keeps producing the same thing to the same specs. The last innovation spec they did (RDRAM) is well . . .

ADDED: But really is this a discussion of AMD Vs. Intel or is it about AMD and the innovation of adding memory pipelines while at the same time ramping up the quad core boards and also about AMDs support for the end user at home instead of targeting the business server market?

[lordsmurf]

Just because you've never seen it, does not mean it's not the way it is.

The whole experiment was AMD reacting to Intel. It's very much Intel vs AMD.

Beyond that, the CPU only does so much. Your major bottlenecks are still the OS, the RAM, the motherboard, and other aspects. Add as many CPUs and cores as you want, but you quickly reach a point of diminishing returns, and do little more than piss away money for something that you neither need or can nor feasibly use.

At this point, consider multiple cores (and/or multi CPUS) proof of concept, rather than something practical.

This is one of those areas where a person's level of common sense really shines. Those that lack common sense will blow wads of cash because it sounds powerful. Those with common sense will quickly realize multi cores/CPUs is pomp and circumstance, unless they dramatically change all the components found in current computers.

Software will always be the biggest issue.

[ROF]

Seen it . . . used it . . . can't recommend it(Intel). Your point being?

[ROF]

Beyond that, the CPU only does so much. Your major bottlenecks are still the OS, the RAM, the motherboard, and other aspects. Add as many CPUs and cores as you want, but you quickly reach a point of diminishing returns, and do little more than piss away money for something that you neither need or can feasbly use.

Re-read the article. Two of those bottlenecks is being smoothed out by AMD(The RAM and the MoBo) while Intel's chips to be introduced later this year do not innovate but just add to the bottleneck. Who's the leader?

[lordsmurf]

ROF, you have to smooth them all out, not just 1-2 of them. You still have a major battle with hard drives, OS, software, and other hardware aspects. If you only fix 1-2 of them, you may as well not do any of them.

In the past 5 years, the only advances made in home/office computers is about on par with a farmer using chicken wire and duct tape to "upgrade" something. At most, a new coat of paint on the barn. We've not seen anything extraordinary in a while now. Given what I just read, we won't for at least several more years either.

[ROF]

This is one of those areas where a person's level of common sense really shines. Those that lack common sense will blow wads of cash because it sounds powerful.

I can't agree more but there are some really dim witted people out there. I see them every so often in my computer building business. For me there is cost level I will not break. That is that my core will not cost more than 4 times the mainboard cost. If the Mainboard is a $60 board I will not spend more the $240 on the CPU. If you need a bigger chip, buy a bigger, better, more expensive board. Some people have to have the latest and the greatest. This is not a bad thing. It is those people who have "wads of cash to blow" that drive the market and the innovation ever onward. I am not one of them.

[ROF]

In the past 5 years, the only advances my in home/office computers is about on par with a farmer using chicken wire and duct tape to "upgrade" something. At most, a new coat of paint on the barn. We've not seen anything extraordinary in a while now. Given what I just read, we won't for at least several more years either.

Agree again but it is a step in the right direction. If all the worlds problems could be fixed all at once . . well . . . there would not be any humans left.

[Steve Stepoway]

ROF, you have to smooth them all out, not just 1-2 of them. You still have a major battle with hard drives, OS, software, and other hardware aspects. If you only fix 1-2 of them, you may as well not do any of them.

Yes and no. The history of high-performance computing has always been that one component gets improved, and the bottleneck moves someplace else. Then that gets improved, and the bottleneck moves again. Quantum leaps in technology are rare. Mostly it's incremental. If we didn't do anything until we can do everything, we'd still be writing binary on IBM 704s.

That said, yes, many people will throw away a lot of $$$ thinking it will make a huge difference, and it won't. A difference? Yes. How much? It depends.

Steve

[ViRaL1]

I'm just glad there are plenty of people out there foolhardy enough to pay for bleeding edge so that the technology becomes mainstream and less expensive more quickly. :P

Now let's get all that Wireless-N stuff down to 802.11g prices.

[ranchhand]

Well, I guess I have a soft spot in my heart for AMD, because way-y-y-y back in the dark ages Intel was the hardware version of Microsoft-no comptetition. Do we realize that an Intel Processor in the 80's could go for as high as $1500 for something around 100MHz?? AMD was a little fly buzzing around the giant's head that nobody took seriously. Then AMD suddenly came out of nowhere with some serious processing power and kicked Intel in it's pricepoint. As a result, prices dropped radically and we all are winners and that is why Homebuilding took off like a rocket and is now a major source of revenue for Intel and AMD. Next time we decide which processor we want to choose, remember that we have that choice because of good old American competition which came from AMD!

[MJA]

I don't care what Intel makes.never ganna buy any of their crap

[deadrats]

Beyond that, the CPU only does so much. Your major bottlenecks are still the OS, the RAM, the motherboard, and other aspects. Add as many CPUs and cores as you want, but you quickly reach a point of diminishing returns, and do little more than piss away money for something that you neither need or can nor feasibly use.

this is one of the most repeated lines and has been for years and it's as wrong today as it was when it was first uttered.

for tasks that are extremely cpu intensive, such as rendering 3d scenes or video, the limiting factor is still the cpu.

if what you say is true, then custom compiling Gentoo, Cinerrela and Blender and using a Ram Drive would greatly improve encoding/rendering performance.

i can tell you from personal experience that is not the case.

At this point, consider multiple cores (and/or multi CPUS) proof of concept, rather than something practical.

total and complete bullshit. try rendering any HD compliant video using a properly multi-threaded app (like Procoder, which can use up to 16 cpu's) and come tell me how impractical multiple cores are.

This is one of those areas where a person's level of common sense really shines. Those that lack common sense will blow wads of cash because it sounds powerful. Those with common sense will quickly realize multi cores/CPUs is pomp and circumstance, unless they dramatically change all the components found in current computers.

what "dramatic" changes would you like to see? even 15,000 rpm SCSI hdd's in a RAID 0 configuration on a workstation class motherboard (like one of those sweet Tyan's) don't improve performance as much as a good dual core cpu.

i will grant you, that if money were no object the smart way to go would be to buy a hardware encoder, but considering hardware encoders capable of rendering 1080p WMV HD in real time cost 35 thousand dollars and hardware encoders capable of rendering mpeg-1/2/4 cost about 10 grand, they really aren't cost effective for the average user.

this is what you don't seem to understand, the dual core and quad core offerings from intel and AMD aren't aimed at the high end professional market, they are aimed at guys like me who can afford to spend $500 on a cpu but not 35 grand on a hardware encoder.

Software will always be the biggest issue.

software and hardware are equally important. even the best coded software in the world is worthless if you don't have hardware powerful enough to fully exploit it and even the best hardware in the world is worthless if the software can't take advantage of it.

[lenti_75]

SWEET, I can't wait to buy one.....by 2010 when the price will be around 100.

[lordsmurf]

Beyond that, the CPU only does so much. Your major bottlenecks are still the OS, the RAM, the motherboard, and other aspects. Add as many CPUs and cores as you want, but you quickly reach a point of diminishing returns, and do little more than piss away money for something that you neither need or can nor feasibly use.

this is one of the most repeated lines and has been for years and it's as wrong today as it was when it was first uttered. for tasks that are extremely cpu intensive, such as rendering 3d scenes or video, the limiting factor is still the cpu.

Yes... but you can't see the forest through the trees.

CPU can be limiting, absolutely .... but as soon as you boost the CPU a good bit, you run into severe limitations in all the other areas I just mentioned. Adding extra cores and more CPUs only has so much value. Until EVERYTHING gets a power boost, your improvements are only nominal.

This remind me of those kids that spend $1000 on some car part to "get better gas mileage" but results in maybe a savings of 1-2 mpg at best. Improvement? Sure. Worthy of even being measured? Hell no.

And you encode video, you don't "render" video. Rendering is for graphics.

[roma_turok]

From PC WORLD:

http://blogs.pcworld.com/staffblog/archives/002475.html

Jerry Jones
http://www.jonesgroup.net

Iit just stupid article about 4 cores in 1 CPU without any AMD Demos "4 x 4" .

[Faustus]

AMD branded core not the inferior intel thingie.

Tell that to my CCE encode times.