12 replies to this topic

[maestro949]

We're getting to the point where we could build some serious graphical nano / molecular engineering tools with this stuff.

CPU: Core 2 Extreme QX6850 - Quad Core 3.0 GHz, 8MB L2 Cache Cost : $1040

Memory: Samsung announces 8GB DIMMs Cost : $2000 est.

RAID Storage: 2TB Lacie Biggest FW800 Rohs U Cost : $1,346

Video Card: PNY VCQFX5500-PCIE-PB Quadro FX5500 1GB 256-bit GDDR2 PCI Express x16 SLI Supported Video Card Cost: $2400

GPU Server: Tesla S870 GPU Server w/4 G80 cards @2.0TFlops Cost $12,000

Edited by maestro949, 27 September 2007 - 08:31 PM.

[maestro949]

Hell, why not throw in a 46" monitor for $4,000

Monitor: NEC Display Solutions LCD4620-BK-IT Black 46" 16ms DVI Widescreen LCD Monitor 500 cd/m2 1200:1

[electric buddha]

I just wish the infrastructure of the internet were keeping pace. I'd love to more viably use a cellphone and it's camera as a dumb terminal over to one of those beasts for real time video processing and display. I doubt it'd be viable with today's system, at least in the US/Canada.

[Cyberbrain]

Hell, why not throw in a 46" monitor for $4,000

Monitor: NEC Display Solutions LCD4620-BK-IT Black 46" 16ms DVI Widescreen LCD Monitor 500 cd/m2 1200:1

Nice! But if you wait a little longer, you could get Samsungs 40" OLED monitor

Wow, imagine a 1.0" flat 65" OLED ... [sfty]

[niner]

Yeah, technology marches on. It's amazing what you can get these days. With the exception of the CPU, most molecular modeling can be done on more modest hardware than that described above. I say exception of the CPU because we can always use more cycles. It would still be a fun package though. The Internet could stand some juicing up.

[Ghostrider]

Umm...it's all about the software.

Nice NEC monitor, but only 1920 x 1080? Many hi-def flat panel TV's will hit that resolution for under half the price. Plus, I think 42" is too big, I'll probably get a 24" within the next two years though. I want to get a 24" glossy-black screen monitor.

[maestro949]

Umm...it's all about the software.

Which is constrained by the hardware. When we get to a point where the serious nano-engineering is being done on a computer the progress of the software development will be closely tied to the available hardware capabilities. Parallel algorithms taking advantage of the multi-cores and GPUs might be capable of being the base for a workstation where self assembly and other types of molecular engineering can be done in a CAD/CAM like IDE.

Nice NEC monitor, but only 1920 x 1080?  Many hi-def flat panel TV's will hit that resolution for under half the price.  Plus, I think 42" is too big, I'll probably get a 24" within the next two years though. I want to get a 24" glossy-black screen monitor.

Yeah I'm not sure whether there's anything else that distinguishes this from an HD LCD. I'm not even certain a screen that big would be useful for engineering as the eyes can only take in so much. How many atoms in motion can a human watch at any given time?




image source: http://www.ks.uiuc.edu/Research/STMV

[Athanasios]

I have been behind the times. I had been telling people of the doubling of performance every 18 months. Then I realized that this was only true when I started telling people that. Now it is more like 12 months.

Faster cheaper computers will make people buy more computers and willing to replace them sooner. This is good news for microsoft as they sell many subs via replacement. I also think it will be the norm soon for everyone to have a laptop and desktop, including the kids. I dont see demand for the latest and greatest dying off soon.

P.S.: Nice credit for the image [tung]

[maestro949]

I have been behind the times. I had been telling people of the doubling of performance every 18 months.  Then I realized that this was only true when I started telling people that. Now it is more like 12 months.

I would argue that Moore's law really doesn't capture the true nature of advancing tech and is a bad example. The real thing to keep an eye on is the emergence from the intersection of numerous technologies and concepts. These are difficult to measure but I would suspect that they'll show up in larger demographic studies over the next few decades in regards to impact on medical advancements, social trends and disruptive technologies like renewable energies, electric cars, etc.

P.S.: Nice credit for the image  [tung]

[thumb] The computational attack on biological complexity will produce more medical advancements and rejuvenation therapies in the 2020s than everything we've discovered and will discover up to 2019.

[FunkOdyssey]

Moore himself actually predicts his law will no longer hold true (progress will slow down) in the near future.

Moore's Law, one of the most widely known laws in the computing world, may not have much longer to live. According to Gordon Moore, co-founder of Intel and best known for his theory predicting that transistor sizes would decrease by 50% every 18 to 24 months, fundamental physical limits will prevent engineers from further chip shrinkages within the next 15 years. He announced this prediction at the Intel Developer Forum.

"In another decade, decade and half or something, we will hit something that is fundamental," Moore said when asked if there would be an end to his 'law'. But he also pointed out that there always have been fundamental barriers that prevented chip technologies from further advancing. "There really are some fundamental limits. It's been amazing to me how the technologies have been able to keep pushing those out ahead of us. As long as I can remember, the fundamental limits are two, three generations out. So far we have been able to get around them."

[maestro949]

I suspect that moore's law will level off when we hit the atomic scale. IBM researchers are almost there already where some in the 80's and 90's predicted that it would take until 2050 to reach that level.

IBM Nanotech Breakthroughs Point to Tech's Future Building Blocks

But so what if going smaller plateaus for a while. We have access to lots of atoms. There's no reason we couldn't mass produce a few hundred billion microchips and engineer mega supercomputers the size of Rome.

[Cyberbrain]

I suspect that moore's law will level off when we hit the atomic scale.

Then comes the subatomic scale and quantum computing [sfty]

[maestro949]

This 16-core workhorse must be blazing fast. It can take 128GB of RAM.

Sun Fire X4450 2U Server - Quad 4 Core Xeons (16 Cores Total)

It might even run Vista.