Sign in with
Sign up | Sign in
Your question

Intel throws good money after a bad chip

Last response: in CPUs
March 31, 2006 6:25:44 PM

Posted already...
March 31, 2006 6:29:37 PM


I'm never the first to post. :cry: 

Do you have a link?
Related resources
March 31, 2006 7:30:29 PM

So can you name any better processors?
March 31, 2006 7:32:01 PM

So can you name any better processors?

AMD Athlon XP
AMD Athlon MP <= Greatest CPU Ever! :D 
AMD Athlon 64
AMD Opteron 64
Intel Xeon (yea..I know)
Pentium 4
Pentium 3
AMD Sempron
Intel Celeron

Want me to keep going?

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:32:41 PM

Knowledge ownz you
March 31, 2006 7:33:19 PM

Knowledge ownz you


~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:38:51 PM

Knowledge ownz you


~~Mad Mod Mike, pimpin' the world 1 rig at a time

Ignore him. He is as anoying or perhaps more annoying than DVDpiddy. I have never once seen Ycon make a post that is more than 1 negative statement long.
March 31, 2006 7:39:46 PM

No truer werdz... :roll:
March 31, 2006 11:13:45 PM

In other news CNN post article when lacking any technical understanding of Itanium, IA-64, and what it brings to the table at 45nm/32nm.

They needed to start on it during 180/130nm (1/8th the transistor count of todays CPUs, but IA-64 is still 130nm today) just to get it ready for 2007 - 2012 period.

Intel - Pentium Pro - Pentium 4 - using medium, then moved to longer, pipeline length, with 2-issue cores, some tweaks to get near 3 issue performance.

AMD K7 - See K8

AMD K8 (Athlon 64 / Opteron) is using medium length pipelines, using 3-issue cores

Intel 'Core Duo' (Mk II) (Conroe, Merom, Woodcrest, etc) are using even shorter length pipelines and 4-issue cores, with optimizations.

Intel IA-64 (Itanium series) is using short 11-issue cores (only sustains 6 though, but it is still 'work in progress'), :trophy: , optimiziations, significantly less transistors per core, (=more cores per chip), and a totally new IA-64 architecture. It just needs 32 MB of cache to start scaling better than IA-32 (x86_64) does today. Which, with 45nm and 32nm is where it'll hit a starting point and more people will realise what ****wits they've been. Comparing a early 180/130nm processor (Itanium) to current 65nm ones with 4-8 times the transistor count at equal cost. :p 

Intel claim 100's of core per chip within 10 years.

AMD 'acquires' Itanium / IA-64 engineers, Intel may take them back within 5 years.

You see where I am going with this right ?

If anyone wants to know why IA-64 will be a nuke'n'pave for the IT industry by 2012, fire me a PM or é-mail, or MSN me.

The only thing I fail to understand, is why 99% of people fail to understand IA-64. (Of course most people are sheep - That's in the bible :p  - boom tsché).

A single 4x IA-64 processor on 65nm is possible, and clocked at only 2 GHz it would have greater MIPS performance, and provide better response, than my 2 x dual-core (quad-core) machine.
March 31, 2006 11:25:06 PM

If anyone wants to know why IA-64 will be a nuke'n'pave for the IT industry by 2012, fire me a PM or é-mail, or MSN me.

Speak on the open forum, lets discuss about that
March 31, 2006 11:44:24 PM

Read the bits I just made bold, in chronological order.

Understand that

180nm^2 = 32,400
130nm^2 = 16,900 <--- Itanium / IA-64 here
90nm^2.. = 8,100
65nm^2.. = 4,225 <--- Normal CPUs here
45nm^2.. = 2,025
32nm^2.. = 1,024

Note: Each level also increases clock speeds, and decreases power consumption, in addition to permiting twice the transistors in a given 2D 'layered' processor design. (sometimes more as they add extra layers every so often, since the design is really a 3D object, but treated as layers of 2D circuits).

If Itanium / IA-64 was made on the same size size, they could pack in 4x the number of cores, 4x the cache (move it to smart shared cache too, stops cache coherency within CPU 'issues'). Maybe add a dedicated IA-64 FPU core and only have 3 x standard IA-64 cores, etc

Read the articles on Itanium / IA-64, the technical data all adds up, once it gets cache it scales.

Look at the jump from Itanium (1) at 800 MHz to Itanium 2 at 1600 MHz [hint: It was more than double performance]. Also bear in mind these are 180 and 130 nm parts, and thus clocked lower aswell when compared to 90nm and 65nm parts.

Read this article from end to end, with a 15 min break after the 1st read to think about it, then re-read:

Look at the SiSoftware comparisons (I'll get some within 24 hours if people don't start looking), IA-64 1st gen, 2nd gen, skip 3rd gen and go straight to 65nm or 45nm production + 3rd gen feature set.

It would kick systems on par with my own, while costing like 1/3rd as much to market.

The media has the wrong ****ing idea, you don't compare 130nm to 65nm unless you're a retard. If they want to compare a 130nm CPU to another 130nm CPU they should be comparing IA-64 to much older Xeons and the early Opterons, and Athlon MP. Itanium is just a demo of 'what is' possible at 180nm/130nm so people can just multiply that by 4x/6x/8x to see where IA-64 'could' take them, at the same power level, etc were they not fools and just embraced the technology.

AMD have caught on to IA-64, Intel made it (this time), and HP is tied to both companies..... It doesn't take a genius to figure it out, but the media simply are clueless when it comes to computers.

eg: "yeah, it has a 160 GHz Hard Drive, or 2 GHz of RAM, or a 2 Gigabyte processor, etc" - I rest my case on the media covering IT. :p 

If they don't know how to square numbers, then they should go back to grade-school.

I am thinking of using CGI to present this information, in a 3D animation with 'flows' indicated, etc, as PowerPoint just isn't going to cut it (Just got another AU$1,000 in PC 'related' books to help me do just that, many are 'How to present ideas to management' related though :p )