The Common System Interface: Intel's Future Interconnect

[Wildside]

*Warning, heck of a long read, n very techie*

Real World Technologies - The Common System Interface: Intel's Future Interconnect

i just read the intro n conclusion paragraphs n im like already blown away n know exactly what Intel is trying to do.

the way i see it without reading this because it's too tech for me to understand............this is Intel's new arch of FSB to backfire against AMD's success with their HyperTransport for again compeition to stay in the lead of performance. This is like Intel FSB Godzilla vs AMD's Hypertransport Godzilla to see who wins.

if Intel says what i think their gonna do in 2008-9, then it looks like AMD is going to have THEE hardest time of their lives in benchmarking compared to Intel. Looks like doomsday is going to happen sooner then later.

feel free to quote from the report, because idk how lol. WE NEED Apokalipse to summarize this report plz!

*EDIT*

sry if i seem off track lol, its just hard to understand is all.

 

[Poizen22]

in my honest opinion, i know this is gona seem dumb, the cpu batle between amd and intel is gonna be the cause the end of the world lol. They are just gonna keep pushing each other technologically, and will eventually invent some super cpu that will be so powerfull it'l think on its own and it'l take over the world, in efforts to stop either amd or intel to develop a new cpu that will outade it lol.

 

[Apokalipse]

feel free to quote from the report, because idk how lol. WE NEED Apokalipse to summarize this report plz!

Summarise? okay.

They're finally adding an integrated memory controller, and getting a wider bus than FSB, with lower latencies.

It'll allow multi-CPU support

Then they're adding a few things on top, like power management (which does hurt latencies when used - a lot. but should provide significant reduction in power usage along the bus)

Basically, much of it is just doing the same thing HyperTransport already does. Although it's somewhat different (I think CSI is more decentralised - while HTT is more centralised on the CPU and memory)

 

[AnthraX]

if Intel says what i think their gonna do in 2008-9, then it looks like AMD is going to have THEE hardest time of their lives in benchmarking compared to Intel. Looks like doomsday is going to happen sooner then later.

If AMD didn't advance in technologies from now, then yes they would have a really hard time. But the fact is both companies have strong road maps ahead.

This is just another thing that Intel is doing to copy AMD. (Another thing was 64-bit extensions) As long as AMD remains a strong innovator, Intel will always have competition from them.

 

[Apokalipse]

if Intel says what i think their gonna do in 2008-9, then it looks like AMD is going to have THEE hardest time of their lives in benchmarking compared to Intel. Looks like doomsday is going to happen sooner then later.

I highly doubt that. Not after seeing this

If 30,000 points is true on a 3GHZ K10, then that says lots about the K10 architecture. a 5GHZ C2Q has so far gotten 27,000

AMD will be a strong player for a long time to come. And AMD is not stopping after K10.

Take a look at Bulldozer and Fusion.

 

[maroon1]

This is just another thing that Intel is doing to copy AMD. (Another thing was 64-bit extensions) As long as AMD remains a strong innovator, Intel will always have competition from them.

Intel Itanium processor were 64-bit before AMD even released Athlon64

If 30,000 points is true on a 3GHZ K10, then that says lots about the K10 architecture. a 5GHZ C2Q has so far gotten 27,000

AMD will be a strong player for a long time to come. And AMD is not stopping after K10.

First: 3dmark doesn't tell you which processor is better

Second: That news came from The inquirer. The source is questionable

 

[Apokalipse]

Intel Itanium processor were 64-bit before AMD even released Athlon64

but not X86-64. Itanium's were 64-bit only, and not x86 compatible.

They could not work with any Windows or Linux OS that was out, except for some specially designed one by Microsoft which was extremely buggy, had very big problems with incompatibility, and practically nobody used it - It could only be bought with prebuilt, proprietry systems.

When AMD made their Athlon 64's, their 64-bit instructions set was just an extension of the instruction sets that CPU's were already using. Plus, their 64-bit instruction set is based on X86, and is very different to the 64-bit instruction set used by Itanium's.

That allowed them to run perfectly fine with any current OS, and just allow people to implement x64 code and operating systems in their own time.

Intel was basically left with no option but to pay AMD to use their X86-64 instruction set (which they used in the form of EMT64)

First: 3dmark doesn't tell you which processor is better

Not absolutely, but it can give a good indication. Certainly nothing that can be ignored.

Second: That news came from The inquirer. The source is questionable

I remember people saying this kind of thing before Core 2's release.

 

[AnthraX]

^^^^ What he said!

 

[veg1992]

^ rofl...

idk where to side.... AMD or Intel... eh... screw fanboyism... i'lll just take the best bang for the buck one... hehehehe

 

[Wildside]

Summarise? okay.

They're finally adding an integrated memory controller, and getting a wider bus than FSB, with lower latencies.

It'll allow multi-CPU support

Then they're adding a few things on top, like power management (which does hurt latencies when used - a lot. but should provide significant reduction in power usage along the bus)

Basically, much of it is just doing the same thing HyperTransport already does. Although it's somewhat different (I think CSI is more decentralised - while HTT is more centralised on the CPU and memory)

ok thx Apokalipse. Sry for me not responding, been busy.