Updated Roadmaps

This roadmaps are so good!!!

Bookmarked!! :-D

I'm from Brazil/PE/Recife

they seems more or less credible, although the AMD roadmap seems quite optimistic. This guy is also pessimsitic on AMD64 support for intel; not until Tejas (on which I agree) and not until Q205. I thought Tejas was pulled for release end of this year ? Or did I dream that ? Also too bad it doesnt show mobile chips...

Either way, if both roadmaps prove to be accurate, intel has a rough year ahead.. even more so in the x86 server market where it has nothing to gain, and everything to lose (like 90% marketshare)

= The views stated herein are my personal views, and not necessarily the views of my wife. =

Those were my thoughts exactly. I think so too.

  <font color=red><b>M</b></font color=red>ephistopheles

Thanks for the links. Lots of info.

Lots of info indeed, H-Boss, but beware... They could turn out to be flawed... Don't trust predictions that go too far...

  <font color=red><b>M</b></font color=red>ephistopheles

Nope just sit back and wait for the test results to flow in.

Very interesting

thought i might try and throw in the IBM/Apple one too...

so far all i find are articals and speculation...
<A HREF="http://www.theregister.co.uk/content/39/31594.html" target="_new">this one </A> is accuctually very interesting. Begs me to ask some Questions later.

<A HREF="http://www.macrumors.com/pages/2004/01/20040121202659.s..." target="_new">1.1 Ghz FSB...</A> not so bad and with what seems to be a fairly low power consumption...whats the power like on intels and AMDs like?





<b>on the verge of catastrophy (y1.999...k)</b>
ASUS P4S8X - P4 2.4B - 2 x 512M DDR333 - ATI 9500 Pro - WD 80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW

in regards to the first link i posted there.

the Future of CPUs seem as tho they will be coming to a dead end till we find a new alternitive to CPU creations.

What are the alternatives? realisticly

there will eventualy be a minimum size, 45nm seems as tho it might be it, 30nm at most. these limits i would asume will not go much faster then 25-50Ghz with out some for of serious Monitored cooling...

so once we reach this...2010ish AMD, Intel and IBM will be equals..of coarse there will be more components to the computer that will continue to see improvemnts but for the most part we will see these three companies as equals no?

the claim for fastest what ever will be a new battle?

128bit?

CPU sized MoBo's?

Gold Transistors.. or better, if there is?

HT/SMT with Multi CPU core's? i.e. Quad 50Ghz = 250-400Ghz?

It seems like there may be real battle for the best optizied code too...

<b>on the verge of catastrophy (y1.999...k)</b>
ASUS P4S8X - P4 2.4B - 2 x 512M DDR333 - ATI 9500 Pro - WD 80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW<P ID="edit"><FONT SIZE=-1><EM>Edited by cdpage on 02/02/04 02:05 AM.</EM></FONT></P>

IIRC, these kind of roadmaps always proved FALSE

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>

That kind of roadmap from the inquirer that dont add normal delay like 6 to 12 month.

I dont like french test

Hmmm.... true, these might turn out to be false...

These roadmaps have to be taken lightly...

  <font color=red><b>M</b></font color=red>ephistopheles

>the Future of CPUs seem as tho they will be coming to a
>dead end till we find a new alternitive to CPU creations.

This has been predicted for at least 20 years now, cpu developement and Moore's law will grint to a halt in 10 years. But 20 years later, the end is still not in sight (well, again 10 years from here  . Don't underestimate human genius.

>there will eventualy be a minimum size, 45nm seems as tho
>it might be it, 30nm at most

Why ?

>of coarse there will be more components to the computer
>that will continue to see improvemnts but for the most part
>we will see these three companies as equals no

I doubt it. Think diamond substrates instead of silicon, 3D etching, and God knows what they will come up with.

>128bit?

Nope, not by 2010. 64 bit is most likely still massive overkill by then, if not for our lifetimes. "640 Kb is enough for for everyone" might have been a tad optimistic, but I dare say now "309,485,009,821,345,000,000,000,000 terrabyte ought to be enough for all of us". Definately for the next 6 years  

>CPU sized MoBo's?

Yep.

>Gold Transistors.. or better, if there is?

Diamond.

>HT/SMT with Multi CPU core's? i.e. Quad 50Ghz = 250-400Ghz?

All of that, and much much more. Multicore is quite likely, cell-like technologies.

But to answer your real question; I don't think any of us will care at that point. you could say a 1 GHz computer is already more than fast enough for typical usage, excluding gaming and video encoding stuff. There is a part of the market that just won't be interested in 5 GHz cpu's. Similary, I expect at a certain point video encoding can be done almost in real time, which will make another part of the market uninterested. THen maybe cpu's will be good enough for gaming, shifting the bottleneck to gpu's, and mostly, developpers; At that point 95% of the current market won't be interested anymore in something faster anymore, leaving just scientific stuff, simulations and the sorts, and everything else will be commodity. You buy a computer like you buy a microwave. Is anyone really interested in a 5000W microwave ?

= The views stated herein are my personal views, and not necessarily the views of my wife. =

I like that one. I'm a humanist. :smile:
Well, eventually, the etching techniques will get to comparable atomic size - this would make it very hard to contain quantum fluctuations inside the cores. However, this will only be the case if they can't come up with another kind of methodology to building CPUs - and I doubt they won't. Therefore, while 90nm->65nm->35nm->10nm or whatever will eventually reach a limit, yes, there's still very, very much room to grow into in terms of processor sophistication. These limitations will be sidestepped and new fabrication methods will be introduced. It's just a question of time. I'm a humanist indeed.
:smile:

<bitching>
32-bit allows for 4,294,967,296 Bytes, or 4,096MB, or exactly 4GB adressing space.

64-bit allows for (hold your breath for this one) 18,446,744,073,709,551,616 Bytes, or 17,179,869,184GB, or 16,777,216TB. It will never come to that, however, because no more than 40 or 48 bits (can't remember exactly) are used for adressing. 40 bits has a limit of exactly 1TB, or 1,024GB. So therefore, with current technology, we're limited to one terabyte of memory... more or less. Or 256TB for 48-bit adressing.
</bitching>

Don't be angry, bbaeyens; I was just messing around.

I agree; the limits of 64-bit are very far from reached.

>64-bit allows for (hold your breath for this one) 18,446,
>744,073,709,551,616 Bytes, or 17,179,869,184GB, or 16,777,
>216TB.

Correct, the number i quoted was actually for 128 bits. Took them from an old excel chart, and copy pasted the wrong value. Still to put things in perspective, your number corresponds to a pile of 8000 Kilometer high of 1 GB Dimms, or 8 Km of terrabyte "dimms" that surely will hit the market in 10 years  

With a 128 bit cpu, you could address a pile of dimms of a few billion lightyears (no, not exagerrating)

>It will never come to that, however, because no more than
>40 or 48 bits (can't remember exactly) are used for
>adressing.

Depends on the chip, some (opteron I think) have a 40 bit physical limitation (ie no more than 1 TB of ram) and 48 bit virtual memory. However, these limits are artificial, as long as terrabyte dimms are not even on the horizon, there is no sense in making a chip with more than 40/48 address lines. If ever these "Dimmms" become available, its trivial to extend this to support higher capacity. Its an implemention consideration, nothing else.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

I wonder if we will ever get to this... 1 Terabyte Memory DIMMs...

Like, 4 populated sockets each with 1TB on a 128-bit quad-core processor running at 10Ghz... Doesn't technological advancement sometimes seem like magic?

I mean, I still remember thinking that a 200MB-HDD was immensely huge... Or having no HDD, for that matter... 4MB of RAM... Wow... And phosphor screens. Green and black only.

>I wonder if we will ever get to this... 1 Terabyte Memory
>DIMMs...

I am very sure we will. its not a bigger step than from 4 MB simms (I still have laying around) to 4 GB modules available today. I'm not sure if they will still be called DIMMs though  

Actually, [looking in my chrystal ball again], I think memory increase will far outpace clockspeed/performance increase over the next decade. Its just so much easier. Holographic and/or 3 dimensional memory structures should increase density by an order of magnitude with relative little effort, up to the point where you can basically have as much memory as you want. I dont even think it will take 10 years.

>Like, 4 populated sockets

Sockets ? Hu ? what are those ? surely you mean those 32 sugarcube sized computing modules you plug into each other ?

>each with 1TB on a 128-bit
>quad-core processor running at 10Ghz

Does it come with a nuclear power plant to power it ?  

>I mean, I still remember thinking that a 200MB-HDD was
>immensely huge... Or having no HDD, for that matter... 4MB
>of RAM... Wow... And phosphor screens. Green and black only

Yeah I know what you mean. Dreaming of a 360 Kb 8" floppy drive to replace my tape euh.. "streamer" =) on my TRS-80. Or when I did 3D rendering (3DS R1) under DOS on a 486-33 with 4 Megs, I read about those 120 MHz 486 cpu's with 32 Meg, and thought these things ought to render just about everything in real time  

= The views stated herein are my personal views, and not necessarily the views of my wife. =