AMD - A glimpse into the future.....

[mysticsniper]

AMD Opteron CPU with Hypertransport 3.0, an extended AMD64 instruction set and FB-DIMM support is expected to be released in Q2/Q3 - 2007. This processor is expected to be a quad core processor with the cores connected through an enhanced version of the Hypertransport protocol. K9 is expected to interface to DDR3 memory with the initial revision running at clock speeds of up to 3Ghz.

AMD Athlon64/Sempron DDR-3 with Direct Connect 2.0 technology (the replacement for Hypertransport) are expected to be released in 2008.

Nice little source about what was forecasted and what actually happened - take a tour in memory lane if you want...

Roadmap

 

[ajgoodfellow]

Nice find - some interesting info on there!

 

[trojan698]

Nice, wonder where all that info came from. These companies publish their roadmaps?

 

[Jamie1984]

quad-me-bobs at 3 giga-doofers, impressive!

 

[FrankJH]

Am I missing something, why is this so impressive?

Dont get me wrong quad core will be novel, but to take 12-18 months to bring it out at about the speed processors are at now

DDR3 doesnt add that much to the party either (imo)

 

[locutus12]

your forgetting FrankJH, its not about the Mhz (or speed as you put it ), its about the effeciency of the instruction set, and this in my opinion is why AMD will always be better than Intel.

for example, an AMD CPU running at 2.6Ghz in 2 years time may seriously muller your cpu when clocked because the instruction set will be more effecient and more intelligent.

i.e. more instructions calculated per Mhz

 

[G.O.A.T]

You obviously didn't see the conroe preview.

 

[locutus12]

conroe is nothing more than intel waking up to the realisation that they actually do need to make an effecient CPU (as well as removing a pipeline thats the depth of the grand canyon) AMD have always lead with the "more per Mhz" and "higher clocks are not always needed" ideas.

 

[Etaqua]

conroe is nothing more than intel waking up to the realisation that they actually do need to make an effecient CPU (as well as removing a pipeline thats the depth of the grand canyon) AMD have always lead with the "more per Mhz" and "higher clocks are not always needed" ideas.

A 2.66GHz Conroe beat an overclocked FX60 @2.8GHz. I say beat, but I should say absolutely destroyed it. Before people say the test was biased, there was a completely neutral test in there, which the Conroe also beat the FX60 by the same amount.

 

[Zefan]

Am I missing something, why is this so impressive?

Dont get me wrong quad core will be novel, but to take 12-18 months to bring it out at about the speed processors are at now

DDR3 doesnt add that much to the party either (imo)

I'm running a 64 3000+ now, which is stock 1.8ghz. Three years ago we had 3200's at 2.2. It's nothing out of the ordinary to have the same clock speeds. Go ahead and tell me that the 3000 isn't that much of an improvement on the 3200, but I can guarantee you you'll be very very wrong.

DDR3 will make an enormous difference! Have you not seen the leap in graphics card performance with the ontake in this RAM? THe 6800 and x800 series have seen the biggest performance increase from one generation to another do to this RAM. It's ridiculously fast and produces very little heat.

 

[locutus12]

A 2.66GHz Conroe beat an overclocked FX60 @2.8GHz. I say beat, but I should say absolutely destroyed it. Before people say the test was biased, there was a completely neutral test in there, which the Conroe also beat the FX60 by the same amount.

LOL... so the Hex peeps did a review pitting a conroe chip that wont be out untill august, thats running on a new platform, and with DDR2, against an FX60 thats been out for 2 months and hasnt yet been paired up with the new AMD M2 socket or DDR2 and beat it... and your wondering why??? thats like pitting a 7800GS in AGP against an Nvidia 7900GT in PCI-E,

flawed test, try harder.

 

[FrankJH]

your forgetting FrankJH, its not about the Mhz (or speed as you put it ), its about the effeciency of the instruction set, and this in my opinion is why AMD will always be better than Intel.

for example, an AMD CPU running at 2.6Ghz in 2 years time may seriously muller your cpu when clocked because the instruction set will be more effecient and more intelligent.

i.e. more instructions calculated per Mhz

I am not argueing with that at all, but its only an EXTENDED instruction set, not an entirely new one. Yes the AMD IS has been impressive especially in gaming rigs since the developement of the 64 bit cpu, so it will be a little like when the original MMX extensions came out - brilliant for a few things but ( for example) no improvement in office files and general windows .

I know the MHz rating is misleading, and you your example was a good one, ( or like an AMD chip can muller a much faster {in Mhz} Intel chip in gaming etc but does relatively poorly in encoding) but we are still talking about 18 months time, with a shrink die also within this time frame and yet the amd chip is still only the same Mhz as is available right now.

I agree the cunroe test was floored in my opinion, but you also have to realise that in 6 months after the release of this chip etc the performance should be even better due to improved chipset drivers etc etc-- still a biased and misleading test in my opinion but good from AMD 's point of view ( as they know earlier than they would normally what they have to aim for performance wise - they are the ONLY people who can compare results as they should be able to quantify the missing %'s with DDR2 etc etc)

To keep up with Intel and Cunroe I really believe AMD have to come up with more thats all

 

[Agent WD40]

AMD Opteron CPU with Hypertransport 3.0, an extended AMD64 instruction set and FB-DIMM support is expected to be released in Q2/Q3 - 2007. This processor is expected to be a quad core processor with the cores connected through an enhanced version of the Hypertransport protocol. K9 is expected to interface to DDR3 memory with the initial revision running at clock speeds of up to 3Ghz.

AMD Athlon64/Sempron DDR-3 with Direct Connect 2.0 technology (the replacement for Hypertransport) are expected to be released in 2008.

Nice little source about what was forecasted and what actually happened - take a tour in memory lane if you want...

Roadmap

S-E-X-E-H. All I can say.

 

[queamin]

Wait until it out then you can see how they stack up against each other AMD willnot stand still,they got things in the pipeline aswell,how many items are released on time not many.
2 comanys fighting for your money is healthy like the video card section if there only 1 main company like in pc sound you hardly get any tech advancements.
Q

 

[locutus12]

the CPU market unfourtunatly is very very unbalanced, Intel control a little under 90% of it, and continue to hold that share by bribing the major pc manufacturers to only ship with intel...

which is another reason why i will only ever use AMD, because given the limited resources that AMD has compared to Intel, they have managed to keep up with and at times beat Intel.

 

[NathanE]

conroe is nothing more than intel waking up to the realisation that they actually do need to make an effecient CPU (as well as removing a pipeline thats the depth of the grand canyon) AMD have always lead with the "more per Mhz" and "higher clocks are not always needed" ideas.

"Always lead"? I think not. Clearly you've not been following this industry for any longer than the AMD64 has been around.

flawed test, try harder.

Again, people continually keep falling into the trap of assuming Intel was running full scale comparative benchmark test. They were not. They were using the FX-60 as an examplary baseline. They then demonstrated a Conroe chip running at a lesser frequency actually passing the FX-60's performance. The point of the test was to show the Conroe had a higher IPC and higher performance per watt.

the CPU market unfourtunatly is very very unbalanced, Intel control a little under 90% of it, and continue to hold that share by bribing the major pc manufacturers to only ship with intel...

which is another reason why i will only ever use AMD, because given the limited resources that AMD has compared to Intel, they have managed to keep up with and at times beat Intel.

This socalled "bribing" (your words) is actually more commonly referred to as "competitive business practice".

 

[stigggeh]

the CPU market unfourtunatly is very very unbalanced, Intel control a little under 90% of it, and continue to hold that share by bribing the major pc manufacturers to only ship with intel...

which is another reason why i will only ever use AMD, because given the limited resources that AMD has compared to Intel, they have managed to keep up with and at times beat Intel.

i dont think intel will miss your money....if your happy buying a slower processor just because you dont like intel then go for it.

 

[FoxEye]

"Always lead"? I think not. Clearly you've not been following this industry for any longer than the AMD64 has been around.

Again, people continually keep falling into the trap of assuming Intel was running full scale comparative benchmark test. They were not. They were using the FX-60 as an examplary baseline. They then demonstrated a Conroe chip running at a lesser frequency actually passing the FX-60's performance. The point of the test was to show the Conroe had a higher IPC and higher performance per watt.


This socalled "bribing" (your words) is actually more commonly referred to as "competitive business practice".

Performance per watt is one thing, but performance per £ is why a lot of us have been backing AMD all this time. Unless Intel radically rethink their pricing, I'll continue with AMD for the foreseeable future.