AMD now for better future investment?

[jshuck3]

To start let's pretend it's December and AMD delivers the 65nm parts that are rumored to be coming. Also Nvidia has their new G80 stuff out and Nvidia has the C55 (AMD equivalent) out.

Now....if you had to buy a chip that will be good now, but put you in a better position for the quad core emergence is AMD a better platform to go with?

I've been reading about Barcelona in Q2, Q4, and Q1/08 and AMD's native quad will probably end up beating Kentsfield (I'm guessing/hoping).

Anyone else concur that going AMD is actually a better long term decision?

 

[HopJokey]

Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

 

[tylerw13]

lol

great another INTEL is better.......NO AMD is better topic...dang guys just let time tell dont try to guess about stuff, and listen to ppl that just talk about what they personally like and use!!!

 

[swtethan]

so youre only guessing and hoping.... that makes it a whole lot better!!!

 

[Lonyo]

Now: Intel is better.
In the future: AMD might be better, or Intel might be better, it's the future, we can't know yet.
In the future beyond that: same as above.

The future is too far ahead to be able to predict. If we were looking at quad core from AMD in a couple of months, it might be clearer and more sensible to wait, but we're not.

 

[Viditor]

Originally posted by: HopJokey
Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

AMD is scheduled to release their 45nm in mid 2008, they will be all 65nm in 2007 except for 20% of their single cores which they are keeping at 90nm for contractual reasons.

Intel's CSI based products in 2008 will be Itanium only...Xeon CSI is scheduled for 2009 (and they haven't said if they will produce desktop CSI at all).

To answer the OPs question, on paper it looks like the AMD solution will be better...
That said we are talking about theory right now for AMD (and to a lesser degree for Intel as well), so the question is do you buy a bird in hand or 2 in the bush?

 

[dexvx]

If I can predict the future, I'd be in retirement right now.

 

[BrownTown]

Buying parts just to try to upgrade later is pretty risky when there is currently no word on whether current boards will even support AMDs quad core processors.

 

[Lonyo]

Originally posted by: Viditor

Originally posted by: HopJokey
Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

AMD is scheduled to release their 45nm in mid 2008, they will be all 65nm in 2007 except for 20% of their single cores which they are keeping at 90nm for contractual reasons.

Intel's CSI based products in 2008 will be Itanium only...Xeon CSI is scheduled for 2009 (and they haven't said if they will produce desktop CSI at all).

To answer the OPs question, on paper it looks like the AMD solution will be better...
That said we are talking about theory right now for AMD (and to a lesser degree for Intel as well), so the question is do you buy a bird in hand or 2 in the bush?

The day I believe roadmap time scales is never
Intel have also indicated they will deliver early with quad core, so why not with other things?

 

[HopJokey]

Originally posted by: Viditor

Originally posted by: HopJokey
Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

AMD is scheduled to release their 45nm in mid 2008, they will be all 65nm in 2007 except for 20% of their single cores which they are keeping at 90nm for contractual reasons.

Intel's CSI based products in 2008 will be Itanium only...Xeon CSI is scheduled for 2009 (and they haven't said if they will produce desktop CSI at all).

To answer the OPs question, on paper it looks like the AMD solution will be better...
That said we are talking about theory right now for AMD (and to a lesser degree for Intel as well), so the question is do you buy a bird in hand or 2 in the bush?

If you are saying that Tukwila (next-gen IPF) will come out ahead of a Nehalem based product I would doubt that based on IPF's track record (Tukwila is currently on schedule to release slightly ahead of the first Nehalem product, but with Itanium I wouldn't trust the release date).

As for Nehalem in 2008, Otenlini at IDF stated that Nehalem would be out in 2008 with Gesher in 2010 (of course a lot can happen between now and then). Here is a link, look at the very last picture on the page:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2840&p=3

As for the OPs question, I agree with you Viditor, the future products for both companies look good on paper right now, but it's all theory and until some working parts are available we cannot draw any conclusions.

 

[jshuck3]

So maybe I worded my question wrong because I wasn't really looking for the "intel better vs amd better" thing as much as I'm realizing my question made it sound. I was looking for more of what people are hearing about/reading about coming down the pipe in the next 12-18 months. Based on that trying to make some kind of intelligent decision. Yeah, it's all a crap shoot...i'm well aware of that. I was hoping for more information beyond what I've read and some of you have actually provided me with something to go off of. Sorry for starting another AMD/Intel whos better thread, but it's a tough position to buy new stuff right now with all the hype flying from both parties on their quad core stuff especially when their possible launch dates are so close. Just trying to make a better decision with whatever info is actually available.

 

[Madellga]

From what I understood, the Quad cores from AMD might not run on the current AM2 platform - they are possible meant for socket F.

So your upgrade path is broken from the beginning.

 

[Duvie]

Aren't Kentsfield (Intel quad core) supposed to work on 975x or 965x chipsets? I think guys over at XS have ES kentsfields running on current Intel motherboards....

I would lean towards in tel solution right now if upgradeability is key....

 

[dexvx]

Originally posted by: Duvie
Aren't Kentsfield (Intel quad core) supposed to work on 975x or 965x chipsets? I think guys over at XS have ES kentsfields running on current Intel motherboards....

I would lean towards in tel solution right now if upgradeability is key....

Kentsfield isnt the true native Quad-Core. Its the "glued" solution, IMO, mainly to fight AMD on the 4-socket front (8 cores total). Yorkfield will be the "native" quad-core solution. It will also have a "glued" 8-core solution for servers. No performance data has been given on Yorkfield, so its a blind guess as to how that will do. The "glued" approach isnt bad considering you can get double the number of cores in a relatively short period of time given most any core.

 

[Viditor]

Originally posted by: HopJokey

Originally posted by: Viditor

Originally posted by: HopJokey
Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

AMD is scheduled to release their 45nm in mid 2008, they will be all 65nm in 2007 except for 20% of their single cores which they are keeping at 90nm for contractual reasons.

Intel's CSI based products in 2008 will be Itanium only...Xeon CSI is scheduled for 2009 (and they haven't said if they will produce desktop CSI at all).

To answer the OPs question, on paper it looks like the AMD solution will be better...
That said we are talking about theory right now for AMD (and to a lesser degree for Intel as well), so the question is do you buy a bird in hand or 2 in the bush?

If you are saying that Tukwila (next-gen IPF) will come out ahead of a Nehalem based product I would doubt that based on IPF's track record (Tukwila is currently on schedule to release slightly ahead of the first Nehalem product, but with Itanium I wouldn't trust the release date).

As for Nehalem in 2008, Otenlini at IDF stated that Nehalem would be out in 2008 with Gesher in 2010 (of course a lot can happen between now and then). Here is a link, look at the very last picture on the page:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2840&p=3

As for the OPs question, I agree with you Viditor, the future products for both companies look good on paper right now, but it's all theory and until some working parts are available we cannot draw any conclusions.

I agree with you about Itanium's track record...though that said, CSI would be a very important component for the Itanium and early deployment could very well be part of Intel's strategy to boost sales. CSI is far more important in the server sector than it is in the rest of the sectors.
At the moment, we seem to have slightly conflicting reports, but my own guess is that we will see Nehelam either at the very end of 2008 or the very beginning of 2009...

As to making decisions based on theory (i.e. the OP's question), it really isn't a sound basis. I said the same thing when the specs for Conroe came out, and just because it's now AMD with the new technology doesn't make it any less true.
Being a "fanboy" is a very good way of reducing your analytical abilities...

I own both Intel and AMD shares, though much more AMD than Intel. The reason for this is that AMD has far more room to grow than Intel does, hence logic dictates that it has a larger capacity for share growth. Both companies have brilliant engineers...

 

[BrownTown]

Originally posted by: dexvx

Originally posted by: Duvie
Aren't Kentsfield (Intel quad core) supposed to work on 975x or 965x chipsets? I think guys over at XS have ES kentsfields running on current Intel motherboards....

I would lean towards in tel solution right now if upgradeability is key....

Kentsfield isnt the true native Quad-Core. Its the "glued" solution, IMO, mainly to fight AMD on the 4-socket front (8 cores total). Yorkfield will be the "native" quad-core solution. It will also have a "glued" 8-core solution for servers. No performance data has been given on Yorkfield, so its a blind guess as to how that will do. The "glued" approach isnt bad considering you can get double the number of cores in a relatively short period of time given most any core.

No, you have it backwards, 4x4 is designed to counter Kentsfield nto the other way around. They are both 2 dual cores, except Intels does nto require a special board.

 

[ahock]

I dont see anything wrong if your quad core is glued or native. What matters are all performance and performance per watt. If Intel can make true of their promise, 70% increase for Kentsfield vs current extreme and 50% for Clovertown vs current High end Xeon, who would not want these?

Lets put a big what i here. What if AMD native quad is released the same sched as with Kenstfield/Clovertown BUT only 10% increase vs their high end Opteron/FX, would you still go for native or glued?

 

[Furen]

Originally posted by: ahock
I dont see anything wrong if your quad core is glued or native. What matters are all performance and performance per watt. If Intel can make true of their promise, 70% increase for Kentsfield vs current extreme and 50% for Clovertown vs current High end Xeon, who would not want these?

Lets put a big what i here. What if AMD native quad is released the same sched as with Kenstfield/Clovertown BUT only 10% increase vs their high end Opteron/FX, would you still go for native or glued?

It made a difference with netburst because it was so FSB-bottlenecked but that doesn't seem to be the case with Core. That 70% number is BS, though, since most applications will not gain much of a benefit from it. Quad-core, in my opinion, is only for people who specifically need it, and it will priced as such.

 

[Regs]

All we can say is Intel is better right now. The future holds just another socket change, and possibly a higher rated PSU.

If K8L becomes an Intel core killer then I would only be amazed, but I'll sure won't be shocked.

 

[myocardia]

Originally posted by: Furen
It made a difference with netburst because it was so FSB-bottlenecked but that doesn't seem to be the case with Core. That 70% number is BS, though, since most applications will not gain much of a benefit from it. Quad-core, in my opinion, is only for people who specifically need it, and it will priced as such.

I agree with nearly everything you said, except for the 70% performance boost in some applications, specifically video editing & creation. Of course, those are really about the only apps that even need more than a dual-core atm.

I also expect the quad-core prices to remain quite high for quite awhile, since not many people seriously do that much video creation/editing. Oh, I almost forgot about the people who do CAD at home; they'll benefit about as much as the video creators/editors. Everyone else won't need quad-core for quite some time.

 

[Furen]

Originally posted by: myocardia

Originally posted by: Furen
It made a difference with netburst because it was so FSB-bottlenecked but that doesn't seem to be the case with Core. That 70% number is BS, though, since most applications will not gain much of a benefit from it. Quad-core, in my opinion, is only for people who specifically need it, and it will priced as such.

I agree with nearly everything you said, except for the 70% performance boost in some applications, specifically video editing & creation. Of course, those are really about the only apps that even need more than a dual-core atm.

I also expect the quad-core prices to remain quite high for quite awhile, since not many people seriously do that much video creation/editing. Oh, I almost forgot about the people who do CAD at home; they'll benefit about as much as the video creators/editors. Everyone else won't need quad-core for quite some time.

Yeah, that and the fact that single-core CPUs are quite profitable. If quad-core becomes mainstream then dual-core will become lower-end, which would make single-core CPUs kind of obsolete. Instead both Intel and AMD will keep throwing single-cores at the low-end and keep dual-core in the mid-range.

 

[Viditor]

Originally posted by: Lonyo

Originally posted by: Viditor

Originally posted by: HopJokey
Don't know about AMD's roadmap after Q1 2008, but Intel has 45nm Penyrn core based products coming out around Q1 '08 and Nehalem based products coming out late 2008 (CSI and IMC).

What do you define long term?

AMD is scheduled to release their 45nm in mid 2008, they will be all 65nm in 2007 except for 20% of their single cores which they are keeping at 90nm for contractual reasons.

Intel's CSI based products in 2008 will be Itanium only...Xeon CSI is scheduled for 2009 (and they haven't said if they will produce desktop CSI at all).

To answer the OPs question, on paper it looks like the AMD solution will be better...
That said we are talking about theory right now for AMD (and to a lesser degree for Intel as well), so the question is do you buy a bird in hand or 2 in the bush?

The day I believe roadmap time scales is never
Intel have also indicated they will deliver early with quad core, so why not with other things?

Intel is less than 3 months early with quad core...and this is merely 2 proven CPUs glued together. I don't get how that translates to early on a project as complex as CSI...

As to the roadmaps, I guess we can all believe what we will. I just can't think of any rational reason to dispute them at this point...

 

[Viditor]

Originally posted by: Madellga
From what I understood, the Quad cores from AMD might not run on the current AM2 platform - they are possible meant for socket F.

So your upgrade path is broken from the beginning.

No, the AMD quad cores have already been guaranteed to run on the current AM2 platforms. What you might have seen is that the next-gen AM3 chips (which will also run on the AM2 platforms) won't take DDR3...

 

[ZOXXO]

Originally posted by: Viditor

Originally posted by: Madellga
From what I understood, the Quad cores from AMD might not run on the current AM2 platform - they are possible meant for socket F.

So your upgrade path is broken from the beginning.

No, the AMD quad cores have already been guaranteed to run on the current AM2 platforms. What you might have seen is that the next-gen AM3 chips (which will also run on the AM2 platforms) won't take DDR3...

Link to the "guarantee"?

 

[Viditor]

Originally posted by: ZOXXO

Originally posted by: Viditor

Originally posted by: Madellga
From what I understood, the Quad cores from AMD might not run on the current AM2 platform - they are possible meant for socket F.

So your upgrade path is broken from the beginning.

No, the AMD quad cores have already been guaranteed to run on the current AM2 platforms. What you might have seen is that the next-gen AM3 chips (which will also run on the AM2 platforms) won't take DDR3...

Link to the "guarantee"?

Link

And please let me know if you need help with your Google trouble...