Does AMD have anything special on the horizon?

[Dimkaumd]

I recently bought shares of AMD because I thought the stock dropped significantly and had room for a rise. However, since then I am down about 5% and upon further review I am noticing that AMD doesnt have anything special on the market right now to compete with Intel. Intels Core 2 Duos and the new Quards seem to be taking market share away from AMD again.

Anyone think that AMD can regain its strength through upcoming products? Will 2007 be dominated by Intel chips or AMD?

 

[Duvie]

Even if AMD pulls off a decent launch with its K8L and it is faster clock for clock it still loses...Why? because it will be a whole new socket and perhaps even newer memory ddr3, while for many of us who have scrapped AMD systems for C2D systems can pop a Kentsfield quad core in current systems....

I think this is AMD's year to stop the hemorraging...2008 may be their push back to top...All specualtion ofcourse since we dont know much for sure about K8L...It also doesn't help the fact that AMD hasn't fully emerged their 65nm parts and Intel will be moving to 45nm....Intels process is too far ahead of INtel...AMD almost needs to leap frog over one process to level up the playing field...

 

[Capt Caveman]

Before buying AMD stock, you should have done the research you are asking now.

 

[NuAlphaMan]

Originally posted by: Capt Caveman
Before buying AMD stock, you should have done the research you are asking now.

Touche! :thumbsup:

 

[Viditor]

Originally posted by: Duvie
Even if AMD pulls off a decent launch with its K8L and it is faster clock for clock it still loses...Why? because it will be a whole new socket and perhaps even newer memory ddr3, while for many of us who have scrapped AMD systems for C2D systems can pop a Kentsfield quad core in current systems....

I think this is AMD's year to stop the hemorraging...2008 may be their push back to top...All specualtion ofcourse since we dont know much for sure about K8L...It also doesn't help the fact that AMD hasn't fully emerged their 65nm parts and Intel will be moving to 45nm....Intels process is too far ahead of INtel...AMD almost needs to leap frog over one process to level up the playing field...

Actually...

1. AM2+ chips are backwards compatible to AM2 (the sockets are identical), the only difference will be lack of availability of HT 3.0 on older boards.
2. DDR3 is in 2008 now for AMD
3. A successful launch of K8L will mean that C2D is slower across the boards (both low and high end)
4. While Intel's 45nm begins production in 2007, you won't be able to get one until 2008
Otellini at IDF
5. AMD is releasing 45nm 6 months later, though keep in mind that a node change doesn't help performance...since Intel has to use so much more cache (I believe they are going to 6MB with Penryn) to reduce the latency penalty compared to AMD's ODMC, they have a far greater need for a node change than AMD does, and a majority of that real estate is already spoken for.

Edit: BTW, to answer the question I think that Intel is going to pummel AMD in the first half of 07, and AMD will pummel Intel in the second half.
Unfortunately for Intel, 70% of all chip sales occur in the second half...

 

[Arkaign]

AMD will be back, but 2007 isn't likely to be it. Intel can continue leveraging clock speed increases and a massive clock-for-clock performance advantage over AMD at this stage. K8L is likely to balance this somewhat, but judging from previous major processor launches, I'm estimating Spring '08 for K8L, about 6 months late from the 'expected' launch frame.

It may well be that Core2 performed much better than anyone expected, and that the existing K8L prototype work was only 'on-par' with C2/C2Q clock for clock. This may have forced AMD to bring additional performance enhancements to the board. This is all conjecture though, and just an educated guess on my part.

For now, AMD is in the worst shape they've been in since K6 was getting slapped around by Mendocino Celerons. Oh well, I think that any AMD64-class processor is more than fast enough for most people out there.

 

[Viditor]

Originally posted by: Arkaign
AMD will be back, but 2007 isn't likely to be it. Intel can continue leveraging clock speed increases and a massive clock-for-clock performance advantage over AMD at this stage. K8L is likely to balance this somewhat, but judging from previous major processor launches, I'm estimating Spring '08 for K8L, about 6 months late from the 'expected' launch frame.

Do you have any reason for this, or is just a "gut feeling"?

It may well be that Core2 performed much better than anyone expected, and that the existing K8L prototype work was only 'on-par' with C2/C2Q clock for clock. This may have forced AMD to bring additional performance enhancements to the board. This is all conjecture though, and just an educated guess on my part.

For now, AMD is in the worst shape they've been in since K6 was getting slapped around by Mendocino Celerons. Oh well, I think that any AMD64-class processor is more than fast enough for most people out there.

Here I can assure you that you are quite incorrect...
1. They now have all of the Tier 1 and Tier 2 OEMs as customers
2. Their marketshare has continued to expand, and so have their profits
3. They are now taken seriously (at least!) as a player in servers
4. They have a FAR better balance sheet with much lower cost debt

 

[Duvie]

Originally posted by: Viditor

Originally posted by: Duvie
Even if AMD pulls off a decent launch with its K8L and it is faster clock for clock it still loses...Why? because it will be a whole new socket and perhaps even newer memory ddr3, while for many of us who have scrapped AMD systems for C2D systems can pop a Kentsfield quad core in current systems....

I think this is AMD's year to stop the hemorraging...2008 may be their push back to top...All specualtion ofcourse since we dont know much for sure about K8L...It also doesn't help the fact that AMD hasn't fully emerged their 65nm parts and Intel will be moving to 45nm....Intels process is too far ahead of INtel...AMD almost needs to leap frog over one process to level up the playing field...

Actually...

1. AM2+ chips are backwards compatible to AM2 (the sockets are identical), the only difference will be lack of availability of HT 3.0 on older boards.
2. DDR3 is in 2008 now for AMD
3. A successful launch of K8L will mean that C2D is slower across the boards (both low and high end)
4. While Intel's 45nm begins production in 2007, you won't be able to get one until 2008
Otellini at IDF
5. AMD is releasing 45nm 6 months later, though keep in mind that a node change doesn't help performance...since Intel has to use so much more cache (I believe they are going to 6MB with Penryn) to reduce the latency penalty compared to AMD's ODMC, they have a far greater need for a node change than AMD does, and a majority of that real estate is already spoken for.

Edit: BTW, to answer the question I think that Intel is going to pummel AMD in the first half of 07, and AMD will pummel Intel in the second half.
Unfortunately for Intel, 70% of all chip sales occur in the second half...

1) that is a lot of speculation that even current AM2 boards will be able to run them...there are lots of things to consider on the hardware side that may not make them compatable. I can say with a bnit more certaintycause I have seen reports of ppl using kentsfields on a lot of current 965p and 975x boards. I think you may be bing a bit optimistic on something so far out...

2) OK...

3) Pure speculation...where is your research to prove this...I haven't see any benches..In paper it seems like it should, but that is like theory...wait until we see the real deal...Like most AMD launches supply will be low so it is hard to have a successful launch IMO...

4) OK...but it seems I have been hearing they have been working on this for awhile. I would expect the 45nm part may be able to be had by end of 2007...

5) I will take your word on it...



In the end you are too rosy about AMD....You only see optimiostic and if everything is what they say and everything falls into place perfectly....I dont see it. I am a bit more realistic and follow past performance of what we were told and what was actually delivered.

I say beg 2007 is further hemorraging...2nd half of 2007 may be a slowing of that but is still Intels...2008 will be the sun coming thu the clouds for AMD and righting the ship....

This is al in terms as the leader. I actually think AMD will do financially well. The 65nm parts have reduce d production cost and they will be more of a king of low and budget end pCs which still can make ample money. Average joes dont need dual core speed of the C2Ds or quad core kentsfield. the majority of the market will be ripe for AMD still

 

[Duvie]

Originally posted by: Viditor

Originally posted by: Arkaign
AMD will be back, but 2007 isn't likely to be it. Intel can continue leveraging clock speed increases and a massive clock-for-clock performance advantage over AMD at this stage. K8L is likely to balance this somewhat, but judging from previous major processor launches, I'm estimating Spring '08 for K8L, about 6 months late from the 'expected' launch frame.

Do you have any reason for this, or is just a "gut feeling"?

It may well be that Core2 performed much better than anyone expected, and that the existing K8L prototype work was only 'on-par' with C2/C2Q clock for clock. This may have forced AMD to bring additional performance enhancements to the board. This is all conjecture though, and just an educated guess on my part.

For now, AMD is in the worst shape they've been in since K6 was getting slapped around by Mendocino Celerons. Oh well, I think that any AMD64-class processor is more than fast enough for most people out there.

Here I can assure you that you are quite incorrect...
1. They now have all of the Tier 1 and Tier 2 OEMs as customers
2. Their marketshare has continued to expand, and so have their profits
3. They are now taken seriously (at least!) as a player in servers
4. They have a FAR better balance sheet with much lower cost debt

I agree financialluy they will be fine....They will loose market share in some markets and gain in some others...

As leaders I dont think they will have the best chip come end of 2007....That is a gut feeling, but since you have no proof as of otherwise then it is as viable an opinion as yours....

 

[cmdrdredd]

Originally posted by: Viditor

Originally posted by: Duvie
Even if AMD pulls off a decent launch with its K8L and it is faster clock for clock it still loses...Why? because it will be a whole new socket and perhaps even newer memory ddr3, while for many of us who have scrapped AMD systems for C2D systems can pop a Kentsfield quad core in current systems....

I think this is AMD's year to stop the hemorraging...2008 may be their push back to top...All specualtion ofcourse since we dont know much for sure about K8L...It also doesn't help the fact that AMD hasn't fully emerged their 65nm parts and Intel will be moving to 45nm....Intels process is too far ahead of INtel...AMD almost needs to leap frog over one process to level up the playing field...

Actually...

1. AM2+ chips are backwards compatible to AM2 (the sockets are identical), the only difference will be lack of availability of HT 3.0 on older boards.
2. DDR3 is in 2008 now for AMD
3. A successful launch of K8L will mean that C2D is slower across the boards (both low and high end)
4. While Intel's 45nm begins production in 2007, you won't be able to get one until 2008
Otellini at IDF
5. AMD is releasing 45nm 6 months later, though keep in mind that a node change doesn't help performance...since Intel has to use so much more cache (I believe they are going to 6MB with Penryn) to reduce the latency penalty compared to AMD's ODMC, they have a far greater need for a node change than AMD does, and a majority of that real estate is already spoken for.

Edit: BTW, to answer the question I think that Intel is going to pummel AMD in the first half of 07, and AMD will pummel Intel in the second half.
Unfortunately for Intel, 70% of all chip sales occur in the second half...

And it won't matter when Intel finally abandons FSB for HT based technology.

The problem is, Intel has a big jump on AMD and I'm not sure if getting a K8L that's better than kentsfield is gonna do any good. Intel has alot of money, much more than AMD. They likely will continue to use their ad campaign for the Core2 and get the name out there. regardless of what is actually 'better" or "more effecient" people buy products usually based on brand reputation or a product they know about.

 

[nyker96]

1. I believe that K8L cannot take back performance crown from Intel. Although it will be a drop-in upgrade for AM2 owners. This will make AM2 platform attractive if AMD is willing to adjust price depending on their relative perf to kensfield.
2. Intel will stay on top with native quad expected in 2007 when introduced.
3. AMD might not be doing badly if it can salvage OEM sales from ATI's chipset + AMD CPU platforms in 2007.

 

[PingSpike]

AMD are doing ok right now, but they don't have anything really competitive coming out for the high end and are suffering from the inability to make enough chips to meet demand. (which isn't that horrible of a position to be in really)

The trouble with AMD is they have no fvcking money. If they had money they probably would have blown it on R&D to stay at the top performance wise. But instead they had a ton of debt...

 

[Munky]

I voted Intel because they will continue to be the performance leader in the first half of 2007, and they have more money and production resources. I dont doubt that the K8L may take back the performance crown, but I also dont dismiss the possibility that it may get delayed, and it may not be as much of an improvement over the C2D as the latter was over the K8.

But at the same time, I couldnt care less for Kentsfield bacause more cores is not the solution to the problem when the majority of the apps/games I use dont even put my second core to good use. My next upgrade will be either a C2D or the K8L, depending on which one performs better across the board, in singlethreaded apps as well as multithreaded.

 

[SparkyJJO]

I don't think Intel's move to 45nm means much really, I honestly don't care whether my CPU is 90nm, 65nm, or 45nm, as long as it performs the way I want. Which would you buy, a 45nm fast chip, or a 90nm blazingly fast chip? I know my answer Size isn't everything.

 

[cmdrdredd]

Originally posted by: SparkyJJO
I don't think Intel's move to 45nm means much really, I honestly don't care whether my CPU is 90nm, 65nm, or 45nm, as long as it performs the way I want. Which would you buy, a 45nm fast chip, or a 90nm blazingly fast chip? I know my answer Size isn't everything.

Although the smaller die size and increased performance go hand in hand. So it's win/win.

 

[tylerw13]

Originally posted by: SparkyJJO
I don't think Intel's move to 45nm means much really, I honestly don't care whether my CPU is 90nm, 65nm, or 45nm, as long as it performs the way I want. Which would you buy, a 45nm fast chip, or a 90nm blazingly fast chip? I know my answer Size isn't everything.

sparky is there any thought to what you just stated???? of course it makes a difference...i guess if you dont understand what having smaller cored processors means, smaller core less heat ability to do more with less power loss...if you dont do much with your processor of course you dont care....i would still be with a 660 if i want doing work that is much faster with multi threaded apps and faster processors with less heat...faster processors with less powerloss (heat) its a no brainer

 

[cmdrdredd]

Originally posted by: munky
I voted Intel because they will continue to be the performance leader in the first half of 2007, and they have more money and production resources. I dont doubt that the K8L may take back the performance crown, but I also dont dismiss the possibility that it may get delayed, and it may not be as much of an improvement over the C2D as the latter was over the K8.

But at the same time, I couldnt care less for Kentsfield bacause more cores is not the solution to the problem when the majority of the apps/games I use dont even put my second core to good use. My next upgrade will be either a C2D or the K8L, depending on which one performs better across the board, in singlethreaded apps as well as multithreaded.

It is very very possible to find a Kentsfield in a reasonable price range by the time K8L comes out. I wouldn't completely rule out that possibility.

 

[Viditor]

Originally posted by: Duvie
1) that is a lot of speculation that even current AM2 boards will be able to run them...there are lots of things to consider on the hardware side that may not make them compatable. I can say with a bnit more certaintycause I have seen reports of ppl using kentsfields on a lot of current 965p and 975x boards. I think you may be bing a bit optimistic on something so far out...

Inquirer article
"The AM2+ CPUs are backward compatible with the existing AM2 CPUs and even the new AM2+ CPUs scheduled for 2007 announcement should work in the previous boards, but not at full speed.

They will have to depend on HT 2 not on HT 3 and even today's CPUs will work on future AM2+ platforms"

2) OK...

3) Pure speculation...where is your research to prove this...I haven't see any benches..In paper it seems like it should, but that is like theory...wait until we see the real deal...Like most AMD launches supply will be low so it is hard to have a successful launch IMO...

Remember I said "successful"...that means that the chip matches the paper design. We have discussed this ad nauseum, but the released design specs of K8L suggest that the cores should be equal but the HT and ODMC (to a lesser degree because of Intel's HUGE cache) should make it superior...

From Xbit's article
"The improvements we learned about so far such as increased to 32 bytes instruction sample, improved branch prediction algorithm, introduction of out-of-order reading, will eliminate a few bottlenecks and improve the integer performance. The expansion of SSE instruction sets will help improve the performance significantly in applications dealing with heavy floating-point or integer calculations using SSE instructions, where K8L will be able not only to compete successfully but even to outperform Conroe. The inability to decode and retire 4 commands per clock cycle in some cases may result into tangible performance gaps in integer applications. However, it may not be of that much importance in most cases, because the typical instruction execution pace in real integer applications does not exceed 2-2.5 instructions per cycle because of the data dependence."

4) OK...but it seems I have been hearing they have been working on this for awhile. I would expect the 45nm part may be able to be had by end of 2007...

I would agree with this...I expect them to try and bull them in to Dec 07

5) I will take your word on it...



In the end you are too rosy about AMD....You only see optimiostic and if everything is what they say and everything falls into place perfectly....I dont see it. I am a bit more realistic and follow past performance of what we were told and what was actually delivered.

Fair enough...but let me hold a mirror up to that "realism" if I may.
You are assuming that:
1. Intel is sandbagging and that they will release all of their products early...how often have they done this over the last 5 years?
2. AMD is lying or will make mistakes...when have they done this over the last 5 years?

My assumptions are taking BOTH companies claims at face value...

I say beg 2007 is further hemorraging...2nd half of 2007 may be a slowing of that but is still Intels...2008 will be the sun coming thu the clouds for AMD and righting the ship....

I will counter-predict that AMD will have all of the speed crowns in H2 07, but that Intel will come back at the end of 08 with Nehelam.

This is al in terms as the leader. I actually think AMD will do financially well. The 65nm parts have reduce d production cost and they will be more of a king of low and budget end pCs which still can make ample money. Average joes dont need dual core speed of the C2Ds or quad core kentsfield. the majority of the market will be ripe for AMD still

I partially agree (your analysis is quite correct for the desktop market IMHO).
Remember that the server market is where the big bucks are...I am sure that Intel is far more concerned about the success or failure of Tigerton than anything else at this point...

 

[cmdrdredd]

quoting rumor sites as fact is very wrong...

oh well...live the dream for now.

 

[Viditor]

Originally posted by: cmdrdredd
quoting rumor sites as fact is very wrong...

oh well...live the dream for now.

If you are referring to the quote from the Inquirer, then I suggest you try a google of the terms "am2+ compatible"...this is really well known now. The Inq was just one of the first that popped up for me...

 

[Hard Ball]

Originally posted by: Arkaign
AMD will be back, but 2007 isn't likely to be it. Intel can continue leveraging clock speed increases and a massive clock-for-clock performance advantage over AMD at this stage. K8L is likely to balance this somewhat, but judging from previous major processor launches, I'm estimating Spring '08 for K8L, about 6 months late from the 'expected' launch frame.

Actually, it will be much sooner than that. At Fab36, the second turn on K8L dice are about taking place, if not already. Maybe another 1 or 2 turns of the wafers most likely will fix the majority of the errata. The best estimate is that some production volume will begin to trickle out late spring. Of course, the first volume products will be ear-marked for the server variants, although some SKUs may become available for 4X4 at around the same time-frame.

It may well be that Core2 performed much better than anyone expected, and that the existing K8L prototype work was only 'on-par' with C2/C2Q clock for clock. This may have forced AMD to bring additional performance enhancements to the board. This is all conjecture though, and just an educated guess on my part.

Yeah, I do think that AMD got caught a little off-guard with the IPC advantage that CMA presented, especially with Intel enjoying the large lead in process tech; and may have had to go back and rework some of the elements in the original Rev H blue print.

It seems reasonable that we will see two substantially different iterations of releases. One at mid-year 07 that will emcompass much of what was planned for the original Rev H "hounds" (name inherited from the failed K9 in 2004). While the second in 1H08 with HTT3.0 and DDR3 will yet bring additional IPC improvements, although not too much is known about that second iteration of K8L at this point.

For now, AMD is in the worst shape they've been in since K6 was getting slapped around by Mendocino Celerons. Oh well, I think that any AMD64-class processor is more than fast enough for most people out there.

They are in pretty bad shape. Although their main woes in the next 12-24 months will be production related, and especially moving to 65nm and 45nm nodes. The next two years will bring the necessity for larger transistor budget per die, in each of the releases for K8L (mid 07, and 1H08), and that will in turn depend on the manufacturing side of things, including how things are going at E. Fishkill.

 

[SparkyJJO]

Originally posted by: cmdrdredd
Although the smaller die size and increased performance go hand in hand. So it's win/win.

In general yes, but not all the time. If it is so, then why are some saying that AMD's move to 65nm is just a shrink and not likely to do much for performance at this stage? Shrinks can provide the pathway towards higher performance but just because I shrink a process doesn't mean it is gonna be better performing if the actually design inside the core is the same thing.

Originally posted by: tylerw13
sparky is there any thought to what you just stated???? of course it makes a difference...i guess if you dont understand what having smaller cored processors means, smaller core less heat ability to do more with less power loss...if you dont do much with your processor of course you dont care....i would still be with a 660 if i want doing work that is much faster with multi threaded apps and faster processors with less heat...faster processors with less powerloss (heat) its a no brainer

No need to throw an insulting statement in there Yes I was thinking. It makes a difference yes, smaller does equal less power usage and all that. But that wasn't my point. My point was it is possible to have a 90nm chip that happens to be better performing than a 45nm chip. It was a silly example because that won't happen, but it is entirely possible to have a 65nm chip outperform a 45nm chip. All depends on the inner workings and designs and all that stuff.

Sorry if my statement was taken wrong, all I meant was just because a CPU is 45nm doesn't make it inherently "faster" than a larger process CPU, it is much more complicated than process size alone.

At any rate I didn't vote, I don't think we have enough info about AMD's K8L stuff for me to vote either way. I'm just gonna wait and see

 

[cmdrdredd]

Originally posted by: SparkyJJO

Originally posted by: cmdrdredd
Although the smaller die size and increased performance go hand in hand. So it's win/win.

In general yes, but not all the time. If it is so, then why are some saying that AMD's move to 65nm is just a shrink and not likely to do much for performance at this stage? Shrinks can provide the pathway towards higher performance but just because I shrink a process doesn't mean it is gonna be better performing if the actually design inside the core is the same thing.

Originally posted by: tylerw13
sparky is there any thought to what you just stated???? of course it makes a difference...i guess if you dont understand what having smaller cored processors means, smaller core less heat ability to do more with less power loss...if you dont do much with your processor of course you dont care....i would still be with a 660 if i want doing work that is much faster with multi threaded apps and faster processors with less heat...faster processors with less powerloss (heat) its a no brainer

No need to throw an insulting statement in there Yes I was thinking. It makes a difference yes, smaller does equal less power usage and all that. But that wasn't my point. My point was it is possible to have a 90nm chip that happens to be better performing than a 45nm chip. It was a silly example because that won't happen, but it is entirely possible to have a 65nm chip outperform a 45nm chip. All depends on the inner workings and designs and all that stuff.

Sorry if my statement was taken wrong, all I meant was just because a CPU is 45nm doesn't make it inherently "faster" than a larger process CPU, it is much more complicated than process size alone.

At any rate I didn't vote, I don't think we have enough info about AMD's K8L stuff for me to vote either way. I'm just gonna wait and see

There's no reaspn to think that Intel is going to do 45nm and NOT increase the performance at the same time

 

[Viditor]

Originally posted by: cmdrdredd
There's no reaspn to think that Intel is going to do 45nm and NOT increase the performance at the same time

True, but there's no reason that they will either...they didn't with the 65nm node on Netburst.

 

[harpoon84]

Originally posted by: Viditor

Originally posted by: cmdrdredd
There's no reaspn to think that Intel is going to do 45nm and NOT increase the performance at the same time

True, but there's no reason that they will either...they didn't with the 65nm node on Netburst.

True, but 45nm C2D will feature 6MB L2 cache, 1333MHz FSB and SSE4, so it should be at least 5% faster clock for clock compared to current C2Ds. SSE4 benefits are still unknown at this stage so I won't bother speculating.

Of course with a die shrink there is also the advantage of lower power consumption, which in turn generally leads to higher clockspeeds. Intel is aiming for 3.5 - 4GHz on 45nm DC and 3.46 - 3.73GHz on 45nm QC according to the latest roadmap.