PDA

View Full Version : AMD's 90nm Processors to Dissipate 105W?


Malfunction
12-16-03, 09:47 PM
http://www.xbitlabs.com/news/cpu/display/20031215184404.html

While Intel Corporation ran into very serious heat dissipation issues with its NetBurst architecture CPUs and 90nm strained silicon technology, looks like its arch rival AMD will also have some hot chips in future.

Sources close to AMD’s partners said the company is going to employ some new power requirements for mainboards designed to support future AMD64 processors, such as Athlon 64 4000+ and the Athlon 64 FX-55, both most probably to be made using 90nm technology. Changes are necessary for mainboards set to come in the second half of 2004.

According to currently available details, maximum current (IDD) of future 64-bit processors from AMD will be 80A, while maximum thermal power will be 105W. These are mandatory requirements for mainboards in the second half of the year. The requirements are only for Socket 939 mainboards, as 754-pin and 940-pin products have very limited future, as AMD roadmap for 2004 revealed.

Apparently, AMD’s 90nm chips will have Vcore at 1.20V – 1.35V, in contrast to current core voltage of 1.40 – 1.55V for 64-bit chips. The first 90nm chips are expected to hit 2.60GHz, but AMD currently does not specify the top speed bin for its new CPUs.

Intel's Prescott 3.60GHz processors are expected to dissipate up to 103W of heat early next year.

Before its 90nm chips emerge, AMD will bring out some new versions of its current 0.13 micron chips. The new CG revision targeted to be in production in late Q1 2004 will resolve some issues with current AMD64 performance-mainstream and high-end processors, such as 2T DRAM timing enhancements to enable desktop customers to achieve higher density and/or higher speed memory (the change may be adored by overclockers) and removal of requirement for use of identical DIMMs for the 2nd and the 3rd DIMMs in 3-DIMM mainboards. There will be CPUID change to “00000F4AH” as well as OPN change for the new processors.

Representatives for Advanced Micro Devices do not usually comment on future products.

Ummmm... :rofl

Peace,

:spongebob

Viral
12-16-03, 11:07 PM
There are major differences between this and the quoted prescott 103watt...

1) That 103watt figure for prescott was not the minimum standard for motherboards, it was just a leaked figure of what the prescott could run at.. it was later confirmed that the prescott would most likely require 100watt+

2) Socket 939 will be around for a long time! AMD are just trying to make sure the first socket 939 mobos will be compatible with as many future A64's as possible. It is most likely that if any K8 actually reaches this amount, it will be towards the end of K8's life, ie mid-late 2005. This would mean possibly an FX-61?? i don't care if 3.0ghz K8's require up to 105watt of energy.. for that performance and time it will be rather acceptable. It is likely that such a processors competition will require a much higher number, as has been shown before.

So all this means is that early socket 939 motherboards will be compatible with new K8's for years to come.

Besides.. look at those v-cores, 1.20 - 1.35v!!!

Remember todays socket 940/754 boards have a 89watt power requirement. This is still not being met by todays processors, but they do come rather close. Because these two standards are not required to remain compatible for years to come, this indicates that the new standard is only set higher because it will be around so long.

So please, learn what these articles are stating.. i wouldnt say its something to laugh at. In fact there have been many indications this would happen and i have been expecting exactly this to happen. All AMD are trying to avoid is what will happen with prescotts compatibility with current mobos. Many simply do not meet the power requirements to be compatible with prescott.

Malfunction
12-16-03, 11:49 PM
Originally posted by Viral
All AMD are trying to avoid is what will happen with prescotts compatibility with current mobos. Many simply do not meet the power requirements to be compatible with prescott.

Ya mean like AMD's Socket 754 and Socket 940? :rolleyes: Talk about trying to show a better light on the face of AMD even though they are currently building/marketing mobo(s) and CPU's that have an extemely short life span compared to the Intel Socket 478.

Nice try Viral...

Peace,

:wtf:

Viral
12-17-03, 12:31 AM
Socket 754 and 940 will continue for some time, only in different market sectors to what some of the processors released today use these sockets for.

All signs point to 754 remaining a socket with enthusiast options (as it is currently) up until the 3400+, then it will change its targetted market sector to mainstream - performance.

940 has always been the socket for opteron. Opteron looks to remain solely on socket 940 until at very earliest mid 2005. Even after that time it is unsure whether 940pin opterons will just remain being produced at the speed grade they reached, or continue progression as normal as an alternate/cheaper/more retro-compatible option.

Socket 754 i partially agree with you on.. but none the less it's life is not coming to an end any time soon. It will meerly shift from an enthusiast platform to a performance mainstream one. For current 754pin A64 owners, this isn't a great thing, but the sockets life is not coming to an end.

Oh and even mentioning socket 940 and "extremely short life span" in the same sentence is a complete joke. AMD are dedicated to keeping socket 940 as opterons platform for a long time. Nothing has shown otherwise.

Nice try Viral...

You picked out one point from my post of many points contradicting your uncertainty about AMD keeping future processors cooler than intels. The point you argued about was not even related to the original article, and your argument was severly flawed and downright wrong. How is that a 'nice try'? I'm sorry but it seemed to me as if you are interpreting this information in a negative light simply because of how it was presented by xbit (judged solely by your reaction). All it is stating is that AMD have created a motherboard standard to ensure compatibility in the future for all 90nm 939pin K8 processors. Is makes special mention as to the what this standard is in the form of maximum thermal power. All this indicates is that AMD believe 90nm 939pin K8's should not top 105watt anytime in the not too distant future.

Oh and this statement:
The requirements are only for Socket 939 mainboards, as 754-pin and 940-pin products have very limited future, as AMD roadmap for 2004 revealed.

Is just completely wrong. That might possibly be the case for 754, because even when that platform gets 90nm processors, they shouldnt draw as much energy as their more powerful siblings because they will be in the form of XP continuums. However, saying the same standard hasnt been set for 940 because it will be short lived is pure ignorance. Opteron is moving to 90nm at the same time as A64. It will most likely have similar requirements, but my speculation is xbit just diddn't get that information. Remember this hasn't been announced yet as the final standard, so i'm sure when the information is properly distributed the socket 940 specs will come with it. We all know how similar Opteron is to AFX, even the 939pin AFX wont differ to greatly from opteron, so why would socket 940 motherboards not require similar standards to support 90nm opterons?? surely they will. I think xbit just diddn't mention this because it's not Athlon 64 orientated, which seems to be all these types of sites are caring about lately.

-=DVS=-
12-17-03, 02:49 AM
thats to much to read :p is it GOOD or BAD thing ? :D

MUYA
12-17-03, 02:58 AM
Originally posted by -=DVS=-
thats to much to read :p is it GOOD or BAD thing ? :D

it's just two forum members having a go at each other..over x-bit's article which is based on rumours and speculation....ya speculation over heat dissipated or thermal power or sumat.

100+ W = can fry an egg on it = includes micro frying pan? = must be teflon coated?

sxotty
12-17-03, 07:35 AM
Malfinction no one has compared with intel for shortlived motherboards, so don't even bother.

edit: and btw athlon 64 4000+ mmm now that sounds toasty err nice :)

Malfunction
12-17-03, 08:00 AM
Originally posted by Viral
You picked out one point from my post of many points contradicting your uncertainty about AMD keeping future processors cooler than intels. The point you argued about was not even related to the original article, and your argument was severly flawed and downright wrong. How is that a 'nice try'? I'm sorry but it seemed to me as if you are interpreting this information in a negative light simply because of how it was presented by xbit (judged solely by your reaction). All it is stating is that AMD have created a motherboard standard to ensure compatibility in the future for all 90nm 939pin K8 processors. Is makes special mention as to the what this standard is in the form of maximum thermal power. All this indicates is that AMD believe 90nm 939pin K8's should not top 105watt anytime in the not too distant future.

Is just completely wrong. That might possibly be the case for 754, because even when that platform gets 90nm processors, they shouldnt draw as much energy as their more powerful siblings because they will be in the form of XP continuums. However, saying the same standard hasnt been set for 940 because it will be short lived is pure ignorance. Opteron is moving to 90nm at the same time as A64. It will most likely have similar requirements, but my speculation is xbit just diddn't get that information. Remember this hasn't been announced yet as the final standard, so i'm sure when the information is properly distributed the socket 940 specs will come with it. We all know how similar Opteron is to AFX, even the 939pin AFX wont differ to greatly from opteron, so why would socket 940 motherboards not require similar standards to support 90nm opterons?? surely they will. I think xbit just diddn't mention this because it's not Athlon 64 orientated, which seems to be all these types of sites are caring about lately.

We are all aware of AMD and Intel's transistion to 90nm. The thing I find hillarious is we all know why Intel is having their difficulties with heat. It will just be a matter of time before AMD will reveal theirs. I figure the same face that Intel put on the process with continue with AMD till the cat is out of the bag. (ie. First working samples of 90nm chips being reviewed.) :thumbsup:

Malfunction no one has compared with intel for shortlived motherboards, so don't even bother.

Yes I know this sxotty, behaving similar is not exactly great in my book either. ;)

Peace,

:afro:

Dazz
12-17-03, 12:25 PM
The SK 939 & 940 will have a long life, the 939 will be for ALL desktop users, while the 940 will be for servers and later, dual core CPU's.

Head_slinger
12-17-03, 12:32 PM
Hey malfunction, whats with knocking AMD's ive had nothing but great experiences. :)

Malfunction
12-17-03, 12:38 PM
Originally posted by Head_slinger
Hey malfunction, whats with knocking AMD's ive had nothing but great experiences. :)

No knocking them bud, just the peeps with the mentality that all is good and holy on the otherside. They paint a pretty picture till the poo hits the fan or Intel announces a change in their roadmap. Happens everytime...lol, that's all. ;) I am actually waiting till the 939pin FX's appear being hosted by a bus lockable mobo. If that never comes, I might just stick with Intel and their overclockability if I am to deal with the heat situation no matter what solution I choose. :thumbsup:

Peace,

;)

Head_slinger
12-17-03, 12:44 PM
fair enough :cool:

Viral
12-17-03, 07:28 PM
We are all aware of AMD and Intel's transistion to 90nm. The thing I find hillarious is we all know why Intel is having their difficulties with heat. It will just be a matter of time before AMD will reveal theirs. I figure the same face that Intel put on the process with continue with AMD till the cat is out of the bag. (ie. First working samples of 90nm chips being reviewed.)

Well i guess we better wait and see... It still isn't final that prescott will run anywhere near as hot as it did compared those early samples, so i guess it's a wait and see thing for both sides.

Don't get me wrong, i'm not always over-positive and 'with' anything AMD do, i was heavily critical of them afew months before the A64 launch because i believed they would release only a 2.0ghz A64 laybled as a 3400+.

AMD's 90nm transition could still turn out to be a horrid ordeal, i'm not claiming it will be completely smooth and produce fine results straight away, but i don't think this article provides any evidence to the contrary. All it may proove from what i gathered, is that 90nm 939pin K8's could produce up to 105watt at some point in time.. i think its safe to say AMD wouldn't create a standard that would only last for one speed grade, so this number is more likely to be met by the end of such a generations life span.

Rob_0126
12-29-03, 08:23 PM
Why dont they, instead of raising the ghz/mhz, why not just raise the processing speed?

I mean, how high are they gonna go, before you need a frig cooling your processor, out of the box?

Robert

Soylent
01-01-04, 08:36 PM
Increasing processor speed by use of more transistors increases power usage too, and having a large CPU die is expensive.

Using a smaller process has usually reduced power consumption per transistor, but if you increase the number of transistors used and the clock frequency you will probably end up using more power than previously anyway.

It appears to me that processor manufacturers try to increase both complexity and clockspeed as much as they can afford(i.e. increase the number of transistors but don't go unreasonably high because that is expensive, increase the clockfrequency as much as can be reliably done and use speed binning to be able to offer high end and low end chips of the same kind).

Dazz
01-02-04, 03:31 AM
At the moment AMD has an advantage, it's more expensive due to it's large die core but gives more contact area to diperse heat.