AMD Athlon 64 delayed until September (Clawhammer)

[Cusqueno]

The Inquirer reports the Clawhammer will be delayed until September. They feel the Barton will be fast enough to compete with Pentium 4 chips until then. Take appropriate size grain of salt.


http://www.theinquirer.net/?article=7530

 

[Soulkeeper]

nooooooooo!!!!!!!

 

[robg1701]

http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543~66302,00.html

 

[dullard]

That is no surprize at all. AMD has been paper launching products - a 2 month delay on average. They were planning on paper launching the Athlon 64 in July. Well recently AMD stated that they are no longer going to paper launch products. Just a week later they say the Athlon 64 will be delayed 2 months. No surprize at all.

 

[KGB1]

What else is new? Though that means intel is in no rush to decrease prices nor raise the speed on their processors significantly until july/august. I would think AMD should let the hammer out of the bag much quicker. It won't help their quaterly income at all. AMD announced the death of Socket A.. at this rate, Socket A would be phased out in 2005.

 

[bgeh]

Originally posted by: Soulkeeper
nooooooooo!!!!!!!

 

[Alptraum]

I would think AMD should let the hammer out of the bag much quicker.

Desire and ability can be pretty far apart. I am sure if they could release it sooner they would.

 

[dullard]

Originally posted by: KGB
What else is new? Though that means intel is in no rush to decrease prices nor raise the speed on their processors significantly until july/august. I would think AMD should let the hammer out of the bag much quicker.

Intel has already announced two price cuts in the next two months. So I'm not sure what your first comment means. On Feb 23, Intel will cut all the current P4 prices by about $50. Sure that means very little on the high end but the $50 cut on the 2.53 GHz from $243 to $193 is a 20.5% cut. Then late March we will see a second price cut which will finish the job of the first cut - and fully push all chips down one ladder level from their current places. So the 3.06 GHz P4 which is currently $637 will drop to $589 and then to ~$401. The two cuts together on that chip will make a total of about 37% off. Note: these are official Intel prices and street prices do vary a bit.

I think AMD should let Hammer out as soon as possible. AMD thinks so too. But they have been struggeling with several issues (many are still rumors but several of those rumors have been confirmed by AMD). If Hammer was flawed on its release that will instantly kill any hopes that AMD has to get into the server market. So they will release Hammer when it is ready.

 

[KGB1]

The 865/875 chipsets will be out in May and June, there will be new Processors by then. I don't think those processors will be sold anything less than $400. Thats what I'm talking about. Sure of course I agree the price will decrease overtime, but not at the pace it was last year. Personally I will not buy a hammer this year, most likely for me the summer of 2004, all the bugs will be ironed out, maybe more stable motherboards also. I can live with mediocre products a bit longer.

 

[AndyHui]

Hhmm....

Not good.

 

[Cusqueno]

This is pure speculation, but will this give AMD time to add support for PC3200 DDR memory into the desktop chips?

 

[thorin]

This is no big deal. 99.9% of home users have no use for a 64bit processor anyway (99.99999% of home users don't use 4GB of RAM let alone 16 ExaBytes). Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel), and the architectural changes (esp on the 32bit side) are minimal (compared to Barton).

Thorin

 

[Mavrick]

Since the 2800+ is not even available yet, it doesn't surprise me much.
They also state that the Opteron is going to be released at 1,8 GHz and less, which is still less than expected...

Not to be pessimistic, but AMD is definately going down (their stock is a 5$ today)
At the rate at which they lose money, they will be in big trouble by midyear...

 

[Nemesis77]

Originally posted by: thorin
This is no big deal. 99.9% of home users have no use for a 64bit processor anyway (99.99999% of home users don't use 4GB of RAM let alone 16 ExaBytes). Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel), and the architectural changes (esp on the 32bit side) are minimal (compared to Barton).

I don't really care for 64-bit right now, I care about the double GP-registers X86-64 brings along with it. But OTOH, few years ago 128megs was the standard RAM-configuration. Nowadays 512megs is common. It wont be long 'till regural users will have machines with over 1 gig of RAM, and we will start needing 64bit CPU's then!

As for me, I'm not troubled. My upgrade is just delayed by few months, and I can wait . I'll just have more time to save money and I will get even faster machine for my money.

 

[Macro2]

RE:"This is no big deal. 99.9% of home users have no use for a 64bit processor anyway (99.99999% of home users don't use 4GB of RAM let alone 16 ExaBytes). Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel), and the architectural changes (esp on the 32bit side) are minimal (compared to Barton)"

True. OTOH, the slippage does mean that yields aand bins on Hammers aren't that good. Otherwise they would have just moved to Clawhammer and skipped Barton which is a lot like K6-III was.
Even with Barton, Clawhammer could exist like The Pentium Pro did until they get it up to speed.


Mac

 

[n0cmonkey]

Damnit. Of course this is going to happen. Now I have to wait another couple of months before a decent x86 based chip is released.

 

[thorin]

Originally posted by: Nemesis77

Originally posted by: thorin
This is no big deal. 99.9% of home users have no use for a 64bit processor anyway (99.99999% of home users don't use 4GB of RAM let alone 16 ExaBytes). Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel), and the architectural changes (esp on the 32bit side) are minimal (compared to Barton).

It wont be long 'till regural users will have machines with over 1 gig of RAM, and we will start needing 64bit CPU's then!

Why? with 32bits of address space you can address up to 4GB of RAM (assuming chipset/bios support). (Not to mention adress translation schemes etc that allow for 36bit memory addressing on 32bit processors etc....). Did you just pick that 1GB = need for 64bit proc out of thin air?

Thorin

 

[Nemesis77]

Originally posted by: thorin
Why? with 32bits of address space you can address up to 4GB of RAM (assuming chipset/bios support). (Not to mention adress translation schemes etc that allow for 36bit memory addressing on 32bit processors etc....). Did you just pick that 1GB = need for 64bit proc out of thin air?

In few years time, the amount of RAM in a powerusers machine has increased by 4-8 times what they were before. With a growth-rate like that, we will soon be at the limits of the maximun addressable memory, and that's why we need 64bits. RIGHT NOW average user doesn't need it, but few years down the road he does.

Where exactly did I say 1GB = 64bit?

Regural 32bit addressing gives uf 4gigs of RAM. That's not really that far in the future. Workstations are already at the limit (Doesn't the new PowerMAC have 2gigs?). And how about servers? 64bits gives us 4,500 terabytes, that's more than enough (for now at least )

 

[Mem]

Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel),

It`s always the case of software trying to catchup with new hardware ,so software will always be dragging its feet.I`m pretty sure AMD more or less have the Athlon 64 ready right now, but quantity and also availability of boards etc is a different story,it`s just a case of no real demand for it right now so until AMD decides when`s the best time to release it we will just have to wait.

 

[andreasl]

Originally posted by: thorin

Originally posted by: Nemesis77

Originally posted by: thorin
This is no big deal. 99.9% of home users have no use for a 64bit processor anyway (99.99999% of home users don't use 4GB of RAM let alone 16 ExaBytes). Plus there is very little 'consumer' 64bit software (let alone a OS....MS has been draggin their feet on x86-64 Windows even though they jumped on EPIC/IA64 Windows for Intel), and the architectural changes (esp on the 32bit side) are minimal (compared to Barton).

It wont be long 'till regural users will have machines with over 1 gig of RAM, and we will start needing 64bit CPU's then!

Why? with 32bits of address space you can address up to 4GB of RAM (assuming chipset/bios support). (Not to mention adress translation schemes etc that allow for 36bit memory addressing on 32bit processors etc....). Did you just pick that 1GB = need for 64bit proc out of thin air?

Thorin

64 bits shouldn't have anything to do with the Hammer release to the desktop. What AMD needs is a higher performing CPU on *32-bit* applications. Something they can slap a higher model number on. Having that allows them to charge a higher price for it. It's obvious that they are having some issues with Hammer related to yields/speed bins. Otherwise they wouldn't have delayed it.

There is less pressure on performance for a server CPU since that will be competing with lower clocked Xeon MPs. So in that regard they can compete well enough to release Opteron.

 

[Alptraum]

I`m pretty sure AMD more or less have the Athlon 64 ready right now, but quantity and also availability of boards etc is a different story,it`s just a case of no real demand for it right now so until AMD decides when`s the best time to release it we will just have to wait.

I disagree. At least to an extant. They have so much riding on this I can't see them waiting if they actually have it ready. If it truely is a board issue, thats abit different. But usually if the big board makers are given the specs they can churn out pretty much anything in a pretty short time. So I doubt that boards are the hold up.

 

[dullard]

I don't really care for 64-bit right now, I care about the double GP-registers X86-64 brings along with it. But OTOH, few years ago 128megs was the standard RAM-configuration. Nowadays 512megs is common. It wont be long 'till regural users will have machines with over 1 gig of RAM, and we will start needing 64bit CPU's then!

You do realize that Clawhammer cannot use more than 4 GB right? Even though it is 64-bit, it wasn't made to use 64 bit memory addresses. Opteron is a different story - it can use more than 4 GB of memory.

Typical great computer in the home (entusiasts have these and they are considered nearly the top of the line):
1988: 256 kB memory
1993: 4 MB memory
1998: 64 MB memory
2003: 512 MB memory
In the 5 years from 1988 to 1993 memory use increased by 16 times. In the 5 years from 1993 to 1998, memory use increased by 16 times. In the 5 years from 1998 to 2003, memory use increased by only 8 times. So actually memory growth has slowed down. But lets pretend the 16 times in 5 years trend will continue. That is about a 75% growth per year. So lets assume that trend will continue:

Great home computers in the future
2004: 896 MB memory
2005: 1.5 GB memory
2006: 2.7 GB memory
2007: 4.8 GB memory

So it isn't until 2007 when a great home computer reaches the 4 GB memory barrier. And note: the normal home computer is about half that of a great computer. So in 2007, the typical home computer still will be roughly 2.4 GB memory. We have many years and many models of CPUs coming before most people will even worry about that 4 GB barrier.

 

[Mem]

If it truely is a board issue, thats abit different. But usually if the big board makers are given the specs they can churn out pretty much anything in a pretty short time. So I doubt that boards are the hold up.

Look how long it took for the nForce2 boards to arrive,besides we don`t know the quantity of Athlon 64 CPUs available or if all testing has been done or if they`ve found any bugs etc on both the Athlon 64 or motherboards ,also AMD have to take into acount their Barton CPU since they won`t want to hurt their Barton sales or demand,in the end all we can do is speculate & only AMD knows for sure what`s happening.

 

[alexruiz]

I think it is a smart move by AMD. They are releasing the hammer (yes, Opteron will be available by April 22nd). They just decided to delay the desktop version. Opteron get full capacity and attention. Remember, money is the key here, and the opterons will for sure give more money to AMD than the Athlon 64 per piece. Granted, volume is lower, but at a product introduction the capacity ramp up is crucial, and the less demanded opteron should allow AMD to fine tweak the production process in order to allow higher volumes for future demand.

Talking about money, they have invested in Barton, and releasing a product that eclipses Barton so close to it means that basically Barton would be DOA. AMD needs to recover some of the cost associated with the Barton development.

As far as being competitive, I think they are so far. It wouldn't be a smart move to stop socket A when it has reached its highest point.

The question mark here is how welll will the opteron being received.

 

[dullard]

releasing a product that eclipses Barton so close to it means that basically Barton would be DOA.

I've been looking at it the other way. Barton will be significantly cheaper. Barton will work in many current motherboards. Barton will have about the same performance as the initial Clawhammers. Thus releasing them at about the same time means that the Clawhammer would have been DOA.