AMD Roadmap: core questions.

[ku]

Link to AMD Roadmap

Ok, so I just have a few questions about upcoming cores from AMD:

1. What benefits are there to the Thoroughbreds other than the .13u? And exactly what are the benefits of .13u?
2. What benefits are there to the Barton core? (short answer is fine)
3. What exactly is the Clawhammer? Why is it so special?

Thanks.

EDIT:
NVVVMM

 

[Swanny]

I don't see where you get that the MP is cheaper than the XP. MP prices are higher at all speeds.

Now on to the questions:

1. Not many for sure. Some people think that it will also have more cache and maybe a 166Mhz DDR interface.
2. SOI manufacturing is the biggest one. This will allow much greater clock speeds and better performance to boot. If the above improvements are not included in the Thoroughbred core they probably will be in Barton.
3. It's AMD's first 64bit CPU!! It will be quite the performer and be able to do 32bit natively.


Hope that helps some.

 

[ku]

Oops. It seems that i've compared the xp2000 to the mp1900. Stupid me. I always do dumb crap like that.

Thanks. That answered a lot of my questions. I'll look investigate further on those terms u just spit out at me. But please, still post as you wish!

 

[AGodspeed]

1. According to an AMD official, Thoroughbred will be no different than the current Athlon XP's (minus the .13u process). The advantages of .13u for people like us are better overclockability, more headroom for MHz in the future, and a cooling running core. The benefits for AMD is that having a .13u Athlon will be very cheap to produce, because the processor is smaller than the current Athlon XP's and smaller processors are always cheaper to make, which will probably translate into lower prices for us in the future.

2. The benefits of the Barton processors are too far away at this point to make any final judgement. All we know is that Barton will encorporate an additional feature called Silicon On Insulator (SOI) technology. Whether AMD will implement architectural changes to Barton is unknown as of this moment (architectural changes like adding more cache (SRAM) and a higher FSB (front side bus) increase performance).

3. ClawHammer is the desktop version of AMD's 8th generation processor architecture (the Athlon is AMD's 7th generation). It is the stripped down version of the 64-bit SlegeHammer processor (a more powerful processor meant for servers, etc. only to be released this time next year). Go here for all of Hammer's specifics. People are excited about Hammer because it is supposed to incorporate a much more advanced silicon process (.13u SOI) in addition to a much more advanced processor architecture (built-in Northbridge with integrated memory controller) than the Pentium 4. If you read up on all the info that's in the link above, you'll get what I'm talking about.

Hope that helped.

 

[ku]

Thanks AGodSpeed. That linky isn't working, maybe it's just temporary beceause the site isn't working if I go through the search either.

Also, I do believe the Palomino cores are 4th generation. Thoroughbreds are 5th. Bartons would be 6th. Correct me if I'm wrong.

 

[mrman3k]

Right now I have an Athlon Classic 900MHz, I want to do a major upgrade sometime between the summer and winter of this year, now what I was thinking about is should I go for the Thoroughbred, Barton, or Clawhammer? I am not totally convinced that the Clawhammer will be comming out at the end of 2002 and I wonder if I should wait for the Barton or just go for the Thoroughbred.

 

[Vic]

If you are holding out for a big change to do a major upgrade, here are some things to consider:
Sometime this summer, probably in the middle of Tbred cycle, AMD should move to 166MHz FSB. These will be supported by VIA's KT333 chipset. Before the fall, both AMD and Intel are supposed to move to 8xAGP, with Nvidia and ATi releasing new vid cards to support this (Geforce4). VIA is supposed to support 8xAGP with the KT333A chipset. So this summer/early fall should see some really good upgrade opportunities with higher bus speeds, 8xAGP, and a new line of vid cards.
As far as the AMD roadmap is concerned, I really don't see Barton becoming big until late this coming fall, and Hammer probably won't hit the desktop for another year.

In answer to the originally poster, what good is the switch to 0.13micron?? Why don't you ask the Intel fan-boys about how excited they are over Northwood? Higher speeds, better OC's, less heat, lower power consumption... a die-shrink is always a good thing.

 

[Rand]

<<

Also, I do believe the Palomino cores are 4th generation. Thoroughbreds are 5th. Bartons would be 6th. Correct me if I'm wrong. >>



Quite wrong. The Palomino is the fourth core based on the original K7 microarchitecture. But all of the cores you listed are classified as 7'th generation X86 microprocessors.
The entire K7 lineup from the initial SlotA Athlon 500MHz, up to the future Barton core are all 7'th generation processors.

I don't feel like typing up everything all over again so i'll just quickly post over my posts from another thread on a similar topic.



<<
AMD's official statement is that current 266MHz FSB SocketA chipsets will support both the Barton and Thoroughbred cores, providing the motherboard will support the decreased core voltage.
Officialy they've stated they intend to scale the .18u Palomino core to 1.8GHz/2000+ by Q1 2002. The .13u Thoroughbred core is expected to take them to "at least" 2.07GHz/2600+ by Q3 2002. The Barton core is planned to emerge shortly thereafter and further scale the K7 to beyond 2.07GHz, though specific clockspeed have not yet been announced.
AMD is presently planning t continue production of the Barton core in the low end after the ClawHammer debutes in very late 2002/early 2003.

Officially there are planned to be no further changes to the K7 core beyond the expected die shrinkage, and SOI with Barton.
>>



and



<<
The following are AMD's official roadmap statemments and are open to interpretation. Also, these statements may well change depending upon yields and competition from Intel. My personal opinions are mixed in and marked as such.


Palmonio will scale to 1.8GHz/2200+ in Q1 2002. Thoroughbred hits a maximum of 2600+/1.8GHz.
ClawHammer is due to launch about 8 months prior to SledgeHammer. ClawHammer introes using .13u process, and later transfers to .08u process. SledgeHammer introes on a.08u process. ClawHammer is set to debute at a performance rating of 3400+.

It is rumored to clock in at 2GHz actual clockspeed, but the latest statements by Hector Ruiz have begun to put this in doubt.
ClawHammer performance rating may very well change considerably from now until it's actually produced as all statements now are based upon software simulations of the ClawHammer and SledgeHammer and yield expectations.

As stated previously, AMD has maintained for almost 1.5 years now that the K7 core will not exceed 133MHz DDR FSB until at least 2003, and so it is extremely dubious that it will ever see anything faster as ClawHammer is expected to be available Q4 2002.
If Intel were to open up a large performance gap between them and AMD with at minimum half a year left before ClawHammer then I might consider it a possibility, otherwise it's extremely unlikely at best that we will ever see a 166MHz DDR FSB on the K7 core.



Officially, Q2 2002 is the release date of the thoroughbred. I would guesstimate it will tag in at the very end of that period, so probably June I'm thinking.
Without a better idea as to the release date I cannot hazard any more then a guess as to when the websites may recieve Thoroughbreds for testing.

Edit:
One further statement. Last I heard ClawHammer engineering samples were expected to be in the hands of reviewers by the very end of 1H 2002 (June?).
That statement is almost two months old though, and I've no further word as to when they might be seen. I've not the faintest idea as to when major websites will see final production samples of ClawHammer.
>>

 

[KenAF]

Thanks rand...I was just getting ready to respond to this thread and you saved me the time.

It's good to see that at least a few members here actually listen to (or follow coverage of) AMD's conference calls, rather than relying on the unfounded rumors perpetuated by various fan sites.

 

[AGodspeed]

Some Thoroughbred info from today...

Anand mentions in his Northwood vs. Athlon 2000+ article today that:

According to AMD, their Thoroughbred (0.13-micron) Athlon core will begin shipping by the end of this quarter. We can extrapolate that to mean that retail availability is still on track for the first half of this year.

And Tomshardware also mentions that Thoroughbred

...will probably make its debut as the "Thoroughbred" Core at this year's CeBIT [March 20th].

I expect full availability of 1.8GHz and possibly 1.866GHz Thoroughbred processors by sometime this April.

 

[KenAF]

AGodSpeed,

Why even bother with third party sites? Everything Rand said as far as release dates is essentially fact, or at least, it's what AMD has publicly announced. During the last conference call, AMD said virtually everything that he posted (so far as cpus and release dates are concerned). As he notes, his reference was AMD's published roadmap.



<< I expect full availability of 1.8GHz and possibly 1.866GHz Thoroughbred processors by sometime this April. >>



I think that's probably a wee bit optimistic.

The next Athlon is the 2200+, based on the Palomino core at 1800MHz. It's to be announced end 1Q, or late March. AMD's roadmap shows 2Q 2002 for the Athlon 2400+ based on Thoroughbred, so conceivably it could come as early as April; that said, this isn't likely (June is far more probable). When AMD has specified quarters in the past for milestones, it's always been 'by the end of the quarter.' Moreover, I can't imagine that AMD would release the 2400+ at the same time as the 2200+, or even within a few weeks of that cpu.

 

[Rand]

<<
Why even bother with third party sites? >>



Potentially because AMD may be somewhat more willing to give out information to large influential third party sites then they would be to make the same information publically available to anyone that cares to do the research. Some of that information may not be protected under NDA and so they'd be free to pass it on.
The above is merely a possibility of course, as naturally I have no way of confirming where or how Dr. Pabst, and Mr. Shimpi acquired their information.

The vast majority of the information I've listed is via AMD's conference calls. They tend to be virtual treasure troves of information of future plans and technical specifications. The roadmap listed on their site is actually surprisingly vague and out-of date.

The noteable exception being the bit on ClawHammer engineering samples, as that bit of info comes direct from Drew Praire (PR Rep for AMD). Drew Praire is often quite willing to speak regarding AMD to anyone representing the online press in any manner.

 

[AGodspeed]

AMD's roadmap shows 2Q 2002 for the Athlon 2400+ based on Thoroughbred...

Really, what roadmap are you looking at? AMD's official stance (according to their last conference call) is that they will make Thoroughbred available during Q1 of this year. It is not inconceivable that we see 1.8GHz Thoroughbred processors by April (which is in Q2 btw).

 

[ShadowFox]

You guys think i'll be able to pop one of the new Tbred's in mt kt266a mobo sometime this summer? I'm thinking yes, but i'd like to hear you guys' thoughts

 

[AGodspeed]

You guys think i'll be able to pop one of the new Tbred's in mt kt266a mobo sometime this summer? I'm thinking yes, but i'd like to hear you guys' thoughts.

I'm thinking yes as well, however it depends if the Thoroughbred's voltage requirements aren't a problem. I believe (and I'm not 100% on this, but I'm pretty sure) that the Tualatin PIII/Celeron .13 micron processors require either a BIOS update or some type of voltage regulator to be able to function on i815 boards (these boards were not "meant" for .13 micron processors, but they do work).

I'm thinking Thoroughbred will work well in a KT266A board because the bus subsystem will be the same (namely a 133MHz FSB). Otherwise, the only problem I see is a voltage issue.

 

[FatMan42]

You guys think i'll be able to pop one of the new Tbred's in mt kt266a mobo sometime this summer? I'm thinking yes, but i'd like to hear you guys' thoughts.
...a recent AT thread.

 

[KenAF]

<< Really, what roadmap are you looking at? AMD's official stance (according to their last conference call) is that they will make Thoroughbred available during Q1 of this year. It is not inconceivable that we see 1.8GHz Thoroughbred processors by April (which is in Q2 btw). >>

Did you actually watch the presentation?

AMD Roadmap From November 2001 Annual Analysts Meeting

They made pretty clear that the first Thoroughbreds would see release by the end of 2Q. You can watch the presentation right here:

November 2001 Analysts Meeting Presentation

 

[AGodspeed]

Did you actually watch the presentation?

AMD Roadmap From November 2001 Annual Analysts Meeting

They made pretty clear that the first Thoroughbreds would see release by the end of 2Q. You can watch the presentation right here:

November 2001 Analysts Meeting Presentation

Except that you're interpreting it wrong. According to AMD, they will be shipping Thoroughbred processors sometime in Q1 (March?) and full availability will start at the beginning (as indicated in the links you gave me) of Q2, or April, as I said before.

So, since AMD also reveals that they will release a 1.8GHz 2200+ .18u Athlon XP in Q1 sometime, it's not that far fetched at all to expect that same speed 1.8GHz Athlon in the form of .13u sometime in April, if not early April.

I still don't understand how you interpret AMD's graphs as saying "the first Thoroughbreds will therefore see release by the end of Q2." The only reason AMD is presenting Thoroughbred as "throughout Q2" is because the very next quarter will see .13u SOI Barton Athlon XP processors. Otherwise, AMD would have extended the Thoroughbred bar into the end of Q3. But they don't indicate that anywhere obviously.