When desktop Hammer is coming and try to gues its name

[Trigor]

Does anyone has a idea when we can expect Atholon Hammer. We know it will bee at the end of the year but can anyone be more specific. And wich name you think will AMD assaign to it.

 

[paralazarguer]

It will carry the name Athlon. The official story is that it will be availabe at the end of this year only to major OEMs but it will not be availbe to consumers until 2003. Hoever, every major AMD processor release has always been late by about 2-3months. Thouroughbred was supposed to come out in March and many tech sites even confirmed a launch date. Three months later....With AMD there's no way to tell. I'm rather pessimistic and I say march of next year if they rework the memory controller which they will have to do to stay competitive (they would have to mass produce now to release at end of year) but the official line is that it will be available to consumers in 2003.

 

[broadwayblue]

where the heck are the thoroughbreds? weren't they supposed to be out months ago? it's july and as far as i know there's only one version available. amd might be in some trouble if they don't get their act together. intel is zipping along with new chips and leaving them in the dust.

 

[Duvie]

I agree...Where is the 2300+,2400+, and the 2500+????

I mean amd needs to get those 3 cpus out, plus fit barton in there somewhere as a hybrid of what, who knows, and then move on to the hammer...I don't see how it all fits personally...something is up and something has to five.

I bet the oems don't get them to q1 2003 and consumer start seeing reviews in late q1 2003 but can't get them until q2 2003....

Like mentioned above they have been tardy over the last 6-9 months with all their processors....

If they wait too long the hammer wont have the lead many think it will have....Intel will be to 3ghz by years end and maybe even shifting to the 166fsb chips by then, and who knows where into q2 2003....

 

[paralazarguer]

Not only may Intel be on the 666 FSB but SIS will provide a dual channel DDR333 chipset to compliment that bus perfectly. We may also see hyperthreading, a 1mb cache, and .09u. That's what Intel can produce by Q1 of next year. They won't, though, if they can sit on what they have now and make as much money from that. As many tech sites have pointed out Intel could easily release a 3ghz chip today but they haven't since their 2.53 is tops and they can get just as much from it.

 

[Duvie]

Very good point!!!!

Rumor had it, that the hammer was going to debut with a pr rating of 3400+....Does that sound familar??? If it is true then INtel has a target in site of where it needs to be to conteract a hamer launch....I think they will be 3ghz plu and 666fsb but not likley to .09 process yet or 1mb cache in the desktop model....Dual ddr chipset for 333mhz ddr will be arriving at that time....


Like I said...I think the hammer will be delayed to q2 at the earliest...

 

[gate]

But -- there are some folks who need 64 bit math, who don't need the absolute newest memory interface. I think AMD is not going to put the highest priority on redesigning the memory interface, as the newest thing there is always a moving target. I believe they will release desktop Hammer as soon as the current CPU speed issues are handled. Those issues are definately being addressed right now.

I'm guessing that you will be able to buy desktop Hammer in December 2002. The Opteron chip, for servers, is another matter. It might go until Q2 2003. I'd expect a second version of the desktop Hammer later in 2003, with a newer memory interface. This isn't being shown on AMD's current roadmaps, but I think it will become a necessity.

They need to get the desktop Hammer out as soon as they can, to prove their 64 bit capability assertions. They have advertized this thing for too long, and public expectation is too high to slip it into next year.

 

[paralazarguer]

I don't know what you mean by "some people need 64 bit math" 64 bit computing is 5% slower per clock than 32bit computing. It only offers the benefit of addressing more than 4GB of memory. Only useful for servers and databases. Please read this excerpt from another thread. BTW Intel has an x86-64 CPU but it's not known if it will even see the light of day.
"piasabird, none of what you said makes any sense. For most applications 64bit computing will be slower than 32bit. "64bit game out that really requires the 64bit bus video card" - that phrase makes no sense. It's jibberish. In fact, the x86-64 architecture is not a true 64bit architecture as even in 64bit applications it will use 32bit word lengths to negate some of the inefficiencies of 64 bit. The myth that 64bit is faster was propogated by the game industry but it's simply not true at this point. It's main benefit is that it can address more than 4GB of memory which is very useful for database management. In today's applications for most users, it's nothing more than marketing I'm afraid. When the Hammer is released in the Athlon encarnation it will be used 99% of the time in 32bit mode and will still show 20-30% more efficiencies over the Athlon XP clock for clock. If you would like to learn more about this architecture I would STRONGLY suggest that you read THIS. as it's clear that you're very misinformed about it."

 

[paralazarguer]

"I'm guessing that you will be able to buy desktop Hammer in December 2002."
That's a completely false guess. Word from AMD is that major OEMs will see a few select chips by the end of this year but that they will not be available to consumers until Q1,03. It will likely be pushed back by at least 1-3 months as every single AMD processor ever has. what makes you think this one will be any different?

 

[Athlon4all]

I really think that if Windows XP Service Pack 1 features x86-64 support, we will see Athlon 64 be the name for Hammer. That really would be a golden marketing oppertunity for AMD to give the teir-one OEM's a big reason to go with AMD (ie Dell). We'll see. and it will be shipping to OEM's by the end of the year, with retail avialibility in Q1-2003.

 

[paralazarguer]

The link that was referred to in the second to last post can be found Here You should definitely read it.

 

[gate]

While it is true that there is no purpose, at present, in converting games to 64 bit computing, and that it takes a tiny bit longer to do 64 bit math than 32 bit math, just as for present integer versus long integer math in most processors - there are very legitimate needs for 64 bit math. Yes, for the database folks, and for any others needing large memory footprints, large mamory space addressing is a real strength of the Hammer processor. But some of us need the 64 bit math. In fact, my only reason to purchase a Hammer in the near future is to get the 64 bit math.

As for when : Two AMD reps in the same week said opposite things. One said that the processor will not be out to the public before Q1 2003. The other said that the processor is on schedule for Q4 2002 public release. I'm guessing that if you want one, you'll be able to find the processor on pricewatch.com by the end of 2002. Just what you might have to pay in the first month of release is quite another matter. I haven't seen any solid price estimates from AMD yet. I imagine that it will start high, and come down once they see what the demand is.

I don't expect that Dell or the other systems houses is going to be advertizing a $1600 Christmas special with a Hammer processor. But I do expect to find at least one motherboard manufacturer, and at least one CPU supplier, that could supply the necessary parts to build an evaluation system by Christmas.

 

[paralazarguer]

Guy, there is no way that Dell is going to sell Hammers. There's some important info for AMD fanboys in today's Tomshardware article. I suggest that you give it a read:
http://www.tomshardware.com/smoke/02q3/020704/index.html

 

[Athlon4all]

I agree that there is no real world benefit for 64-bit computing but the publidc sees 64-bit and says "WOW!!!! that must be 2x performance." Thats why I can see a marketingh oppertunity, but I dunno.

 

[Duvie]

Thanks for the read!! One of the better things I have read out of that shithole site...Still think most of his reviews suck!!!


BUt the author hit it on the head!!!!

Amd needed to string up consecutive successes after the athlon, tbird, and xp...however the obvious problems and delays switching to .13 with the t-breds has allowed intel to flex the muscle of the longer pipeline to ramp the distance so far amd's own pr rating can't catch it...So are they going to change the pr rating to get back up there??? Maybe with the bartons!!! May be a big mistake though....

As for the hammer being the saviour...I have to agree again, the marketshare is slipping and all the fan base here in the enthusiast market wont buy enough hammers to justify 1/100 of the money spent to design it. They need to wine and dine the corporate world. That is who is going to save amd not us here. Many ppl forget that the hammer is likley to come out in the 300 plus range when desktops come out....opteron is likley going to be higher since it is aiming for the server IT market..

 

[Soulkeeper]

i swear tom is drunk when he writes some of his articles
it's kinda hard to read sometimes

 

[paralazarguer]

He's german

 

[mechBgon]

It will be interesting to see how fast Hammer is at SETI, which likes low memory latency, lots of cache and a good FPU. Currently the Pentium4 and AthlonXP/MP families are both quite fast at SETI@home, but the real heavy hitters for SETI are CPUs like the Alphas (leaving price out of the equation, of course). A 64-bit-optomized SETI client on x86-64 could be a real eye-popper.

I'd also be interested in seeing what Hammer can do in the realm of 3D modelling, animation and CAD, where it will carry over the strengths of the Athlons (strong FPU) and also feature SSE2 optomization, an improved memory setup, support for more RAM, and 64-bit support.

As for arrival, AMD might not be on time, but then again, look at Intel's venture into the 64-bit world. They took a few time-outs themselves. Whether AMD can live up to their claim that Intel will have nothing that can touch their Hammer at time of release is questionable, but I guess we'll see. Pricing will be interesting too... will it undercut Intel on a price-performance footing?

 

[paralazarguer]

As the article said, nothing that AMD could ever have planned for the future is as important as the here and now, not to mention the next 6 months. They could lose all of the progress they've made and then some. Don't forget that intel has an x86-64 CPU of their own. When they release the P5, reliable rumor is that the extra 5% silicone needed for x86-64 will be there but disabled just like P4s have hyperthreading disabled. Maybe it will even be ENABLED. Intel also has the sweet .09u ace up it's sleeve and 1 meg l2 cache (they seem to want to keep double the cache of AMD processors) as well as speeds climbing up to 4ghz with current prototypes. The next P4 will have a 666mhz bus and dual channel DDR333 (sis) to compliment it perfectly. Then it will go to 800mhz. The hammer will be stuck with DDR333 single channel and a 400mhz bus. Hmmm....I think the hammer has a lot of potential but I think that Intel has really planned the next few years out well....good times ahead!

 

[mechBgon]

From XBit Labs, whom AnandTech Ace Reporter NFS4 says is one of his reliable sources:

Friday, June 21, 2002

No Yamhill to Come [1:19 pm] Rat

According to The Inquirer, Intel was going to respond to the upcoming ClawHammer processor from AMD with some product based on x86-64 architecture. This project was known as Yamhill and was intended solely for competition with hammer family. However, the company seems to have changed its plans now, because The Inquirer claims they canceled their x86-64 Yamhill.

At the recently held meeting in New-York, the head of Intel Company announced that they will not have any 64bit processors capable of performing the existing 32bit code. Instead Intel will focus on improving the IA-64 processors, because they don?t feel like giving up what they have been developing for 10 years. At present Intel is testing Madison samples, which should later come to replace McKinley.

This announcement of Intel?s sounds more than strange, because AMD Hammer is destined to be a success from the very beginning due to its architecture. This means that Intel can lose a pretty tangible part of the desktop PC market if they have nothing to strike back, because their Pentium 4 will never be able to run 64bit code, no matter how fast it is.

We would like to mention separately the following. Some analysts claim that the introduction of registers similar to those used in AMD Hammer processors could increase the Prescott die size only by 5%. However, the launching of Prescott-Yamhill by Intel could help the competitor to promote x86-64 a lot, because the software developers couldn?t disregard such a nice couple.

As we see, Intel decided to help neither its competitors, nor itself, hoping that the software developers will not react in any way to the innovations made to the CPUs from a small Sunnyvale Company. All this reminds us of 3DNow! and Quake3 game, which showed the world what high-quality SIMD instructions could do. Two years later Intel introduced Pentium III processor with the Streaming SIMD Extensions set.

And I think this is unfortunate for the users of either platform, since it splinters what appears to be a versatile, backwards-compatible approach to expanding the capabilities of computers. Intel's been working on their flavor of 64-bit longer, but if it's really as easy as tacking 5% more silicon (not silicone btw), then heck, they should do it.

 

[paralazarguer]

Yes, and as the SAME article says, they're going to probably leave the silicon there but just have it deactivated like hyperthreading is on the P4. As the SAME article says, they aren't really cancelling the processor. They're just saying it to support Itanic.

 

[imgod2u]

If Intel does have an x86-64 in progress, it will probably show up in the Xeon family. That is the only viable market for x86-64. The market does just demanding enough processes (integer-wise) to need 64-bit integers while still being a mainly x86 market. Programmers for the workstation market are use to x86 and would probably not want to let go.

 

[alexruiz]

Originally posted by: 7757524
The next P4 will have a 666mhz bus and dual channel DDR333 (sis) to compliment it perfectly. Then it will go to 800mhz. The hammer will be stuck with DDR333 single channel and a 400mhz bus.

Well, obviously you haven't read the hypertransport papers completely.... there is no "bus" as exists today in CPUs using hypertranport. An estimate of the "equivalence" in a hypertranport bus would be "800 MHz", going as high as "1.6 GHz' just enabling more pins.... there should be a previous thread covering the subject.

Also, remember that dual channel helps IF and only IF the CPU can handle the added bandwidth. The NForce has dual channel DDR, and it barely beats the sis 735......

 

[paralazarguer]

The nforce is not a dual channel chipset. That's a huge misconception. It's actually a very similar technology but not the same. Besides, having dual channel PC2100 on athlon is useless since the system bus on the Athlon can't handle more than 2100megs a sec anyway. The reason that chipset has that much bandwidth is for the integrated graphics. If you want evidence of the speed of dual channel, look at RD ram. When RD ram was single channel DDR beat it hands down. The only way RD was able to get ahead was using dual channel.