Should I Be Excited About Hammer?

[Poritz]

I was wondering what clock speed the new Hammer would début at. But my main question is whether or not people think they are excited about it. Will the Hammer perform so much better than other chips that many people outside the focused tech community want one? What do you think the price will be? I want to know what people are thinking of it because I am wondering if I should be getting excited about it.

Poritz

 

[Poritz]

It is always like this. Nobody ever wants to respond to my posts.

 

[BuddyAtBzboyz]

I think it will kick some serious ass as to what it will debut at my guess would be anywhere from 2800+ to 3200+ but nobody who hasn't signed an nda knows so we will just have to wait.

 

[AmdisGreat]

Well, I am an AMD employee and even I don't know what speed it's gonna debut at......

 

[mastertech01]

If you had a pair in hand with a mobo right now you would be real excited.. LOL.. no sense getting excited till they are avaliable and you see what Intel trumps with.

Meanwhile try something interesting with what you have.. you might be surprized what you can do with it.

 

[WarriorX]

Hey AmdisGreat, how do you like working for AMD? I live about 5 miles from AMD HQ in Sunnyvale, and I have "some" connections there from high up. I really want to work there after getting a MS in Computer Engineering. I am going to attend Santa Clara Univ. and want to get a summer internship there after Soph, Junior, & Senior years. Maybe you could PM me your email address and we could talk? Or AIM/MSN username? I would appreciate it.

 

[Duvie]

Wasn't the talk about the hammer being 20% greater then same speed palomino chip??? Maybe I am wrong but I know I have read that number a couple of times....

20% is not the end all and by the time it gets here intel may have 20% plus leads in most the categories with 2.8ghz coming out end of august and 3ghz likely around time of hammer....


I am getting less and less excited about it everyday as my p4 chips performs so well...I likley will upgrade to a 2.4b later in th fall if the prices drop enough and may alreay be at 3ghz oc'd....That hammer will have to be really impressive +10% more then intel for me to switch ou all new components....


NOt to mention prescott 166fsb chips and dual channel ddr mobos...I mean how much is the hammer really going to be over the top of that???? I think most of us will be surprised one way or the other....

Too much hype for a product we saw questionable german site give the only benches to date on the hammer.

 

[AmdisGreat]

Working in AMD is great. But I guess, like any big tech company , the work environ varies from group to group. Like any company on the verge of coming out with a new flagship product , the energy level is on the higher side right now.
Of course we are anxious to know about Hammer performance much more so than you guys. Our future depends on it!
WarriorX, you should add your email to your personal profile so that I can pm you.

 

[meson2000]

The benchmarks that were leaked of the clawhammer earlier this year (around May) showed that it had
a 40% higher performance than a equally clocked Athlon MP. Keep in mind that the hammer
was at 800mhz at that time. I recently read that IBM demoed DB2 on an opteron system clocked at 1.1ghz
a couple of weeks ago. So at least AMD is ramping up the mhz. I don't think AMD will have
a problem releasing the clawhammer at the end of this year to OEMs at 1.6ghz like many sites have
stated. People like us should be able to pick one up in late January 2003. I believe it will carry a rating of 2800+ and have 256k of L2 cache.
At least that is what most of the stuff I have read said.

 

[nemo160]

if those opterons were sledgehammer cores that should bode well for how the clawhammer core will ramp as the sledge core is signifigantly more complex (more hypertransport links for interprocessor transfers)
also since the first 800 mhz samples were locked there to prevent premature benchmarking, not because the chips wouldn't run any faster 1.6 ghz + doesn't seem unlikely

 

[alexruiz]

Well, I am extremely excited about the hammer. If my guess about market segmentation is correct, AMD will position dual hammers as high end desktops.... because of the glueless SMP capability, courtesy of hypertransport.


http://forums.anandtech.com/messageview.cfm?catid=28&threadid=827893

Check this thread, there are several interesting comments. In fact, there is a challenge set, the Intel lovers will get a granite bay + P4, and some AMD boys will get a hammer machine of our choice for the same money.... benchmarks?? Anything, and only brute force will prevail...... Thas is how much I trust the specs of the hammer.

 

[mechBgon]

I'm pretty excited about Hammer's potential. Can't wait to see what kind of rendering/animation power it has, with its combo of strong FPU, low memory latency and SSE2. If nothing else, I'd expect it to deliver lots of performance for the dollar/yen/etc.

Amdisgreat, if you're responsible for passing on customer input at all, let it be known that your retail heatsink/fan units are generally regarded as being too loud. Could they switch up to some lower-rpm 70mm fans or something?

 

[BuddyAtBzboyz]

I think the mainstream clawhammer will have 512 L2 and the Opteron will have 2x that. However this is just my opinion not fact.

 

[MajinVegeta]

Hammer wont be blessed with a huge cache. AMD intergrated the whole northbridge into the core of the chip, ya that chip thats 1/2 the size of your processor with the passive heat sink on it that stuck to you MB.

Hammer isnt really anything new just a reworked Althon for the 3 or 4 time.

 

[Rainsford]

Originally posted by: MajinVegeta
Hammer wont be blessed with a huge cache. AMD intergrated the whole northbridge into the core of the chip, ya that chip thats 1/2 the size of your processor with the passive heat sink on it that stuck to you MB.

Hammer isnt really anything new just a reworked Althon for the 3 or 4 time.

Uh, or a completely new chip. Onboard memory controller is just "reworked"? Sounds like one of a long list of big changes to me.

 

[SickBeast]

i'm worried about hammer, simply because it's trying to be all things to all people (64bit and 32bit combined). i think that a pure 64bit chip from intel will be much faster clock for clock. it's a shame that amd can't compete with intel right now. i bought an athlon 1333 last summer when they had intel soundly defeated, but this year the tables have turned. the p4 is where it's at right now, plain and simple. if the hammer is released at 1600mhz like people here are saying, and if it's 40% faster than the athlon xp, that makes it equivalent to an athlon xp running at 2400mhz, which would be about the same as a p4 3000, i think anyways. i don't know how to convert to amd's pr ratings. so...it looks like at launch it will be as fast as intel's top 32 bit chip, and it will have 64 bit capability to boot. i just hope that software is made for the amd version of 64 bit...

 

[Swanny]

Hammer isnt really anything new just a reworked Althon for the 3 or 4 time.

Obviously you don't quite know your stuff. This is a completely new processor. Hence K8 instead of K7.

 

[Athlon4all]

Originally posted by: Rainsford

Originally posted by: MajinVegeta
Hammer wont be blessed with a huge cache. AMD intergrated the whole northbridge into the core of the chip, ya that chip thats 1/2 the size of your processor with the passive heat sink on it that stuck to you MB.

Hammer isnt really anything new just a reworked Althon for the 3 or 4 time.

Uh, or a completely new chip. Onboard memory controller is just "reworked"? Sounds like one of a long list of big changes to me.

Well, there is a pretty decent arguement that the ClawHammer isn't a true 8th generation CPU. It has much in common with the Athlon, but rest assured, Hammer has been developed independant of the K7, and thus any things that are similar between the K8 and the K7 are simply because they are the best option for AMD.

In any case, Hammer has several significant ehancements that should allow for higher clock speeds and a significant increase in IPC(without even taking into account x86-64). SSE2 support, on-die memory controller that will over time likely get even higher bandwidth utilization numbers than PC1066 on the P4, while having lower memory latency than any other CPU+Chipset solution in the Desktop CPU market, 12 stage pipeline (K7 has 10), 512K of L2 Cache (with a striong possibilitiy of the L2 bus width widened to 256-bits (athlon is 64-bit)), and more for the Server/Workstation SledgeHammer.

The potential is there for ClawHammer, but only time will tell whether it will be a hit or hype.

 

[Booster]

I don't know whether you should be excited or not, but the Hammer is a very important step in overall CPU development, IMO. Three things are the main issues here:

1 Power consumption -- how much juice will the Hammer draw? Probably a ton or so, would be good if 500W PSU wasn't a must.

2 Thermal protection -- will clock throttling be built into the core? Or will it have a PIII-like shutdown feature only? Or maybe no thermal shutdown circuit at all?

3 How hot will it run? Probably would require a huge heatsink with a l o u d fan.

All this makes me think that the P4 would still be a better alternative b/c of these possible issues.

 

[MajinVegeta]

its just like the K6-II to K6-III.

They slightly modify the core slap a new name on it and they have "new" core, when its clearly based off the old one.

Its still faces the same limits the old one does.

http://www.geek.com/news/geeknews/2002Jul/bch20020705015267.htm

AMD is still pushing High IPC by forcing the chip to do more work in the same time(or less per step) increasing th burden on the chip this showing hammer is still limited by the high IPC nature of the K7 core.

x86-64 op codes takes up 2% of the die space on the chip so why is it 2x the size of the AXP, and only features 265K of L2 cache. They added the northbridge, and 3 HT links, thats a MB part so now Intels Prescott will be smaller than Hammer and not have the extra cost of SOI and SI-28, and be made on the smaller 90nm process and larger 300mm wafers.

 

[alexruiz]

Isn't funny how all the *ntel fans that attack the hammer blindly are the ones who haven't read the papers???

Go read the hammer papers, and then come back and argue!! Read about HT and then ask about "FSB".... :|

 

[MajinVegeta]

Like this one that show 3 seprate Ht links?
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/MPF_Hammer_Presentation.PDF

oh and HT isnt a FSB because each transaction is put on a list then excuted so it takes even longer to access the DRAM hence 20 Entry MCQ.

So now you graphic card if you use the on die northbridge just like amd paper says its transaction will be queued until a spot opens up for it, and to use the CPU it will have to go to the SRQ.

now when you go to MP sytems it gets even uglier, with DRAM fragmentaion and massice queue delays.

 

[mechBgon]

After making CPUs for a reasonable length of time, like 20 years or so, I somehow doubt that either AMD or Intel are naive enough to design themselves into a corner they can't get out of. Intel needed a little time to get the P4 platform on course at 512kb cache and 533MHz bus, for instance, but now it's working pretty well. Give AMD their chance too.

 

[Rainsford]

Originally posted by: MajinVegeta
its just like the K6-II to K6-III.

They slightly modify the core slap a new name on it and they have "new" core, when its clearly based off the old one.

Its still faces the same limits the old one does.

http://www.geek.com/news/geeknews/2002Jul/bch20020705015267.htm

AMD is still pushing High IPC by forcing the chip to do more work in the same time(or less per step) increasing th burden on the chip this showing hammer is still limited by the high IPC nature of the K7 core.

x86-64 op codes takes up 2% of the die space on the chip so why is it 2x the size of the AXP, and only features 265K of L2 cache. They added the northbridge, and 3 HT links, thats a MB part so now Intels Prescott will be smaller than Hammer and not have the extra cost of SOI and SI-28, and be made on the smaller 90nm process and larger 300mm wafers.

I don't pretend to know what AMD is doing with the design, and I doubt you know either (really, all of us are just guessing). But I do know that a high IPC is not a "limitation". More IPC is good, just because Intel killed the IPC of the P4 doesn't mean it's a good design strategy. It's one approach, yes, but AMD may have another approach that yields better results.

But my main point is we don't know! Making all these pronouncements makes you look foolish.

 

[BuddyAtBzboyz]

Rainsford is right we are all just guessing but I think MajinVegeta is just stretching. Of course the k8 has some of the k7 in it just like the p4 has some of the p3 in it. If you wanted to make a truely new cpu you would have to fire your whole team and hire new people. Of course they are going to reuse stuff that works well. If it ain't broke don't fix it!