ClawHammer to Perform 30-50% Better Than Athlon XP at Same Clock Speed

[majewski9]

I dont think AMD is releasing their .13 micron Athlons since number 1 Intel hasnt really delivered. I mean northwood was suppose to be a major improvement, but the Athlon XP 2000+ is very competitive with the Northwood 2.2ghz. Number 2 they are determining if they should move to a higher bus speed and cache size. It would be a good bet that the .13 micron will have more cache since northwood has 512k. I dont know about the bus speed. If they were going to move to a 333mhz bus then they would probaly release a new chipset supporting that bus until other chipset makers released 333fsb chipsets. I have heard that the VIA KT400 or KT333A will support the 333fsb offically. I do think that most boards today can do the 166fsb with support of DDR333. I think that AMD should move the FSB to 166 double pumped and 512k cache, but they are probaly betting on Hammer and not so much on the Athlon.

The Duron should move to the 266fsb with its .13 micron core as metioned on its roadmap! Does this mean that the Athlon will get and FSB bump as well? Who knows for sure?

 

[NFS4]

<<

<<
No more floppy drive or serial port!
>>



Whether or not you like them, they need to stay there. Whenever you bring up whether or not a floppy is useful it always turns into an argument. The fact that there is an argument means the port should stay (afterall why piss off half of your potential customers just to not add something that people who don't like it don't have to use). Serial ports are also still used a lot (TI calculators still largely use Serial port graphlinks for PC's. Last time I checked the USB links didn't even work on PCS, only Macs).

This doesn't even affect me as I don't use serial stuff and my floppy drive is IDE based, but it would me mindless for them to remove these connections. >>


I haven't used a floppy in about 6 months. I don't even have one in my computer. I don't use serial or parallel ports either. All USB. Hell, motherboards don't come with ISA slots anymore. B/C they're outdated.

With the advent of Zip, CDR, and CDRW, floppies need to be shot dead. No longer do we get AOL diskettes, we get AOL coasters. That should be reason enough right there to see that the market has shifted.

GET RID OF IT!!

 

[SexyK]

One other question I have about this report is, are we talking about on specific x86-64 code? or for everyday 32-bit windows code? because if its a huge boost, only under dedicated 64-bit code, that's not going to benefit anyone for a while at least. Are these 32bit numbers?

Kramer

 

[Rand]

<< One other question I have about this report is, are we talking about on specific x86-64 code? or for everyday 32-bit windows code? because if its a huge boost, only under dedicated 64-bit code, that's not going to benefit anyone for a while at least. Are these 32bit numbers?

Kramer >>


Strictly 32bit performance.
There is really nothing to appropriately compare the Hammer's 64bit performance to as it will be the first 64bit X86 microprocessor.



<< Number 2 they are determining if they should move to a higher bus speed and cache size. >>



How do you figure AMD is still deciding on how to implement the K7 in .13u? They have been stating consistently for approaching 1.5 years now that the Thoroughbred will be nothing more then a dumb shrink of the existing Athlon core.
If one is to believe AMD then both the Throughbred and Barton cores will be identical to the existing Palomino core.
No larger cache, and no faster FSB or anything of the like.

 

[SteelCityFan]

I'm not a big fan of AMD's rating system since they are the ones making them up... but if I remember right, they base it on the IPC of their chip vs the IPC of Intel's.

When Intel goes to the 533Mhz System Bus and PC1066 RDRAM will they alter their little multiplier since clock for clock the newer P4's will run faster? They didn't alter it when the Northwoods came out which gave the P4 about a what 8-10% boost. See where I am going with this? This is why I don't like their model number system. I see here that they are going to give it a rating of 3400..where are they getting this number. Why stop there... why not give it a rating of 6000. How can they plan a rating so far ahead?

 

[nortexoid]

now, is that 30-50% better when running a 32-bit OS and 32-bit applications, or 64-bit OS and 32-bit applications, or 64-bit OS and 64-bit applications????

 

[AGodspeed]

I suggest you talk to those friends again, to go from initial silicon to full release in less then 1 year would be an incredible feat indeed.
Also, to quote what someone else said that is employed by an unnamed microprocessor manufacturer...
"You can't debug, validate, and attain yields at the same time. To debug and validate consists of changing the process. How can you work on improving yields when nothing is consistant?"

Yup, I misspoke. Reread what I said before in my post before that statement: it'll take AMD another ~ 6 months of debugging, validation, etc. before final ClawHammer's are sent to OEMs and through the retail channels. This would give AMD enough time to work on getting acceptable yields to ramp up Claw and Sledge Hammer supply and it would be enough time for OEMs and such to have ClawHammer systems ready to go.

 

[human2k]

I dont see how you guys are getting so hyped about a workstation cpu unless you own an itanium rig or something................what i care about is thoroughbred! But for now the P4 1.6A is the cpu of the hour.

 

[Bluga]

<< In my opinion, anything under 10% is barely noticible. >>



It will still show in the benchmarks.

 

[formulav8]

<< I'm not a big fan of AMD's rating system since they are the ones making them up... but if I remember right, they base it on the IPC of their chip vs the IPC of Intel's.

When Intel goes to the 533Mhz System Bus and PC1066 RDRAM will they alter their little multiplier since clock for clock the newer P4's will run faster? They didn't alter it when the Northwoods came out which gave the P4 about a what 8-10% boost. See where I am going with this? This is why I don't like their model number system. I see here that they are going to give it a rating of 3400..where are they getting this number. Why stop there... why not give it a rating of 6000. How can they plan a rating so far ahead? >>




AMD has stated many times that the rating system is being compared to the TBird NOT the P4.

 

[AGodspeed]

Just finish out your math though. The AMD PR rating works quite well when compared to a P4 Northwood at 400 MHz fsb. Thus a 1800+XP is really really close in speed to a 1.8A GHz P4.

Not really. An AXP 1800+ (1.53GHz) is more competitive with a 2.0GHz Willamette, in fact a lot of hardware review sites have concluded that the 1800+ is faster than a 2.0GHz Willy.

And with Northwood, an AXP 2000+ is on average ~ 10% than 2GHz NW since the AXP 2000+ is on par with a 2.2GHz NW.

The 533MHz FSB might just take out all the conservativeness of AMD's model rating system.

 

[AGodspeed]

<<

<< I'm not a big fan of AMD's rating system since they are the ones making them up... but if I remember right, they base it on the IPC of their chip vs the IPC of Intel's.

When Intel goes to the 533Mhz System Bus and PC1066 RDRAM will they alter their little multiplier since clock for clock the newer P4's will run faster? They didn't alter it when the Northwoods came out which gave the P4 about a what 8-10% boost. See where I am going with this? This is why I don't like their model number system. I see here that they are going to give it a rating of 3400..where are they getting this number. Why stop there... why not give it a rating of 6000. How can they plan a rating so far ahead? >>

AMD has stated many times that the rating system is being compared to the TBird NOT the P4. >>

AMD's model rating system is meant to be compared to the Pentium 4, not T-birds. Supposedly AMD said the model rating was against a T-bird, but it's quite clear now that it's against a Pentium 4.

 

[MadRat]

PR ratings for the AMD XP-family are based SpecInt scores of the T-bird, and not anything to do with the Pentium 4. The Pentium 4 is a trademark and one cannot go around advertising someone else's trademark. This is why nobody really does direct comparisons anymore, for fear of reprisals.

 

[Rand]

<< I'm not a big fan of AMD's rating system since they are the ones making them up... but if I remember right, they base it on the IPC of their chip vs the IPC of Intel's.
>>



It is based off of the performance off the Thunderbird core, in a suite of typical office/games. The full information is on AMD's site.




<< now, is that 30-50% better when running a 32-bit OS and 32-bit applications, or 64-bit OS and 32-bit applications, or 64-bit OS and 64-bit applications???? >>



32bit OS, 32bit applications. We're talking strict X86 32bit code here. Nothing more then what we've all been playing with since the 386'.



<< I dont see how you guys are getting so hyped about a workstation cpu unless you own an itanium rig or something >>


We are referencing the ClawHammer which is targeted towards the home desktop market. The SledgeHammer is the server/workstation variant.




<< Yup, I misspoke. Reread what I said before in my post before that statement: it'll take AMD another ~ 6 months of debugging, validation, etc. before final ClawHammer's are sent to OEMs and through the retail channels. This would give AMD enough time to work on getting acceptable yields to ramp up Claw and Sledge Hammer supply and it would be enough time for OEMs and such to have ClawHammer systems ready to go. >>



So de-bug and validation complete in September, six months from now?
That would be an incredible pace if they managed it.... I'll cross my fingers for them,, but I doubt they'll manage it.
And even then it only gives them 2 months to work on the manufacturing process and build up enough supply for release.
That most definitely isnt viable.




<< AMD's model rating system is meant to be compared to the Pentium 4, not T-birds. Supposedly AMD said the model rating was against a T-bird, but it's quite clear now that it's against a Pentium 4. >>



People may choose to make such a comparison, but it is based off of the performance of the Thunderbird core in typical desktop application usage.
ie. Quake3, Sysmark2001 etc etc.
The full details are on AMD's site for anyone that cares to peruse it.

 

[SteelCityFan]

<<
It is based off of the performance off the Thunderbird core, in a suite of typical office/games. The full information is on AMD's site.
>>



Now, where is the deed to that bridge I had up for sale.



AGodspeed

The reviews I have seen do not show the 2000XP on par with the 2.2 NW... they show the NW winning the vast majority of the tests.

 

[MadRat]

<<The reviews I have seen do not show the 2000XP on par with the 2.2 NW... they show the NW winning the vast majority of the tests.>>

Well, the vast majority of websites seem to point one or the other as faster, depending on which manufacturer they garner the most support. Its very hard to find review sites that are not skewed by the promise of exclusive releases at the cost of NDA's and fair (read favourable) opinions.

I tell you what, start a new thread and make that statement. Just humor us all and provide links. We'll all enjoy the laughs one way or another.

 

[Rand]

<<

<<
It is based off of the performance off the Thunderbird core, in a suite of typical office/games. The full information is on AMD's site.
>>



Now, where is the deed to that bridge I had up for sale.



AGodspeed

The reviews I have seen do not show the 2000XP on par with the 2.2 NW... they show the NW winning the vast majority of the tests. >>



I guess it depends which sites reviews you favor. Personally, I favored Lost Circuits, AnandTech, Tom's Hardware, and Aces Hardware reviews to be the most reliable... and overall from what I've seen I'd say the AthlonXP 2000+ + KT266A is roughly on-par with the 2.2GHz Northwood + I850. Each platform has it's distinct advantage, but overall at stock clockspeeds I'd put their performance as being extremely similar.

 

[BD231]

Sorta like the K6 core's supposed superior FPU performance over the P6 core. Or the K6-2 300's comparable FPU performance to the PII 400MHz?

I wasn't aware of such claims, those statements were made when I was to young to care about computers. I'm sure AMD learned their lesson in making false statements though. When I made my statement:

If AMD says it will be 30 to 50% faster than an XP at the same clock then you can trust that, they have been really good about delivering what they promise.

I was referring only to the Athlon line. I don't remember AMD ever lying about performance on their K6-2 and K6-3 procs, in fact they pretty much kept their mouths shut for a long time. In this age, if someone promises a 30 to 50% speed jump and dosent deliver, that would cause a great amount of people to loose faith in the company making that statement. I think AMD is a little smarter than that, they wouldent throw around such figures if they knew they couldent deliver at least something close.

 

[sandorski]

Yup, the k6-2 300=PII 400 claim was when using 3Dnow, which was the case in the highly 3Dnow optomized Quake 2 exe provided by AMD. It is also my understanding that the K6 fpu unit was superior in many ways(how much performance that it would acheive, I dunno), but no one ever chose to optomize their code for it.

 

[SuperSix]

I'll believe it when I see it. AMD is as good at spin and FUD as Intel.

 

[Barnaby W. Füi]

am i the only one who thinks its cool that in the pic of the linux hammer demo, they were using blackbox?

 

[x86]

<< now, is that 30-50% better when running a 32-bit OS and 32-bit applications, or 64-bit OS and 32-bit applications, or 64-bit OS and 64-bit applications???? >>



I think he has a point. Who knows if this can be another one of Jerry's flabergastic controvertial statements. He could most probably mean that on 64 bit applications it will be 30-50% faster than a P4 running 64-bit applications (using a emulator perhaps). Jerry says things to make his company seem stronger, he does this a lot on CNBC.

-x86

 

[FishTankX]

<< I'd just like to butt in here and point out flaws with extending a performance rating system to architectures and MPUs at frequencies not due for quite some time. Namely, average memory access time gets worse as clock frequency gets higher, and thus performance does not increase linearly with clock speed. It is also not possible to assume that performance will scale at the same rate with two different microarchitectures at different clock frequencies. It is thus FAR too early to even try to compare the performance of MPUs due 9-12 months from now based simply on an existing performance rating. >>



Uh.. there's a flaw in your logic. AMD's Hammer has an integrated memory controller, which reduces memory access times.

20NS memory access delays..+
6ns*2 for CAS2 or 3 for CAS3=~43NS

By contrast, the Northwood gets like..
:Q 200NS :Q
4times memory acccess... wow!
Combine that with more execution units than the P4, and you have a competitive processor regardless of memory controller latency issues, since the memory controller is on die..

 

[FishTankX]

<< I'd just like to butt in here and point out flaws with extending a performance rating system to architectures and MPUs at frequencies not due for quite some time. Namely, average memory access time gets worse as clock frequency gets higher, and thus performance does not increase linearly with clock speed. It is also not possible to assume that performance will scale at the same rate with two different microarchitectures at different clock frequencies. It is thus FAR too early to even try to compare the performance of MPUs due 9-12 months from now based simply on an existing performance rating. >>

 

[AGodspeed]

I think he has a point. Who knows if this can be another one of Jerry's flabergastic controvertial statements. He could most probably mean that on 64 bit applications it will be 30-50% faster than a P4 running 64-bit applications (using a emulator perhaps). Jerry says things to make his company seem stronger, he does this a lot on CNBC.

Actually, Jerry's estimation is pretty accurate according to Paul Demone.