AMD gains on Intel, sets Hammer speeds

[AGodspeed]

nm

 

[Wolfsraider]

forgive me i'm stumped here: amd is releasing hammer >will debut at 2GHz or higher>intel is at 3 gigs>at release time<

now for the questions

is amd rating a 2000 mhz chip at 3400?

am i lost here please explain isn't the hammer = opteron?

please straighten me out here

also here is this

<If they can get Thoroughbred out and it can make it to 2000MHz that would translate into a performance rating of 2500+. If they can push that core to 2133MHz that would be 2700+.>

is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating?

i know something is wrong here but please explain it to me

 

[Katana]

<< is amd rating a 2000 mhz chip at 3400? >>



It looks that way.



<< am i lost here please explain isn't the hammer = opteron? >>



Sledgehammer, the server cpu, is going to be called Opteron, the Clawhammer, desktop cpu, is going to have the Athlon name.



<< is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating? >>



Yes

 

[Sid03]

<< is amd rating a 2000 mhz chip at 3400? >>

apparently so.


<< is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating? >>

actually, at 2000mhz the hammer will get a 900 point jump over the palimino.

if i recall correctly, they said that hammer will be ~25% faster than a palimino. so, i have no idea how they come up with a rating almost double that of a palimino (500 extra for an athlon xp, 900 extra for a hammer)

 

[RaynorWolfcastle]

AMD will have a problem in its number scheme in that the numbers are supposed to be comparable to a thunderbird processor of that frequency, I don'r know about you but to me a 3.4 GHz thunderbird means nothing. It will be similar to if intel said this P2 2.4 GHz is equivalent to a P-MMX 4.2 GHz, which is totally insignificant.

AMD has 3 choices at this point:
1) Call an ace of spades what it is and admit that it is comparing its processors to Intel's offerings
2) Release a new numbering scheme that uses a more significant baseline processor architecture
3) Continue with what will soon become a numbering scheme that means nothing.

Should be interesting to see what AMD does with this

-Ice

 

[ElFenix]

i think i posted in the other thread about this, something about me not expecting a 3.4 from intel until Q1. i could be wrong though. pretty sure 3.4 will be pushing the limit of NW.

 

[Hulk]

I based the PR extrapolation on the current AMD scheme of each 100 point PR equaling 67MHz.

This extrapolation may be overestimating for Thoroughbred if it offers no performance increase over Palomino other than clock speed increase. AMD started the PR rating at 1500 for the 1333MHz Palomino. The XP1500+ is waaayyyyy faster than a P4 1.5. Since the difference between the PR rating and actual MHz increases for AMD's current rating system, the fastest chips in the current XP line are more or less comparable to current Northwood offerings.

Now, depending on enhancements to the Thoroughbred core, such as larger L2 cache, Thoroughbred could effectively "recallibrate" the PR rating scale and make the first few T-breds "underachievers" just like the first few Palomino speed grades.

Further efficiency of the Hammer core will no doubt force AMD to recalibrate the PR rating scale for Hammer. AMD has been very honest with the PR ratings for the Palomino line and I have no reason to believe they won't be with Hammer. So, it appears that 2000MHz worth of Hammer will equal 3400MHz of P4, more or less. P4 will most likely be there when Hammer is released.

The good news for Intel is that they have the core, the process yeilds are high, so they can release the faster speed grades on their schedule until AMD pushes their hand. The bad news for Intel is that 0.13u P4 will most likely be topping out at 3400MHz or somewhere around there. If they don't have 0.09u process ready AND Hammer begins to scale they will not have an answer for Hammer, for a while anyway. We saw this when the 0.25u Athlon left the 0.18u PIII in the dust. Intel had to scramble out the P4. From what I have heard there is no P5 just around the corner for this eventuality.

In addition, as P4 clockspeeds scale ever higher they will require ever faster memory subsystems to keep up. AMD on the other hand by going with a more efficient chip is not to constricted by memory subsystem performance. Sure, they will have to have fast memory subsystems for hammer, it just won't be as critical as for P4. Current DDR P4 systems still don't perform as well as RDRAM systems.

All of this of course depends on AMD delivering. If they can't deliver on time then Intel will simply stay a few speed grades ahead, corner the highend market, and have the time to develop an answer to Hammer.

Just my thoughts of course. Should be an interesting year for the CPU. P4 scaling to unheard of speeds and a new core for AMD.

 

[MadRat]

The HAMMER is supposed to run x86-32 code extremely efficiently. Hopefully the 15% speed boost over he K7 architecture is true. I'm having a hard time believing an 800fsb will add a 15% speed boost unless the K7 architecture is extremely cache starved as it is now.

 

[mastertech01]

Official release in Q1 2003? I guess that means we will see OEM versions on the F/s Trade forum by December 2002? LOL

 

[Athlon4all]

<< The HAMMER is supposed to run x86-32 code extremely efficiently. Hopefully the 15% speed boost over he K7 architecture is true. I'm having a hard time believing an 800fsb will add a 15% speed boost unless the K7 architecture is extremely cache starved as it is now. >>

First off, the 800fsb term is very misleading. Second, the 15% (or I've heard 25% actually) comes in 20% of it from the Integrated Memory controller, and 5% from other core enhancements. Now about the 800fsb. The 800fsb you speak of is still the connection between the CPU and the rest of the system, but it does not have the responsibility of maybe the most crucial thing of the fsb on todays CPU's, and that is moving data between the CPU and Memory. So really, a major part of the fsb's job has been taken away from the Hammer's because of the Integrated Memory Controller. One thing that this changes with this, is that today, as memory speed increases, that speed's usefulness would be limited by the fsb, however with Hammer theoretically, a SledgeHammer would be able to fully use for example 5.4GB/ps of bandy of Dual Channel PC2700 DDR, where as the Athlon would not be able to use it at all due to the fsb being limited. Just thought I'd add that.

Now, right now, roadmaps say there will be no 3GHz NW P4 until Q1 2003 until after CH's launch. Now, I see it very possible that Intel will move that up if ClawHammer hits the shelves in Q4 2002, we'll see.

 

[Degenerate]

<< can't imagine an AMD chip debuting at 2ghz, that's just gonna scream. >>



I hope that is with sarcasm because a 2Ghz from AMD now is not big deal.

 

[Texmaster]

My Future chip can beat up your future chip!!!

Oh yeah?

Yeah!

Oh yeah?......

 

[LikeLinus]

<< can't imagine an AMD chip debuting at 2ghz, that's just gonna scream. >>



<<I hope that is with sarcasm because a 2Ghz from AMD now is not big deal.>>

Yeah i feel the same way. 2GHZ is not going to be that much faster than my 1733mhz (1600+ @ 165mhz FSB).


Their PR system is a shame. AMD needs to fire their whole Marketing/PR department and move on.

 

[hungrygoose]

i don't really see a problem with the PR rating since the only people who really pay attention to it is the uneducated consumer, which is who it is targeted at.....i don't care what they call it or wat it's codename is or wat it is compared to.....or any of that crap.......benchmarks tell all for the educated consumers which is wat most of these forums consist of

 

[wefixitalways]

I don't care what anybody says INTEL still RULES and I hope it stays that way. They build the best quality CPU'S

 

[moocat]

I got this in a fortune cookie at lunch yesterday...

"The Hammer line from AMD will be used in some interesting and suprising applications this coming year."

Of course, you can't always put too much stock in what a fortune cookie tells you.

 

[ElFenix]

i got this in a fortune cookie the other day:

"Admire those who succeed, and I earn from their success."


i thought that was a little odd.

 

[Pariah]

That's obviously a typo that should say "learn from their success." Just an errant space after the "L".

 

[spanner]

AMD's PR rating has great potential, I mean if it is used consistently between different CPU's then it will end the PC industry scamming uneducated customers when they say a Duron 700 is faster then an athlon 700 (yes it happens on a daily basis). This would make life so much easier. So far AMD has been extremely honest with their PR ratings (every real world benchmark shows it) and if it stays that way then I give the PR rating my full support. I do believe AMD needs a new extremely cheeky marketing department to compete with intel (All that intel FUD really gets to me)

 

[majewski9]

IM already plannig a hammer upgrade from my Tbird!

 

[Wolfsraider]

oops lol postcount +1 for double posting

 

[Wolfsraider]

<<

<< is amd rating a 2000 mhz chip at 3400? >>



It looks that way.



<< am i lost here please explain isn't the hammer = opteron? >>



Sledgehammer, the server cpu, is going to be called Opteron, the Clawhammer, desktop cpu, is going to have the Athlon name.



<< is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating? >>



Yes >>



and

<< is amd rating a 2000 mhz chip at 3400? >>

apparently so.


<< is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating? >>

actually, at 2000mhz the hammer will get a 900 point jump over the palimino.

if i recall correctly, they said that hammer will be ~25% faster than a palimino. so, i have no idea how they come up with a rating almost double that of a palimino (500 extra for an athlon xp, 900 extra for a hammer)

thanks guys i was lost

i look forward to the hammer also but wish the ratings were easier to understand

lol i know wait for the benchmarks lol

 

[Sid03]

<<

<< is amd rating a 2000 mhz chip at 3400? >>

apparently so.


<< is opteron supposed to be so much faster that it gets bumped an add 800 mhz rating? >>

actually, at 2000mhz the hammer will get a 900 point jump over the palimino.

if i recall correctly, they said that hammer will be ~25% faster than a palimino. so, i have no idea how they come up with a rating almost double that of a palimino (500 extra for an athlon xp, 900 extra for a hammer) >>



wait a minute...i misspoke...

if the above is true, the hammer will have a rating almost three times that of the athlon above the actual clockspeed. (500 vs 1400)

i don't get it.

 

[AGodspeed]

Sanders said ClawHammer will launch at at least 2GHz.

At least 2GHz does not = 2GHz.

 

[CedarTeeth135]

<< 3400 rating >>



Damn, that is double my 1700 Xp. Hopefully they won't be too expensive.