[quote] Because it would be embarassing for Apple to officially say that their machines match mid-range Wintels, and top range prices. <hr></blockquote>
At the current top configurations from Dell and Apple, when you add the Powermac's Dual Processors together, the resulting number is almost as high as the Dell's 3 ghz flagship. Apple almost doesn't need to be so adamant about this Mhz Myth thing, seeing as they are [only] 200 mhz behind. While that gap is pretty big, add Altivec and the DIMension gets blown away. The Powermac, in a sense, is advancing more quickly than the Dell, because of the dual CPU's. Every Apple update, the Powermac advances 200 mhz per CPU, plus the added performance that the Velocity Engine offers. (Read: advances 600 mhz) But every time the Dell advances, it is advancing about +/- 250 Mhz. I recall that when the P4 was released, correct me if I am wrong, it was clocked at about 1.2 ghz. The fastest Powermac, however, was at about 600 mhz. The Powermac cleaned the P4's clock at most graphic operations. The mhz difference was approx. 600 mhz. Right now, with the Powermac Dual/1.42 and the P4/3.0, the difference is about a total of 200 mhz. At this rate, the DP powermacs will supercede the P4 next update. But of course, the Next Update will be a 970, we hope.
**Please don't criticize me on my incorrect reasoning**
bunge is right in saying AMD's rating is based on relative performance compared to the original (K7) Athlon.
[quote]Originally posted by Eugene:
<strong>
This is an outright lie, since even the old Thunderbird chips were even faster per clock than the Athlon XPs in many tasks. AMD like Intel, lengthened its pipeline with the move to Athlon XP and took a performance hit for it.</strong><hr></blockquote>
No pipeline change since Thunderbird to Palomino (Athlon XP). In fact no pipeline change has been made in the Athlon family of processors so far (except for very minor ones). Palomino is a minor enhancement to Thunderbird with added features such as SSE, hardware prefetch and multiprocessor support. There is no fact that Palomino is slower than Thunderbird in the same frequency.
If you compare K7/K75/K76 with Thunderbird, there was some performance/clock degradation because K7* Athlons had external 512K (though slower) L2 while Thunderbird's L2 is 256K (faster) on-chip.
<strong>bunge is right in saying AMD's rating is based on relative performance compared to the original (K7) Athlon.</strong><hr></blockquote>
That's what AMD wants you to think. But where is this mythical 3000 MHz K7? What good is a comparison to a chip that never existed?
[quote]No pipeline change since Thunderbird to Palomino (Athlon XP).<hr></blockquote>
Oh?
[quote]In fact no pipeline change has been made in the Athlon family of processors so far (except for very minor ones). Palomino is a minor enhancement to Thunderbird with added features such as SSE, hardware prefetch and multiprocessor support.<hr></blockquote>
Isn't a hardware prefetch an additional pipeline stage?
[quote]There is no fact that Palomino is slower than Thunderbird in the same frequency.<hr></blockquote>
Some of the very basic integer benchmarks like RC5 claim otherwise.
Forgive me for asking, but I thought Windows XP's abstraction layer only supports Intel chips. Isn't this an unfair speed advantange for Intel. I thought that was what AMD's "XP" chips were all about......to appear "Intel" to Windows XP.
Since your post did not question its validity, it doesn't matter. You lied. AMD lied. Don't play stupid now...It doesn't matter if you knew whether it was a lie or not.
Do you believe AMD or not? Tell me what good it is for AMD to compare its current CPUs to a product that doesn't exist. And what about now? Does adding 256K of L2 cache speed up the Athlon XP enough to warrant a change in the Quantispeed curve? Go to Tech Report or any other hardware site. They all say one thing, Thoroughbreds are faster than Bartons at equal Quantispeed more often than not.
The real reason why they changed the scale was marketing, to attempt a comparison between their chips and Intel's. There's no comparison though, since the old 2800+ is faster than the new 3000+...and both aren't really in the same league as the 3.06 GHz P4.
And a side note, Intel's ready to officially launch i865PE 800 MHz FSB mobos and corresponding speedbumped P4s any time. The only thing holding them back is AMD's inability to compete.
Shuttle.com recently leaked a product page for such a board with goodies like integrated S-ATA directly off the south bridge and not off the PCI bus... They're ready and waiting.
Do you believe AMD or not? </strong><hr></blockquote>
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.</strong><hr></blockquote>
Too bad AMD officials have often stated otherwise in candid interviews.
Again, why would they change the scale so drastically with the switch to Barton cores when all the initial reviews clearly show the new ratings are bunk?
Isn't a hardware prefetch an additional pipeline stage?</strong><hr></blockquote>
No. Memory access and execution pipeline are separated in current Athlon (and most of the other processors). Interestingly, Hammer (Athlon64 and Opteron) incorporates DRAM access in the execution pipeline. It was made possible because Hammer integrates DRAM controller in the CPU. I think Hammer is the first processor that incorporates memory access in the execution pipeline.
Now this is becoming semantics. I thought prefetching was fetching an instruction before you're entirely certain it'll be executed, which can introduce some time penalties if the BPU is not up to the task.
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.</strong><hr></blockquote>
It is true that AMD officially maintain that QuantiSpeed is only meant as a comparison to their own older chips (if you ask them, that is). But do you really think this is how they believe the average customer would interpret it? I'm pretty sure that implicitely suggesting to less informed customers that QuantiSpeed ratings are equivalent to P4 MHz was not exactly something AMD's marketing deartment wanted to avoid.
<strong>I think with the next release of Macs we will see a lot of 64 bit talk especially OS wise. No matter which chip they use it will be a 64bit and Mhz really don't mean much.
At least they won't have to explain the huge Mhz gap.
Dobby.</strong><hr></blockquote>
Yup can see it now. (s jobs)"Like I told you in the last year Apple is comitted to rebuilding itself through innovation of the great products we here have all come to know and love. In addition, this year has been declaired as the year of the laptop and what an amazing year it has been (recap of sales of the 17" and 12" powerbooks). Well like I've also said in the past we have some amazing products this year and so let me show you what I mean... (lights dim and space 2001 theme plays while case rises from the floor). Today we are introducing a completely new chip design. Apple in conjunction with IBM have produced the first ever ProDesktop 64 bit chip in the history of computer technology. (System bake off and specs. Bake off against previous top of the line Powermac with photoshop 64 bit version or maybe FCP)" Yup now none of you need to watch the product anouncement <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
We can all only hope that this happens this year. Even by then they need to give us powerful 970 systems, aka 1.8Ghz chips and dual configurations somewhere in the line. They could just give us 1.4Ghz and 1.6Ghz singles Then we'd still be treading behind what Intel will be putting out by then.
Comments
My exact thoughts on the issue.
**Please don't criticize me on my incorrect reasoning**
[ 02-19-2003: Message edited by: os10geek ]</p>
[quote]Originally posted by Eugene:
<strong>
This is an outright lie, since even the old Thunderbird chips were even faster per clock than the Athlon XPs in many tasks. AMD like Intel, lengthened its pipeline with the move to Athlon XP and took a performance hit for it.</strong><hr></blockquote>
No pipeline change since Thunderbird to Palomino (Athlon XP). In fact no pipeline change has been made in the Athlon family of processors so far (except for very minor ones). Palomino is a minor enhancement to Thunderbird with added features such as SSE, hardware prefetch and multiprocessor support. There is no fact that Palomino is slower than Thunderbird in the same frequency.
If you compare K7/K75/K76 with Thunderbird, there was some performance/clock degradation because K7* Athlons had external 512K (though slower) L2 while Thunderbird's L2 is 256K (faster) on-chip.
<strong>bunge is right in saying AMD's rating is based on relative performance compared to the original (K7) Athlon.</strong><hr></blockquote>
That's what AMD wants you to think. But where is this mythical 3000 MHz K7? What good is a comparison to a chip that never existed?
[quote]No pipeline change since Thunderbird to Palomino (Athlon XP).<hr></blockquote>
Oh?
[quote]In fact no pipeline change has been made in the Athlon family of processors so far (except for very minor ones). Palomino is a minor enhancement to Thunderbird with added features such as SSE, hardware prefetch and multiprocessor support.<hr></blockquote>
Isn't a hardware prefetch an additional pipeline stage?
[quote]There is no fact that Palomino is slower than Thunderbird in the same frequency.<hr></blockquote>
Some of the very basic integer benchmarks like RC5 claim otherwise.
Is this wrong? Please pounce......I'm ready.
<strong>
Since your post did not question its validity, it doesn't matter. You lied. AMD lied. Don't play stupid now...It doesn't matter if you knew whether it was a lie or not.
</strong><hr></blockquote>
You're full of crap.
<strong>
You're full of crap.</strong><hr></blockquote>
Do you believe AMD or not? Tell me what good it is for AMD to compare its current CPUs to a product that doesn't exist. And what about now? Does adding 256K of L2 cache speed up the Athlon XP enough to warrant a change in the Quantispeed curve? Go to Tech Report or any other hardware site. They all say one thing, Thoroughbreds are faster than Bartons at equal Quantispeed more often than not.
The real reason why they changed the scale was marketing, to attempt a comparison between their chips and Intel's. There's no comparison though, since the old 2800+ is faster than the new 3000+...and both aren't really in the same league as the 3.06 GHz P4.
And a side note, Intel's ready to officially launch i865PE 800 MHz FSB mobos and corresponding speedbumped P4s any time. The only thing holding them back is AMD's inability to compete.
Shuttle.com recently leaked a product page for such a board with goodies like integrated S-ATA directly off the south bridge and not off the PCI bus... They're ready and waiting.
[ 02-19-2003: Message edited by: Eugene ]</p>
<strong>
Do you believe AMD or not? </strong><hr></blockquote>
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.
<strong>
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.</strong><hr></blockquote>
Too bad AMD officials have often stated otherwise in candid interviews.
Again, why would they change the scale so drastically with the switch to Barton cores when all the initial reviews clearly show the new ratings are bunk?
[ 02-20-2003: Message edited by: Eugene ]</p>
<strong>
Isn't a hardware prefetch an additional pipeline stage?</strong><hr></blockquote>
No. Memory access and execution pipeline are separated in current Athlon (and most of the other processors). Interestingly, Hammer (Athlon64 and Opteron) incorporates DRAM access in the execution pipeline. It was made possible because Hammer integrates DRAM controller in the CPU. I think Hammer is the first processor that incorporates memory access in the execution pipeline.
<strong>
I don't even care. AMD doesn't claim that the 'quantaspeed' rating is equivalent to an Intel chip so don't accuse AMD of doing it. The numbers are probably still off as you say, but that still doesn't mean AMD is claiming their chips are as fast as Intel chips.</strong><hr></blockquote>
It is true that AMD officially maintain that QuantiSpeed is only meant as a comparison to their own older chips (if you ask them, that is). But do you really think this is how they believe the average customer would interpret it? I'm pretty sure that implicitely suggesting to less informed customers that QuantiSpeed ratings are equivalent to P4 MHz was not exactly something AMD's marketing deartment wanted to avoid.
Bye,
RazzFazz
<strong>I think with the next release of Macs we will see a lot of 64 bit talk especially OS wise. No matter which chip they use it will be a 64bit and Mhz really don't mean much.
At least they won't have to explain the huge Mhz gap.
Dobby.</strong><hr></blockquote>
Yup can see it now. (s jobs)"Like I told you in the last year Apple is comitted to rebuilding itself through innovation of the great products we here have all come to know and love. In addition, this year has been declaired as the year of the laptop and what an amazing year it has been (recap of sales of the 17" and 12" powerbooks). Well like I've also said in the past we have some amazing products this year and so let me show you what I mean... (lights dim and space 2001 theme plays while case rises from the floor). Today we are introducing a completely new chip design. Apple in conjunction with IBM have produced the first ever ProDesktop 64 bit chip in the history of computer technology. (System bake off and specs. Bake off against previous top of the line Powermac with photoshop 64 bit version or maybe FCP)" Yup now none of you need to watch the product anouncement <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
A@ron