Sign in with
Sign up | Sign in
Your question

Can enlighten...?

Last response: in CPUs
Share
August 17, 2002 11:31:47 AM

I'm an AMD user. Ever since pentium came out with 2.53Ghz with 533 FSB, my royalty fer AMD has shaken a bit. Using for graphic heavy games mostly, I would certainly appreciate a few answers
1. AMD(Thoroughbred) with DDR PC3200 Pentium(Northwood) with RDRAM PC1066. Which is definitely faster?
2. What technology does AMD use. What does intel use?
3. Which is better in overclocking?

More about : enlighten

August 17, 2002 11:45:52 AM

1. At stock speeds, the current fastest Athlon at 1.8 (and soon will be 1.933 and 2.066 supposedly, but rumored to be 2.066 and 2.133) can't really compare to the 2.53 in terms of performance in any application. The 1.933 and 2.133 may have a better chance of matching and maybe even beating the 2.53 but the 2.66 and 2.8's will be out 5 days after their launch. It's a close match.

2. As far as I know, both use the old CMOS technology on silicon wafers. Currently they're both on the .13 micron transistor size with copper interconnects. Quite similar on the base level, dramatically different on the architectural level. There's really too much in that to explain all at once.

3. The P4 hands down is an overclocking machine. You can take the lowest Athlon (even if you take a t-bred and say it started at 1.4 GHz) and have a hard time getting a 1 GHz overclock. With a 1.6a, that's actually a pretty conservative goal.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.<P ID="edit"><FONT SIZE=-1><EM>Edited by imgod2u on 08/17/02 04:46 AM.</EM></FONT></P>
August 17, 2002 6:41:26 PM

Wait just a bit for the next round of chips the 2400+, 2600+, 2.66 and 2.8. Then we will see who comes out ahead. And if you wait and buy later you will save money to get what would cost you a couple hundred more right now.
Related resources
August 17, 2002 6:41:57 PM

when is athlon at 1.933 and 2.133 coming out? cos <A HREF="http://www.pcprogress.com/products.asp?orderid=42761549..." target="_new">they're</A> showing they have p4 2.8 for sale (with no word of preorder in sight, but its probably is).

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
August 18, 2002 7:41:25 AM

Quote:
1. At stock speeds, the current fastest Athlon at 1.8 (and soon will be 1.933 and 2.066 supposedly, but rumored to be 2.066 and 2.133) can't really compare to the 2.53 in terms of performance in <b>any</b> application.

This is not true...for starters...the 2200+ holds its own against the 2.53GHz chip....overall the 2.53 is faster....but not in <b>ALL</b> like u said.....the K7 still holds the crown for 3D Rendering software and even some video editing........thanx to the unmatched FPU unit that the Athlon has =)

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
August 18, 2002 8:14:40 AM

Quote:
3. The P4 hands down is an overclocking machine. You can take the lowest Athlon (even if you take a t-bred and say it started at 1.4 GHz) and have a hard time getting a 1 GHz overclock. With a 1.6a, that's actually a pretty conservative goal.


The 1ghz you get from the athlon would outperform the 1ghz from the p4, so your comparison is not valid.

Now a 1.33ghz tbred(1500+) could reliably hit 2ghz this is about a 55% increase, most 1.6a's can hit 2.4 reliably, this also is about a 55% increase.

While no one disagrees the p4's overclocking is better, the situation you presented actually would be relativly equal in overclockability.(btw the 2ghz athlon would smoke the 2.4ghz p4, and cost less).

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
August 18, 2002 8:16:29 AM

Quote:
This is not true...for starters...the 2200+ holds its own against the 2.53GHz chip....overall the 2.53 is faster....but not in ALL like u said.....the K7 still holds the crown for 3D Rendering software and even some video editing........thanx to the unmatched FPU unit that the Athlon has =)


Not to mention, you probably wont notice the performance difference betwen the 2200+ and the 2.53 pentium, but you sure as hell will notice the 200 bucks extra you had to pony up for the p4.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
August 18, 2002 8:19:35 AM

Ok gentlemen, this is stuff strait of my lab.

I've gotten to test the 2.8 P4 and the 2800+ AMD. The AMD cpu was code named Fred by us in the lab because well we didn't know which core it is from AMD........:) 

Now the P4 used 1066 rambus, 512 megs 9700 ATI card, 2 raid, at 0.

AMD "Fred" used DDR 400 Nforce 2, 9700 ATI card, 2 raid, at 0.

they were within 5 frames in Quake III,

P4 2.8 scored around 15500 in 3dmark SE latest build.
AMD Fred scored around 14500-15000 in 3dmark SE latest build.

At one point they were only 100 points between the two cpus.

I easily believe that beta Nforce 2 board could of ran ddr 400 ram at case 2......they would of been tied.

there is still life in Athlon core......;) its going to get interesting next couple months....

MeldarthX
a b à CPUs
August 18, 2002 10:25:48 AM

Geeze, care to get a Pentium III? There's still life in the Tualatin core after all...

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 18, 2002 6:33:13 PM

Where did you get all that stuff so early? The XP2800 may have been shown live but never have I heard it being sampled by people already!
I find it surprising that they came 5FPS in Quake 3, especially when they were 50FPS different from the highest end AXPs vs highest end P4s in current tests!

However I am not surprised there is still life. Whoever continues to beleive the K7 is at its end is a moron living in the past. Raystonn said it'd die at 2GHZ, doesn't seem the case anymore. In fact I still beleive the P4 was designed to be TWICE faster when it reaches the max speed of the architecture. So far it has been that way. At 4GHZ, the P4 needs a shrink to 0.09m, the Athlon is at 0.13m at 2.2GHZ, it will shrink after that for sure, if not then the Dresden guys must be pretty skilled. But anyway, the 0.09m AthlonXP can pretty much reach 3GHZ, whereas the 0.09m P4 would top around 6GHZ. So IMO, I can see the K7 going on and on like the P4 in die shrinks and in the end, the P4 would be TWICE more because of the pipeline and therefore it proves that no, it does not mean a pipe twice longer, would or can give out 3, 4 times more speed.


--
Is the opportunity to earn money by working, free?
August 18, 2002 6:40:27 PM

OK......but by what ur saying that also means the Scalabilty of the P4 is great...but the thing is [-peep-] IPC compared to the Athlon.....which alos would mean MHz means nothing...which is true......

i dunno about u but id MUCH rather own a 3GHz chip that outperforms a 6GHz chip thats designed for the same platform AKA x86 architecture..........only cuz u THAN know the people designing those things are VERY skilled......

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
August 18, 2002 6:48:28 PM

Hmm that wasn't exactly what I meant.

Intel released the P4 too early, it was hot as hell, giant, required a socket switch, had extra voltage requirements, and sucked ass. Had they waited and worked so hard, the P4 today would be a great competitor, designed to scale twice the Athlons, to speeds beyond 8GHZ, with a great IPC.

The scalability is of course great, but I would like somebody to prove me where other ,than the pipeline, does it make the scaling better! I mean we say it was designed to ramp in clock speeds, but saying that, to me, means there are many things contributing in the NetBurst core that made it have this attribute. So someone prove me that other than twice the pipeline length, something else or other things help. Therefore no I am not saying that if Athlons are twice less clocked, they would still outperform a twice higher clocked P4. I was saying the Athlons will die out at half the P4's dying clock speed. P4s by then will obviously have a high IPC, so a 6GHZ P4 would definitly be a powerhouse. How will AMD compete? I wish I knew, but such clock speed scaling being open, and the possibility to add IPC without sacrificing clock speed, or very little, is a brilliant idea I wonder why AMD doesn't just go for it, even if it takes time to design, at least they have a prototype on the horizon, to strike back.

--
Is the opportunity to earn money by working, free?
August 18, 2002 10:48:54 PM

Quote:
The 1ghz you get from the athlon would outperform the 1ghz from the p4, so your comparison is not valid.

Now a 1.33ghz tbred(1500+) could reliably hit 2ghz this is about a 55% increase, most 1.6a's can hit 2.4 reliably, this also is about a 55% increase.

While no one disagrees the p4's overclocking is better, the situation you presented actually would be relativly equal in overclockability.(btw the 2ghz athlon would smoke the 2.4ghz p4, and cost less).

1 GHz from 1.4 compared to 1 GHz from 1.6 isn't a huge difference in terms of performance increase, 71% vs 63%. Although yes, the point is conceded.
However, your example is quite incorrect. Of the sites that have tried a t-bred overclock, mostly with extreme cooling (such as watercooling or even a vapchill), 2 GHz was reached maybe 4/7 of the time (according to the initial 7 reviews I read). With the newer steppings, let's just give it that it can reach 2 GHz with air cooling and be relatively stable except for the few 20% that may not gain that much with all the voltage increases you want (so long as it doesn't kill the CPU). Of the 1.6a's I've seen, very few have been limited at 2.4 GHz. Some people have stopped there due to limitations in FSB, motherboard settings, or memory limitations but most have gotten 2.6 or higher with voltages of 1.6+ vcore (although below 1.75).

Quote:
The scalability is of course great, but I would like somebody to prove me where other ,than the pipeline, does it make the scaling better! I mean we say it was designed to ramp in clock speeds, but saying that, to me, means there are many things contributing in the NetBurst core that made it have this attribute. So someone prove me that other than twice the pipeline length, something else or other things help. Therefore no I am not saying that if Athlons are twice less clocked, they would still outperform a twice higher clocked P4. I was saying the Athlons will die out at half the P4's dying clock speed. P4s by then will obviously have a high IPC, so a 6GHZ P4 would definitly be a powerhouse. How will AMD compete? I wish I knew, but such clock speed scaling being open, and the possibility to add IPC without sacrificing clock speed, or very little, is a brilliant idea I wonder why AMD doesn't just go for it, even if it takes time to design, at least they have a prototype on the horizon, to strike back.

There is really no way to directly derive the max scalability exactly based on pipeline length of the integer pipelines. There are far too many variables such as heat, cooling, processing technology, transistor size, materials used to make the CPU. However, there will come a point where no matter how good some of those are, you still won't be able to scale much higher.
As for the P4's scalability, all the pipelines on the design was extended. The FP pipelines were extended, the integer pipelines were extended, the L1 caches were reduced and a trace cache was added which removed (arguably) the need for heavy duty (which would slow down scalability) decoding units.

Quote:
i dunno about u but id MUCH rather own a 3GHz chip that outperforms a 6GHz chip thats designed for the same platform AKA x86 architecture..........only cuz u THAN know the people designing those things are VERY skilled......

The difference is, the Athlon at 3 GHz would not perform better, or even near a 6 GHz P4. Of course, either of these are farther than any of the current designs could hope to reach on the current manufacturing process and steppings, but 6 GHz is a pretty low feasible goal for the P7 core (used in the P4) while no matter how much you revise the K7 core, I doubt you'll get much higher than 3 GHz. Although it may get there and maybe even a bit more.
The point was not that in order to gain scalability, you'd have to sacrifice IPC (which is BS to begin with, if it were, no more processors would be designed because with any design that gains in one, you'd loose in the other), but rather, the P7 core design (specifically the P4) is able to gain a better percentage in scalability than it would loose in IPC, which makes it overall a better design than its predecessor (the P6 core in the P3) and arguable better than the K7 design (which seems stuck at below 2.5 GHz, given future revisions and enhancements). As opposed to the P4, which has already been demonstrated by Intel at IDF at 4.1 GHz (watercooled albeit).

Quote:
This is not true...for starters...the 2200+ holds its own against the 2.53GHz chip....overall the 2.53 is faster....but not in ALL like u said.....the K7 still holds the crown for 3D Rendering software and even some video editing........thanx to the unmatched FPU unit that the Athlon has =)

<A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1635&p=7" target="_new">http://www.anandtech.com/cpu/showdoc.html?i=1635&p=7&lt;/A>
The 1.8 held it's own pretty well and managed to beat the 2.4B in a few but the 2.53 is unmatched.<P ID="edit"><FONT SIZE=-1><EM>Edited by imgod2u on 08/18/02 04:15 PM.</EM></FONT></P>
August 19, 2002 12:02:59 AM

" dunno about u but id MUCH rather own a 3GHz chip that outperforms a 6GHz chip thats designed for the same platform AKA x86 architecture..........only cuz u THAN know the people designing those things are VERY skilled......"

A 3GHtz Athlon against a 6GHtz P4 would get blown away. Just look at the Tom's Hardware comparison's between the 2200+ XP and the 2.53GHtz P4. The Athlon is struggling to hold it's own against it in some areas (even though it still gets outperformed) and in others it simply gets left in the dust. Just take a look at this: http://www.tomshardware.com/cpu/02q2/020610/index.html

Knowledge is the key to understanding
August 19, 2002 12:20:56 AM

Quote:
However, your example is quite incorrect. Of the sites that have tried a t-bred overclock, mostly with extreme cooling (such as watercooling or even a vapchill), 2 GHz was reached maybe 4/7 of the time (according to the initial 7 reviews I read). With the newer steppings, let's just give it that it can reach 2 GHz with air cooling and be relatively stable except for the few 20% that may not gain that much with all the voltage increases you want (so long as it doesn't kill the CPU). Of the 1.6a's I've seen, very few have been limited at 2.4 GHz. Some people have stopped there due to limitations in FSB, motherboard settings, or memory limitations but most have gotten 2.6 or higher with voltages of 1.6+ vcore (although below 1.75).


A limitation is a limitation, while I agree many 1.6as can hit higher than 2.4 most motherboards and rdram dont like those speeds, the 2.4 number is a round figure for an average overclock, there is supposed to be a core revised tbred which promised better overclocking.

Regardless, both overclocks are very good, and both are respectable.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
August 19, 2002 2:48:23 AM

Yes that was what I was thinking, proper arranging of the core, proper electrical specifications, and all you specified, contributes to how far a core can go. But I just don't like it when someone states the P4 was designed to ramp in clock speed, and have a text to support it by making it seem that there were other things than the pipeline architecture-wise that majorly contribute. I know what I am saying, though you may have not been here when that was said often, but case in point, I wanted proof, and since you have a very good knowledge of the P7 core, you've laid out very understandable reasons, which do provide more support to my claim that it is MAJORLY pipeline-wise, therefore a good theoretical 2 to 2.5 times higher speeds, and then you add the extra with the core enhancements and fab aids.
Again AMD can go on shrinking the Athlon core, so we can only know when the process shrink cannot help go further, that the core is dead. But until then, shrink on AMD, for we want to know the real limits, and see whether those "prophets" of semiconductors were wrong!

I still disagree with whomever states the K7 will stop at 3GHZ or a tad more.
And why do you say 6GHZ is not gonna be easy to reach goal for P4s? If Athlons can reach 3GHZ, P4s will as well on the SAME process, from current technology standpoints. Perhaps it was designed to top at 6GHZ thanks to the pipeline length, while the extra 4GHZ comes from the new silicon techniques, the BBUL packaging, and many more process enhancements, but not SHRINKS.

--
Is the opportunity to earn money by working, free?
August 19, 2002 4:04:42 AM

Yeah, you can only push an architecture so far...

Knowledge is the key to understanding
August 19, 2002 6:39:56 PM

Hmm you were replying to what claim here?

--
Is the opportunity to earn money by working, free?
August 19, 2002 7:20:31 PM

I was just giving an example dude.........keep in mind the 2200+ runs at 1,800MHz.......2,530 - 1,800 = 730...thats a 730MHz difference....thats VERY respectable on the Athlons part for doing that well already...in MHz....

therefore again, i was statign in general....id rather have a 3GHz chip that performs equal to or better than a 6GHz chip.........understand what i mean ?

<A HREF="http://www.anandtech.com/mysystemrig.html?id=13597" target="_new">-MeTaL RoCkEr</A>
August 19, 2002 9:02:29 PM

Yes yes, your preference towards higher IPC rather than clockspeed. But that's really all it is, a preference. My whole point was that it really doesn't apply here, as the P7 core's theoretical scalability far surpasses the Athlon's theoretical IPC advantage. Now, before you point out Hammer, the P7 core will receive future revisions to help its IPC as well.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
!