• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Are 1.3 TB's different from 1.2 TB in design? & an intriging CPU question

D

Degenerate

Platinum Member
I have seen some bench marks where the two and other athlons dont form a linear line at all. WHY?

Also, just as a side question, how do manufactures increase the spped of the CPU eg from P3 450 to a 1.ghz. Isn't it the same design? do they just permantly overclock it? Does that mean tha from the chips release at 450 mhz, intel could make a 1.0 ghx straight away?
 
The p3 450 design maxed out at 600 Mhz. The next p3 rev brought on a smaller fab process (.18 micron), which allowed the p3 to run up to much higher speeds, and include the L2 cache on die. P3 would eventually max out at 1 ghz. They reach this goal by improving the manufacturing process. (Wingnutpez might be able to shed more detail on this) but from I understand that each revision done on a cpu can be minor tweak to its design( to fix errata), and/or a process improvement where something is change/improve in the process to get better yeilding CPUs.


has for the athlons, the biggest difference can be attributed to the difference in bus speeds. Going from a 100/200 fsb to a 133/266 fsb brings a much larger performance increase than a speed increase in the cpu can do.


All this is from my own understanding, so I may be slightly off or more.
 
Also, as speeds go up in a CPU line and approach the maximum speed of the design, the graph for performance versus speed bcomes more flatline. The graph will look like a logarithmic function. Every step higher will bring that graph ever so slightly more parallel to the x axis (speed).
 
That is correct.

Speed is determined by many design factors, which are seperate from physical limitations.

In the Athlon's case, it's current generation is reaching it's usuable maximum. Athlon's design shows a severe a IPC penality as clock ramps, which explains the lack of regular performance ramping you are observing. In comparison, the P4 was designed to ramp up and continue to ramp up for years.
 
Yea, I would have to say that the benchmarks you saw must have been 133MHz busses vrs 100MHz. So far the Athlon is scaleing pretty nicely up to 1.5GHz. As Adul mentioned, once you reach a chip's limit (was 600MHz for the original Katmai PIII process) then something must be done - normally shrinking the CPU to a smaller process (Katmai was 0.25 micron, Coppermine was 0.18).

TBird is probably only going to go to about 1.5GHz on the current 0.18 process. A shrink to 0.13 will take it to ??? - whatever it is, much higher than 1.5GHz.
 


<< In comparison, the P4 was designed to ramp up and continue to ramp up for years. >>



After the socket change this year, and the year after that, and on and on, etc... 😛
 
That is also so.

And the P4 will continue to be faster after the socket change. This form change didn't bother many of you nearly as much when AMD completely stopped production of Slot-A CPUs in prefrence for the Socket-A designs.

AMD party politics aside, form factor changes a fact of life. Whenever there is a significant improvment to a CPU, pins must be added to support the extra functions, packages changed to reflect new die sizes, cache sizes, etc.
 


<< That is correct.

Speed is determined by many design factors, which are seperate from physical limitations.

In the Athlon's case, it's current generation is reaching it's usuable maximum. Athlon's design shows a severe a IPC penality as clock ramps, which explains the lack of regular performance ramping you are observing. In comparison, the P4 was designed to ramp up and continue to ramp up for years. >>




The athlon biggest problem is the heat it generates. It still has room to go up in clock speed with out to much problem. Granted it is nothing like intel's clock speed of the p4, but the again the Athlon does get more work done for what is has to work with. When the next core rev of the athlon comes out, they expect that to hit 2 Ghz on the current .18 micron process. .13 micron will be needed to continue on.


** information, thansk for not sounding like a troll.
 
The Athlon's heat problem is due to inneficient trace layout and other poor design features.

The Athlon, as a previous-generation architecutre, benifits from not having to deal with a highly latent pipeline, like the P3 does not have to (and you will note that the P3 out-performs the Athlon at the same clock speed). When all of the P4's features are used, the P4 performs substantially MORE results at the same clock speed. But, as we have discovered (and what should be common knowledge by now) legacy applications do not use any of the P4's extra features, so it relies on the strength of it's clock.

The Athlon will not hit 2Ghz at it's current process.
 
information information hehehe
gezzz everywhere i go i see this guy, seems like there is a general consensus that you should change your name to dis-information
 
Information, thank you for being a bit more objective, today. Although you have lost a lot of credibility from MANY people, perhaps posts like these that are attempts to remain completely factual will sway some people back to civility.

Good luck! 🙂
 
gullible readers, welcome, this is Information, your true and bestest friend.
the PIII does not outperform the Athlon in benchmarks even clock for clock, and Athlons are 33% faster in clock now also. the PIII loses in almost every benchmark especially due to the superior Athlon triple pipelined FPU.
the P4 does not rely on the strength of its clock; Intel and the P4 are becoming notorious for the devaluation of processor clock speed.
the P4 performs substantially better than ITSELF using its own enhanced instruction set, but not better than the Athlon. now, did you take a look at how expensive your apps were? and how much it would cost you to upgrade them just to use SSE2 to make the p4 pull its weight?
 


<< The Athlon's heat problem is due to inneficient trace layout and other poor design features.

The Athlon, as a previous-generation architecutre, benifits from not having to deal with a highly latent pipeline, like the P3 does not have to (and you will note that the P3 out-performs the Athlon at the same clock speed). When all of the P4's features are used, the P4 performs substantially MORE results at the same clock speed. But, as we have discovered (and what should be common knowledge by now) legacy applications do not use any of the P4's extra features, so it relies on the strength of it's clock.

The Athlon will not hit 2Ghz at it's current process. >>

<-- I beg to differ on this.

in some case it does, in other cases it doesn't. Also, there was an app called pov-ray thay some members on the board compiled with SSE2 optimizations a while back. While the p4 gain 50% increase in performance, when it came down to clock per clock, the p3 was still faster. While the athlon was a lot more efficient in finishing the 3d renders. Even a duron was doing better on a clock for clock basis. Also the athlon gain from the optimizations done to the Pov-ray app.

I am not going to get in a argument about the Athlon's trace layouts, and whether it is or not a factor in the athlon's heat output. I will say this, there is very little reason for the athlon not to be able to hit 2ghz on the next core with the current .18 micron process. Even with the current athlon people are able to hit 1.5 and 1.6 Ghz using conventional cooling. The next athlon core with have less heat output, improved branch prediction, hardware prefetch, and some additional improvements to the core. This core will be introduced at 1.533 Ghz.


We should also see an increase in performance on a a clock for clock basis vs the current athlon core.
 


<< gullible readers, welcome, this is Information, your true and bestest friend.
the PIII does not outperform the Athlon in benchmarks even clock for clock, and Athlons are 33% faster in clock now also. the PIII loses in almost every benchmark especially due to the superior Athlon triple pipelined FPU.
the P4 does not rely on the strength of its clock; Intel and the P4 are becoming notorious for the devaluation of processor clock speed.
the P4 performs substantially better than ITSELF using its own enhanced instruction set, but not better than the Athlon. now, did you take a look at how expensive your apps were? and how much it would cost you to upgrade them just to use SSE2 to make the p4 pull its weight? >>



nice spin on things travis. 😛

Also intel's market share has shrunk since the Athlon's introduction. AMD now has 21% of the consumer processor market. They would not be there if they did not have a good chip.
 
Well, Palomino will be out (hopefully) soon and P4 won't hit 3Ghz at it's current process either. Saying athlon being poorly designed is kinda stupid: it was clearly superior than P3 (actually P2 with SSE🙂 ) when it came out. You can check the Anand's reviews to see how P3 (mendocino) gets beaten by Athlon (original) in 90% of apps mostly by about 20% margin. P3 (coppermine) outperforms Athlon (t-bird) on Quake3 and some other rare games/apps, there are also places where P3 gets badly smashed by athlon (like 3D rendering where in some places 700Mhz duron performs like 1Ghz P3). Several months have passed since P4 was released but still only few apps are taking advantage of SSE2. What if SSE2 never gets as populas as it needs to be to help Intel out with it's undeperforming P4 ?
 
intel is going to have to push the industry hard to get all apps optmized for the p4. One has to wonder how well these apps will be optimized has well. Intel is betting far too much IMHO on software companies to bail them out here. Just look how long it took for applications to become fully 32 bit. Considering intel released the first 32 processor with the 386.
 
The P3 is indeed faster than the Athlon at the same clock.

This is a known fact which has apparantly left the realm of human understanding now that P3s really aren't benchmarked in comparison anymore. I'm just doing my part here to keep it real.

The P4 is much faster. This has already been established through many web reviews, and I don't feel like posting a URL for each and every major review site on the web.
 
You pic a single benchmark. Also that is an older review, today people are now running the 1ghz athlons on the kt133a, AMD 760, and the Ali magitech.

http://www.anandtech.com/showdoc.html?i=1441&amp;p=7note the bypco bencmark

http://www.anandtech.com/showdoc.html?i=1441&amp;p=8

now take note of the quake 3 and unreal trounament benchmark.

http://www.anandtech.com/showdoc.html?i=1441&amp;p=9

Now look at the serious Sam bencmark, the p3 is choking bad here.
p3 is also choking for MBTR

http://www.anandtech.com/showdoc.html?i=1441&amp;p=10

Ouch, the athlon 1 ghz raped the p3 1 ghz here in both the wave-->MP3 conversion and the wave-->WMA

Now the webmark is a bit strange.

The p4 wins on 2 of the 6 benchmarks, but rates higher overall? 😕
http://www.anandtech.com/showdoc.html?i=1441&amp;p=11

and finally from the boy wonder.



<< First, the Pentium III is an aging platform. It is perfectly suitable for the majority of the users out there, however once you begin to weigh down its 133MHz FSB with multiple concurrent transactions, the architecture will show its age. >>



P3 is only faster under certain situations.

 
&quot;The P4 is much faster&quot;
Ok me get this straight you are claming that the P4 is faster when you have to pull a P4 1.7Ghz out to edge out a 1.33Ghz t-bird processor. we don't need to get into clock for clock we know whose the winner there with today?s apps.

 


<< The Athlon's heat problem is due to inneficient trace layout and other poor design features. >>



Prove that.



<< But, as we have discovered (and what should be common knowledge by now) legacy applications do not use any of the P4's extra features, so it relies on the strength of it's clock. >>



Ha! You mean legacy apps like Win2K (and probably WinXP), Office 2K, current games, current benchmarks, etc, etc?



<< The Athlon will not hit 2Ghz at it's current process >>



We'll see won't we?
 


<< has for the athlons, the biggest difference can be attributed to the difference in bus speeds. Going from a 100/200 fsn to a 133/266 fsb bring a much larger performance increase than a CPu increase only can do. >>



that must be it. Funny though, how this thread has change topic.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top