PDA

View Full Version : Where is my 1.4 T-Bird?


Silver
01-12-02, 12:42 PM
Alright, spent a few hours and gathered reams of paper to come up with the following (of course from the source).:)

The Athlon 1.4 was used as the base line for the xp series of processors as far as performance testing goes. This is to say that the 1.4 Athlon was considered 100%. Each successive xp was designed to increase performance from here by approximately three percent. The xp1500 operating at 1.33 Ghz actually outperformed the 1.4 Athlon at stock settings by about 1%. An additional 3.5% was dependant on the inclusion of Windows media encoder version 7.0 (which enabled 3DNow) for a total of 1.045% over the Athlon 1.4. After this initial step each successive Xp resulted in an increase of 3% including 3DNow utilization. This of course is why my xp1600 shows it has 3dNow as well as SSE.

This also explains why CPU testing as reported by Sisoft and all others testing during reviews show the following.

CPU Dhrystone Whetstone

Athlon 1.0 2746 1363
Athlon 1.2 3295 1636
Athlon 1.33 3750 1848
Athlon 1.4 3904 1887 Baseline 100%
XP1500 Don't Have Yet (1.330 GHz)
Athlon xp1600 3872 1940 (1.400)
XP1700 Don't Have Yet (1.467)
XP1800 4240 2124 (1.533)
xp1900 Don't Have Yet
xp2000 Don't Have Yet

Athlon 1.53 4274 2066 Overclocked

Total speed increase between xp1800 and Athlon 1.4 is reported by AMD (source) at 13.5% (October 5, 2001). This was in "overall Desktop Performance". It is very obvious that the enhancements in the core effected the xp processors but inordinately so due to the implementation of the version 7.0 Microsoft Windows Media Encoder.

Fundamentally Cpu Power is the amount of work done (ICPs)x speed at which it is done (MHz). This explains why Pentiums don't due much work but do it at a rapid speed (higher Mhz). The Athlons and the XPs do more work but at a lower speed. End result, The guy with the big bucket does less but moves more water than the guy with the can going like H......in the same period of time.

To wit, "the AMD Athlon xp processor 1800+ outperforms the Intel Pentium 4 processor 1.8 GHz by up to 50% on an industry standard entertainment application." AMD 10/5/01

Without the use of the Microsoft encoder (7.0) the xp1800 is only 8.5% faster than the T-Bird at 1.4.

Now it is easy to see why processor power on the T-Bird could be overclocked to attain these levels without much trouble as the processing power of the cpus was not that dramatically changed in relation to the Baseline T-Bird 1.4. A 114% increase in processing power over its present rating would put it squarely in xp1800 territory (1.6GHz). Yes, this is faster than the xp1800 at 1.53GHz but it has a slightly bigger bucket. BTW at 115% of the T-Bird 1.4 (1.61GHz) the T-Bird would exceed the xp1800 at stock speed (as indicated by xp1800 max speed-Amd). The important thing to remember here is that the performance of the T-Bird 1.4 was the base line used by AMD for all of the XP ratings and that all revisions were to go up by approximately 3%. Wonder why they dropped the original name "Palomino" used during testing and adopted the Athlon XP system.

Some notes,

" The Athlon XP2000+ is faster than the Pentium 4 2.0GHz by 12% in overall desktop performance" AMD

Testing by AMD showed the XP1800 to be 127% (1.53) faster than a Pentium 4 (1.5GHz) in overall performance testing. This data is possible being misinterpreted by some as anything that is not an xp cpu. Amd most assuredly will not tell you that a T-Bird 1.4 would have to run at 1820 when it recognizes the xps advantage to be only 13%. At 1.582GHz the T-bird catches up.

Athlon 1.4 and all XP1500s to xp1800s were tested on the Gigabyte GA-7DX (rev 4.0) for DDR testing. AMD

Athlon 1.4 and all xp1500s to xp1800s were tested on the
Asus A7V133 (rev 1.05) SDRAM. AMD

Intel processors used on a Intel D850GB mobo. Ram used PC-800 RDRAM.

Ram of the respective type for each AMD board was used from Micron.
Give you the model number if you really want it.

AMD testers are really into testing when they throw Sound Blaster Live cards on all of the test boards. AMD

Conclusion:
Don't let anyone tell you that at this time you are lying about your Sisoft Scores or that you need to overclock your T-Bird to 2.5 GHz just to catch up with their XP1600. I have one and I assure you that an overclocked t-bird at 1.7 will leave the XP1600 looking for an Intel. Lets see xp2000 at 1.663 (stock). T-bird 1400x1.2=1680 ....hmm 1.77GHz ought to cover it.

:)

Silver
01-12-02, 12:56 PM
Just thought I would throw this in. It came off of one of the numerous reviews located everywhere on the xp2000. Interesting that there are no comparisons to the Athlon 1.33 or 1.4. Pr Rating is of course estimated. Repeat, this test ws not done by me or anyone I have heard of or know. It is an independant review.

Silver
01-12-02, 01:02 PM
And my xp1600 at 1.747 tested at Dhry 4874, Whet 2424 with a pr rating of 2324, mmx 9562, 3dnow 11,953. Yep, that would appear to be about right. Problem of course is that it was not stable at that speed (yet) but that T-bird.....

stompah
01-12-02, 03:50 PM
I thought that the XP +rating basically was rated equal to a tbird's MHz? (for example a Tbird@1.6g = 1600+ XP)

I saw this somewhere on XPCentral or AMD's website, but cannot remember where.

Anyhow your explaination is good. I guess the only way to prove it is to get a tbird up to 1.6g and pit it against 1800+ and throw a bunch of tests at it.

Silver
01-12-02, 04:05 PM
Me too however Wild Andy did give me a little lesson. I had to go around the net to find it. Here is how it goes, due to core changes the xp was enhanced. Now this speed increase will show up in Sisoft testing. At this you can tie it. The problem is that some of the enhancement was dependant on a Microsoft revision. This will not show up in testing. Using the T-Bird as the baseline (AMD did) each new revision is to go up 3%, the first (xp1500) a little more at about 4.5%. It is rather easy to do the math as far as overclocking it after that and the gap is not so large that a descent 1.4 ca'nt overtake it. I.E. xp1800 is 4 steps up 4.5+3+3+3=13.5%. 1400x1.135=1589. So at 1589 you are playing with the xp1800 at 1530. Now Intel on the other hand has a good deal farther to go.

Silver
01-12-02, 04:08 PM
Andy, if you ever get in here, I was wrong and you were right (though not to that degree). I am sure that there are alot of people that thought the same as I. Yes a T-bird has to run faster than a xp though not as high as was indicated and not NEARLY as high as intel. Thank you for this info and I'm sure alot of overclockers would like to know this.

Diu
01-12-02, 08:04 PM
lol, Nice. :D

Silver
01-28-02, 03:27 PM
Bump

res0r9lm
01-28-02, 04:09 PM
I'm getting 1383 out of 1.4 t bird@1.41

res0r9lm
01-28-02, 04:28 PM
Originally posted by res0r9lm
I'm getting 1383 out of 1.4 t bird@1.41

Angryviking
01-28-02, 07:20 PM
Whoa! I thought OCing my 1.2 was just fun... :D

I didn't actually think it would measure up to the XP line in any capacity :eek:

If your numbers are right Silver, I am quite happy...

My 1.2 @ 1.53 = 4255/ 2110
XP1800 @ 1.53 = 4240/ 2124

Yee... haw? :beer:

Silver
01-28-02, 08:51 PM
Believe me I spent a whole day working on this as I was unaware of it as well. The fact of the matter is that the xps are faster in desktop applications. AMD did use the 1.4 T-bird as the baseline for xp testing 1500 through 1800. AMD furthermore shows the increase in cpu due to restructuring of the core as well as the percentage due to the software implimentation. The larger increase was due to the software. The speed at which a t-bird would have to run in order to match the xp is given (as a percentage) by AMD. BTW, the new sisoft will show this increase at least in the pr rating while the old sisoft does not.

Silver
01-28-02, 09:04 PM
Originally posted by Angryviking
Whoa! I thought OCing my 1.2 was just fun... :D

I didn't actually think it would measure up to the XP line in any capacity :eek:

If your numbers are right Silver, I am quite happy...

My 1.2 @ 1.53 = 4255/ 2110
XP1800 @ 1.53 = 4240/ 2124

Yee... haw? :beer:

Now in order to make up the software advantage you have to go a little farther. About 1.589 should do it. This is of course in desktop applications soooooo. Short of it is 1.589 should match it according to AMD. I personally would'nt worry about it too much as you are in the same ball park. You will see it really showing in the pr rating. An xp1600 at 1.7 is doing pr2400? My T-bird at 1.72 is presently doing about 2290. This definately shows the advantage given to this core by the software. You can definately play with at least some of the xps.

Angryviking
01-28-02, 11:15 PM
I like your thinking Silver. :beer:

with the figures to back it up too !

:burn:

Thanks

Kendan
03-06-02, 04:01 AM
Bump!!!

bobt17
03-07-02, 12:58 AM
BUMP good work silver

Silver
03-07-02, 03:29 AM
Spent a day working on this when it was first brought to my attention.:)

bobt17
03-07-02, 04:26 AM
yeah b4 reading this i thought a tbird would have to run at 1.6ghz to be equal to an athlon xp 1600+

Eriksson
03-11-02, 11:01 AM
Good post, thanks for the info :)