savyj

08-17-02, 09:45 PM

I decided (based on my other post asking whether or not AMD could temporarily take back the flagship crown from Intel) to try to extrapolate some data on the matter. Based on Anandtech's Athlon XP 2200+ review (http://www.anandtech.com/cpu/showdoc.html?i=1635&p=1) which pitted the XP 2200+(KT333) against the Northwood P4's with RDRAM (PC1066). I ran the numbers and this is what I came up with.

Calculations based on:

NVIDIA's claim for the nForce2 to beat KT333 performance by "at least a few percent". "A few" is generally defined as a small number usually between 3 and 5. For these tests, I used 3.

I graphed out the average percent increase per 66mhz increase in Athlon XP speed (based on the Anandtech numbers) and for sake of efficiency loss, ROUNDED DOWN. Therefore if the average was a 3.68% increase per 66mhz increase I used 3% for my calculations.

DISCLAIMER: These are simple math EXTRAPOLATIONS. Do NOT make purchase decisions based on these numbers, they WILL NOT necessarily resemble the actual benchmark numbers of the shipping products (once shipped :-)

SYSMark2002 O.P. - Athlon XP increased 2% per 66mhz increase

2600+/nForce2=177

2600+/KT333=172

2400+/nForce2=170

2400+/KT333=165

P4-2.53/850E=178

SYSMark2002 I.C.C. - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=278

2600+/KT333=270

2400+/nForce2=262

2400+/KT333=254

P4-2.53/850E=330

Lame MP3 Encoder - Athlon XP decreased 3% per 66mhz increase

2600+/nForce2=1.48

2600+/KT333=1.52

2400+/nForce2=1.57

2400+/KT333=1.62

P4-2.53/850E=1.65

Adobe After Effects - Athlon XP decreased 3% per 66mhz increase

2600+/nForce2=12.1

2600+/KT333=12.5

2400+/nForce2=12.9

2400+/KT333=13.3

P4-2.53/850E=10.1

3D Studio MAX - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=71.2

2600+/KT333=69.1

2400+/nForce2=65.9

2400+/KT333=64.0

P4-2.53/850E=57.1

Maya 4.0.1 - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=71.2

2600+/KT333=69.1

2400+/nForce2=65.9

2400+/KT333=64.0

P4-2.53/850E=57.1

The two Lightwave 7.5 Benches - P4 wins both

(I didn't do these because the optimizations for the P4 are so extensive the Athlon XP CAN'T catchup without SSE2).

Unreal Perf. Test - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=59.8

2600+/KT333=58.1

2400+/nForce2=56.2

2400+/KT333=54.6

P4-2.53/850E=50.0

Jedi Knight 2 - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=134.1

2600+/KT333=130.2

2400+/nForce2=125.4

2400+/KT333=121.7

P4-2.53/850E=116.6

Commanche 4 - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=54.4

2600+/KT333=52.8

2400+/nForce2=50.8

2400+/KT333=49.3

P4-2.53/850E=51.2

That's 11 tests total - here's the total

Athlon XP - 6 wins (including all 3 games)

P4 - 5 wins (mostly SSE2 optimized media tools)

If you use +5% instead of +3% for the nForce2 equations, the Athlon XP 2600+/nForce2 combo takes one addtional test (SYSMark2002 I.P.) away from the 2.53ghz P4.

If these results turn out to be true, it seems that AMD would indeed take back the flagship performance crown (just barely). That is, until Intel ships the 2.66ghz P4 ;)

Calculations based on:

NVIDIA's claim for the nForce2 to beat KT333 performance by "at least a few percent". "A few" is generally defined as a small number usually between 3 and 5. For these tests, I used 3.

I graphed out the average percent increase per 66mhz increase in Athlon XP speed (based on the Anandtech numbers) and for sake of efficiency loss, ROUNDED DOWN. Therefore if the average was a 3.68% increase per 66mhz increase I used 3% for my calculations.

DISCLAIMER: These are simple math EXTRAPOLATIONS. Do NOT make purchase decisions based on these numbers, they WILL NOT necessarily resemble the actual benchmark numbers of the shipping products (once shipped :-)

SYSMark2002 O.P. - Athlon XP increased 2% per 66mhz increase

2600+/nForce2=177

2600+/KT333=172

2400+/nForce2=170

2400+/KT333=165

P4-2.53/850E=178

SYSMark2002 I.C.C. - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=278

2600+/KT333=270

2400+/nForce2=262

2400+/KT333=254

P4-2.53/850E=330

Lame MP3 Encoder - Athlon XP decreased 3% per 66mhz increase

2600+/nForce2=1.48

2600+/KT333=1.52

2400+/nForce2=1.57

2400+/KT333=1.62

P4-2.53/850E=1.65

Adobe After Effects - Athlon XP decreased 3% per 66mhz increase

2600+/nForce2=12.1

2600+/KT333=12.5

2400+/nForce2=12.9

2400+/KT333=13.3

P4-2.53/850E=10.1

3D Studio MAX - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=71.2

2600+/KT333=69.1

2400+/nForce2=65.9

2400+/KT333=64.0

P4-2.53/850E=57.1

Maya 4.0.1 - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=71.2

2600+/KT333=69.1

2400+/nForce2=65.9

2400+/KT333=64.0

P4-2.53/850E=57.1

The two Lightwave 7.5 Benches - P4 wins both

(I didn't do these because the optimizations for the P4 are so extensive the Athlon XP CAN'T catchup without SSE2).

Unreal Perf. Test - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=59.8

2600+/KT333=58.1

2400+/nForce2=56.2

2400+/KT333=54.6

P4-2.53/850E=50.0

Jedi Knight 2 - Athlon XP increased 3% per 66mhz increase

2600+/nForce2=134.1

2600+/KT333=130.2

2400+/nForce2=125.4

2400+/KT333=121.7

P4-2.53/850E=116.6

Commanche 4 - Athlon XP increased 4% per 66mhz increase

2600+/nForce2=54.4

2600+/KT333=52.8

2400+/nForce2=50.8

2400+/KT333=49.3

P4-2.53/850E=51.2

That's 11 tests total - here's the total

Athlon XP - 6 wins (including all 3 games)

P4 - 5 wins (mostly SSE2 optimized media tools)

If you use +5% instead of +3% for the nForce2 equations, the Athlon XP 2600+/nForce2 combo takes one addtional test (SYSMark2002 I.P.) away from the 2.53ghz P4.

If these results turn out to be true, it seems that AMD would indeed take back the flagship performance crown (just barely). That is, until Intel ships the 2.66ghz P4 ;)