Explain to me this. If THG is comparing the top of the line new P-M 2.13Ghz CPU against the Athlon64/FX and the P4, why does he always use the older 130nm cores? The P-M and P4 use the latest 90nm chips and fastest chipsets, but he uses usually a slower board and older CPUs. Now why do I complain? Well a magor part of this test is power draw and the 90nm Venice and Winchester cores use MUCH less power (about 1/3-1/2 the power). Performance wise they also use SSE3 and a faster memory controller. He also uses an ASUS A8N-SLI. This is not the fastest NF4 board out there, but also it generates a ton of heat. I bet the chipset draws more power than the CPU would at idle. With the stock cooler at 8000RPM it still runs about 70'C. I replaced the cooler on mine and it is still around 60'C (but much quieter at 2500). If you are not running SLI, then why use this power hog. Heck even a ATI, VIA, or SiS chipset can compete just as good and likely draw half the power.
So lets see... You take that P-M running at 66w and compare it to THG results and the Athlon64 3500+ looks really bad at 201w. Now you drop the GeForce6800GT and you take off 40w, the chipset is likely another 20w and the CPU is likely at least 40w. Now you are looking at 100w. Does that P-M look that good now? That puts the Athlon64 CPU and chipset drawing only about 40w more than the P-M system. Now also take into acount that the P-M board and CPU cost a lot more and things start to look even better for AMD. He also pointed out that playing a DVD was a 25% load. Well with Cool-n-Quiet on on my Athlon64 3500+ Winchester core it sits at 1Ghz and 1.1v and still uses about 10% load and the stock all aluminum heatsink (which is much smaller than the older Athlon64 heatsinks) can also run with the fan turned off and tops out at about 50'C. To make that more impressive, that is in a closed case, which THG did not do. Heck I even run the 120mm 1600RPM Antec fans at about 5v. These fans run at about 850RPM. The case tesmps float around 33'C (I wanted quiet more than cool). He runs the P-M in open air at likely 21'C in his A/Ced lab.
He also will not touch 64 bit for any of these tests. Windows64 bit is out and he should have no problems getting drivers. The performance got games is pretty close to 32 bit, but in a lot of other tasks the Athlon64 would see huge jumps in performances. I have seen many applications show 5-20% quite often with some running twice as fast. Even the Yonah will not get 64 bits until the end of next year.
Now take that a step forther. He used the fastest P-M at 2.13Ghz which costs about $700. For that price you could get an Athlon64 X2 4400+ easy. Now people say the P-M overclocks great... well most will do 2.3-2.5Ghz, but only a few have actually made it into the 2.6-2.7Ghz on air cooling. Well the 4400+ overclocks quite well too. AMDZone got theirs to 2.8Ghz with the air cooling and only a tiny votage bump (1.425 if I recall right).
The P-M clock for clock is VERY close to the Athlon64, but in 64 bit applications the Athlon64 will pull ahead quite easy. If you run thread applications or multiple applications the X2 will smoke the P-M (which is a fare comparison based on cost). If you compare overclocking, likely the Athlon64 E6 cores will still overclock better (on average). Then there is SLI. Now P-M system has SLI. ASUS has no BIOS update to support the addaptor in an NF4-SLI P4 board either, so for gaming performance you could compare a P-M 2.13Ghz with a GeForce 6800GT against a similar priced Athlon64 3500+ with 2 GeForce 6800GTs in SLI. Which do you think will win in games now?
Overall I would say this was an unrealistic slanted article that showed nothing more than that yes the P-M is a good low power CPU that performs well, but the comparisons were a joke. If I was making a HTPC, then a P-M would make a great choice, but otherwise it is not worth the added cost and with more applications going 64 bit an threaded over the next few years, the P-M will become obsolete much faster.
So lets see... You take that P-M running at 66w and compare it to THG results and the Athlon64 3500+ looks really bad at 201w. Now you drop the GeForce6800GT and you take off 40w, the chipset is likely another 20w and the CPU is likely at least 40w. Now you are looking at 100w. Does that P-M look that good now? That puts the Athlon64 CPU and chipset drawing only about 40w more than the P-M system. Now also take into acount that the P-M board and CPU cost a lot more and things start to look even better for AMD. He also pointed out that playing a DVD was a 25% load. Well with Cool-n-Quiet on on my Athlon64 3500+ Winchester core it sits at 1Ghz and 1.1v and still uses about 10% load and the stock all aluminum heatsink (which is much smaller than the older Athlon64 heatsinks) can also run with the fan turned off and tops out at about 50'C. To make that more impressive, that is in a closed case, which THG did not do. Heck I even run the 120mm 1600RPM Antec fans at about 5v. These fans run at about 850RPM. The case tesmps float around 33'C (I wanted quiet more than cool). He runs the P-M in open air at likely 21'C in his A/Ced lab.
He also will not touch 64 bit for any of these tests. Windows64 bit is out and he should have no problems getting drivers. The performance got games is pretty close to 32 bit, but in a lot of other tasks the Athlon64 would see huge jumps in performances. I have seen many applications show 5-20% quite often with some running twice as fast. Even the Yonah will not get 64 bits until the end of next year.
Now take that a step forther. He used the fastest P-M at 2.13Ghz which costs about $700. For that price you could get an Athlon64 X2 4400+ easy. Now people say the P-M overclocks great... well most will do 2.3-2.5Ghz, but only a few have actually made it into the 2.6-2.7Ghz on air cooling. Well the 4400+ overclocks quite well too. AMDZone got theirs to 2.8Ghz with the air cooling and only a tiny votage bump (1.425 if I recall right).
The P-M clock for clock is VERY close to the Athlon64, but in 64 bit applications the Athlon64 will pull ahead quite easy. If you run thread applications or multiple applications the X2 will smoke the P-M (which is a fare comparison based on cost). If you compare overclocking, likely the Athlon64 E6 cores will still overclock better (on average). Then there is SLI. Now P-M system has SLI. ASUS has no BIOS update to support the addaptor in an NF4-SLI P4 board either, so for gaming performance you could compare a P-M 2.13Ghz with a GeForce 6800GT against a similar priced Athlon64 3500+ with 2 GeForce 6800GTs in SLI. Which do you think will win in games now?
Overall I would say this was an unrealistic slanted article that showed nothing more than that yes the P-M is a good low power CPU that performs well, but the comparisons were a joke. If I was making a HTPC, then a P-M would make a great choice, but otherwise it is not worth the added cost and with more applications going 64 bit an threaded over the next few years, the P-M will become obsolete much faster.