Polyloop - 3D & 2D Forums - View Single Post

andywright · 14th February 2006, 09:49

You are right, AMD has Hypertransport not hyperthreading. The FX-55 is a single core processor and therefore will show up as one processor. If you want dual processing, you need one of AMD's Dual processors.

Note - a hyperthreaded P4 will still only perform equivelent to a similar specd AMD processor, as it dosen't have the cache and pipelines to act as true 2 processors, even their new P4 dual core dosen't improve as much as it could as it has a shared cachE memory for the two processors- and runs very hot too !

If you want the equivelent of your FX55, its the new FX60 which is dual core, but extremely expensive (but easily outperforms the new P4 dual core extreme version - by 10-20% in some tests).

The other option is to go for a standard dual processor such as the AMD 4800+ which has very good reviews and will probably halve your render time over even the the FX55 and costs around £450 in the UK. Ensure you get the 1Mb+1Mb cache version as this will be better for memory intensive work like rendering (there is a cheaper 512Kb + 512Kb version available)

See this comment below for comments on 3D modelling:

Arguably, what works for one architecture may not necessarily be the all encompassing solution for the other architecture. Suffice it to say that as a rule of thumb multiple threads, running simultaneously as in the case of multimedia encoding, will experience more improvements than applications that are targeting single threads or even use the same resources for all threads running. Cases in point are 3D rendering programs such as Caligari TrueSpace or 3dsmax that may show logical processors in the form of independent scanlines, whereas the overall runtime is completely oblivious to the number of such units. The benchmark results will always be the same. No rules without exceptions, and the best known or publicized benchmark showing results to the contrary is Maxon's Cinebench 2003. In that case, HyperThreading makes a huge difference, at least for the raytracing

This is from :

http://www.lostcircuits.com/cpu/amd_x2/

14th February 2006, 09:49	#8 (permalink)
andywright Extrusion Join Date: Jan 2006 Location: Oxfordshire, UK Posts: 154	You are right, AMD has Hypertransport not hyperthreading. The FX-55 is a single core processor and therefore will show up as one processor. If you want dual processing, you need one of AMD's Dual processors. Note - a hyperthreaded P4 will still only perform equivelent to a similar specd AMD processor, as it dosen't have the cache and pipelines to act as true 2 processors, even their new P4 dual core dosen't improve as much as it could as it has a shared cachE memory for the two processors- and runs very hot too ! If you want the equivelent of your FX55, its the new FX60 which is dual core, but extremely expensive (but easily outperforms the new P4 dual core extreme version - by 10-20% in some tests). The other option is to go for a standard dual processor such as the AMD 4800+ which has very good reviews and will probably halve your render time over even the the FX55 and costs around £450 in the UK. Ensure you get the 1Mb+1Mb cache version as this will be better for memory intensive work like rendering (there is a cheaper 512Kb + 512Kb version available) See this comment below for comments on 3D modelling: Arguably, what works for one architecture may not necessarily be the all encompassing solution for the other architecture. Suffice it to say that as a rule of thumb multiple threads, running simultaneously as in the case of multimedia encoding, will experience more improvements than applications that are targeting single threads or even use the same resources for all threads running. Cases in point are 3D rendering programs such as Caligari TrueSpace or 3dsmax that may show logical processors in the form of independent scanlines, whereas the overall runtime is completely oblivious to the number of such units. The benchmark results will always be the same. No rules without exceptions, and the best known or publicized benchmark showing results to the contrary is Maxon's Cinebench 2003. In that case, HyperThreading makes a huge difference, at least for the raytracing This is from : http://www.lostcircuits.com/cpu/amd_x2/ __________________ Athlon 64 X2 4400+ Win XP Pro, 2 GB DDR 400 MHz Geoforce 6800GT Xtreme 256 MB. Carrara 5 Pro, Hexagon 2.01, Lightwave 8.5 Poser 6, DAZ Studio, Vu 5 Infinite, Bryce 5.5 www.wright-site.co.uk