There’s a lot of discussion on both issues here in the forum.
Have a browse or do a search,
I’d give you a link but I’m off.
A couple of things to keep in mind about going to a 64-bit OS. First, not all devices have 64-bit drivers so you run into compatibility with hardware. Second, not all software works properly under a 64-bit OS. So both hardware and software compatibility can be a problem.
One the positive side, you can install tons of memory which can really speed things up all by itself. But that won’t do you much good if you can’t connect a printer or scanner, or your application doesn’t function correctly.
I would stick to a 32-bit OS (specifically Windows XP) and go for the quad-core processor.
If you can afford it go for the quad core.
However only some CS3 functions are multithreaded and that is for two cores. The quad core has a massive amount of onboard cache/memory that can be allocated to the one or two cores that are actually doing any work and is probably the reason quad cores test faster than dual cores although rarely would more than two of the cores actually be doing anything. Stay with a 32 bit OS, preferably XP. Vista is very slow with any disc access operation, partly negating the speed of CPUs and fast RAM, is not compatible with many multimedia and CD/DVD authoring programs, and is reported to unload monitor calibration info every time it flashes one of its idiot warning boxes. CS3 runs well on Vista. However some device drivers, including printers (e.g. older Epsons) are not as functional in Vista. With a 32 bit OS there is absolutely no point in getting more than 2 gbs RAM. Save that money and buy at least one, if not two, additional hard drives.
IS money a factor? and do you use more than one program at a time? Quads will help when multi tasking even if the pgm isn’t threaded properly and you have several pgms then it will help a lot. If money is a consideration then you will get better value out of a Dual at the moment, they are HEAP cheeper and perform not very far behind.
In most tests a Quad out performs a Dual by any where between 0-15% but are around 30% more expensive (depending on models). And in a few cases they are slower.
BRETT
Ok, here’s the deal. Whether a quad will outperform a dual is in the details, that is, how well the processor is married to the mobo and Mch and Ich (Northbridge and Southbridge). What versions are we talking about? Is it Intel or third party silicon, and how good is it?
There are certain limits on just how good the quad is going to be, so for my money, the Core 2 Duo E6600 or 6800 is just about optimum for now. But that’s only one opinion.
The same considerations apply to those as well.
I recently upgraded my PS-machine. A new MB and an Intel Core2 Quad processor (Q6600).
I have no trouble at all running Ps CS3, Bridge, Lightroom etc.
To max out the processor, I tested with the Radial Blur filter. In the Windows Performance window I could see all four processors were deployed.
However, my new processor won’t change the climate back to what it was, as it consumes well over a 100 Watts.
Rob
how well the processor is married to the mobo
that’s why i stick w/intel on intel. 🙂
But that’s only one opinion.
two.
I’m using a Q6600 with an Intel BadAxeII, things do not really change for PS’s use of processors. When I am using a filter, the CPU usage is all over the map, it normally averages about 35% usage but you can clearly see ‘spikes’ that are at 80%, if the average usage gets around 50%+, I can expect a freeze or a crash. It has always been that way.
Compared to Maya rendering, Maya pins all four CPUs at 100% usage, and does not freeze or crash during the rendering.
PS is not all it is cracked up to be for the optimization of CPU utilization. The engineers are not to blame, it is just that full optimization is not ‘sexy’ and does not have a ‘wow’ factor to increase sales so they cannot devote the time needed to correct these problems.
Personally I would like Adobe to step back, take a deep breath and create a working group to solve these less than stellar issues.
Good post, Conrad. I did 8 months FSB validation for Intel and the article is exactly right, especially comparing video to graphics performance. The bottleneck is at the MCH, where data is being moved from all the cores through one device. Think of it as a vast area needing 4 airports to serve the needs, but a single mountain pass which must be used for both departure and approach to those terminals.
Having two methods of operating multi-core operations, ie AMD, with memory controller on the chip, vs Intel with it’s FSB and external MCH, does pose problems for a programmer, as the behavior of each system is different. So, how do you run on both?
