I am finishing up a new computer today and started thinking that it has been quite a while since I have seen a thread discussing computer speed while running Photoshop. Over the past year some rather remarkable advances have been made in CPU and memory speed. I have to believe that PS tests done prior to this past year will not be very meaningful.
The PS Speed Test is a well used test that you can download for free.
When you post, include all the necessary system info:
Hardware: Apple, Windows, etc
OS: list the version of MAC OS or Windows (including 32-bit or 64-bit designation when appropriate)
RAM Data: Total GB, speed (actual according to BIOS reading or CPUz)
CPU Data: CPU version ID, number of cores, clock speed, multi-threads supported
Photoshop Version: indicate 32-bit or 64-bit if appropriate
Don't think we need to go into motherboard specs, but include that if you want.
P.S. Does anyone know if PS is actually coded to take full advantage of multi-core, mutli-thread operation?
Nathan, there are many areas of PS that do take advantage of multi-core but each to a different extent. Most PS image operations (like filters, etc...) will easily saturate 2 cores. A few will fully utilize 4 cores and virtually none 8 cores. Some operations like saving and opening a file are inherently serial and so derive no benefit from multiple cores.
nathanlake wrote:
P.S. Does anyone know if PS is actually coded to take full advantage of multi-core, mutli-thread operation?
Check out http://macperformanceguide.com/OptimizingPhotoshopCS5-cpu.html and go to other pages from there. Lloyd has got a few PS benchmarks for testing computers of different levels and he has a number of write-ups on all sorts of stuff including multi-core capability and scalability to even more cores.
Some aspects of PS such as file access are strictly single-core. Others are multi-core. When you get to this page http://macperformanceguide.com/PhotoshopCS5-performance-cores.html you'll read that having 12 cpu cores can even cause Ps to run slower than 6 cores !
01:01:47 (one minute, one and 47 hundredths seconds).
Adobe Photoshop Version: 12.0.3
Operating System: Windows XP 32-bit
Version: 5.1 Service Pack 3
System architecture: AMD CPU Family:15, Model:3, Stepping:2 with MMX, SSE Integer, SSE FP, SSE2, SSE3
Physical processor count: 2
Processor speed: 2210 MHz
Built-in memory: 4096 MB
Built this box in '05 and, believe it or not, it doesn't seem slow because it's the only one I use. I'm going to build again later this year so I ran this test to see what I have to look forward to.
Interestingly, with the test settings as recommended in the test procedure there is very little difference between the 32-bit and 64-bit version of PS. The difference comes when you change the settings.
Recommended test settings (States = 1, RAM slider = 100%)
32-bit average = 14.63s (3 tests)
64-bit average = 14.40s (3 tests)
This test really needs updating with a larger source image file. Obviously it's a little more troublesome to host but with the power of the new computers we're really seeing minimal gains on editing images this small. Something like an 8bit tiff from a 5DMK2 raw file would be ideal IMO. Hugely popular camera and a file that will give you a better representation of now just how fast your computer is, but how suitable.
AdrianRogers wrote:
This test really needs updating with a larger source image file. Obviously it's a little more troublesome to host but with the power of the new computers we're really seeing minimal gains on editing images this small. Something like an 8bit tiff from a 5DMK2 raw file would be ideal IMO. Hugely popular camera and a file that will give you a better representation of now just how fast your computer is, but how suitable.
Adrian,
The HardwareHeaven image in the PS test I linked above is 12 MB, a really good benchmark of your system resources.
AdrianRogers wrote:
This test really needs updating with a larger source image file. Obviously it's a little more troublesome to host but with the power of the new computers we're really seeing minimal gains on editing images this small. Something like an 8bit tiff from a 5DMK2 raw file would be ideal IMO. Hugely popular camera and a file that will give you a better representation of now just how fast your computer is, but how suitable.
I am not sure I understand what the advantage would be of using a larger image. You could easily create a test that took longer, but not sure how it would better measure the speed of the system.
