ImageJ

CPU or videocard?

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

4 messages Options

Василий Попков

CPU or videocard?

Greeting everyone

We are going to buy a new computer for Zeiss microscope (and its
application) and maybe some heavy image processing (in imageJ or others
application for analyzes images from microscopes). So, the question is:
what is more important: good processor or video card? I am not so sure...

Thank you,
Vasily

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

Gary Sellani

Re: CPU or videocard?

You may want to consider the possibility of using OpenCL with imagej.

‎http://developer.imagej.net/opencl

Original Message
From: Василий Попков
Sent: Wednesday, December 17, 2014 3:29 AM
To: [hidden email]
Reply To: [hidden email]
Subject: CPU or videocard?

Greeting everyone

We are going to buy a new computer for Zeiss microscope (and its
application) and maybe some heavy image processing (in imageJ or others
application for analyzes images from microscopes). So, the question is:
what is more important: good processor or video card? I am not so sure...

Thank you,
Vasily

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

Michael Sarahan

Re: CPU or videocard?

Hi,

Unless you have identified areas of interest with known good GPU-supported
libraries, or you're willing to get into a lot of programming to do things
yourself, I'd say the CPU is a more worthwhile investment at this point.

This said, consumer-grade CPUs and GPUs are generally more than adequate
for research work, and it's quite affordable to get both. A 300$ Intel i7
processor with a 300$ NVidia GTX 970 would be a quite capable system in all
respects - total cost ~$1200 with all the other necessary hardware (power
supply, motherboard, case, memory). AMD CPUs are slower per core, but
would offer you more cores for potentially much less money. Consider how
much your current applications use multiple processors - if your current
applications are efficient at using many processors, then this may be a
good path. AMD graphics cards are also competitive in price, but limit you
to OpenCL applications, while NVidia cards let you use CUDA or OpenCL.

Hope this helps.
Michael

On Wed, Dec 17, 2014 at 9:05 AM, Gary Sellani <[hidden email]> wrote:

>
> You may want to consider the possibility of using OpenCL with imagej.
>
> ‎http://developer.imagej.net/opencl
>
>
> Original Message
> From: Василий Попков
> Sent: Wednesday, December 17, 2014 3:29 AM
> To: [hidden email]
> Reply To: [hidden email]
> Subject: CPU or videocard?
>
> Greeting everyone
>
> We are going to buy a new computer for Zeiss microscope (and its
> application) and maybe some heavy image processing (in imageJ or others
> application for analyzes images from microscopes). So, the question is:
> what is more important: good processor or video card? I am not so sure...
>
> Thank you,
> Vasily
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

Gary Sellani

Re: CPU or videocard?

Also consider the Intel Xeon E3 1200 series. The line is more like the "i" series than Xeon, but it supports more memory and ECC at that. Dell Power Edge‎ or something from Supermicro. Be sure to get the current rev of the CPU, which is V3.

Currently Nvidia benchmarks better on OpenCL, so I really don't see a reason to go to AMD. That said, AMD will be at 20nm next year, so things may change. Nvidia and AMD have been hampered by a lack of a fine geometry process. They have been selling essentially the same chips for 3 years.
Original Message
From: Michael Sarahan
Sent: Wednesday, December 17, 2014 3:05 PM
To: [hidden email]
Reply To: [hidden email]
Subject: Re: CPU or videocard?

Hi,

Unless you have identified areas of interest with known good GPU-supported
libraries, or you're willing to get into a lot of programming to do things
yourself, I'd say the CPU is a more worthwhile investment at this point.

This said, consumer-grade CPUs and GPUs are generally more than adequate
for research work, and it's quite affordable to get both. A 300$ Intel i7
processor with a 300$ NVidia GTX 970 would be a quite capable system in all
respects - total cost ~$1200 with all the other necessary hardware (power
supply, motherboard, case, memory). AMD CPUs are slower per core, but
would offer you more cores for potentially much less money. Consider how
much your current applications use multiple processors - if your current
applications are efficient at using many processors, then this may be a
good path. AMD graphics cards are also competitive in price, but limit you
to OpenCL applications, while NVidia cards let you use CUDA or OpenCL.

Hope this helps.
Michael

On Wed, Dec 17, 2014 at 9:05 AM, Gary Sellani <[hidden email]> wrote:

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html