Intel GPU speed when CPU in use

Peter Hucker
Peter Hucker
Joined: 12 Aug 06
Posts: 40
Credit: 41,915,550
RAC: 43,877
Topic 215144

Just noticed something - if you're using an Intel GPU (built into the CPU), then it runs much slower (a fifth of the speed) if the CPU cores are fully utilised.  So if you're running GPU and CPU projects on it, you need to reduce the number of CPU cores you give to Boinc.

Peter Hucker
Peter Hucker
Joined: 12 Aug 06
Posts: 40
Credit: 41,915,550
RAC: 43,877

Following on from this, is

Following on from this, is there any point?  Does an Intel GPU do more work than an Intel CPU core?

Peter Hucker
Peter Hucker
Joined: 12 Aug 06
Posts: 40
Credit: 41,915,550
RAC: 43,877

Answered my own question by

Answered my own question by running Einstein's LATeah on both.  The GPU core is 4 times faster than 1 CPU core on an i5-3570K.  Probably even more of a difference on the 8th generation chips, as their graphics is twice as fast, but a core is only 1.5 times faster.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 4,912
Credit: 30,221,090,863
RAC: 36,022,582

Peter Hucker wrote:... Does

Peter Hucker wrote:
... Does an Intel GPU do more work than an Intel CPU core?

A lot depends on how you define "more work".  The current Intel GPU tasks are really 'mini-tasks' designed to run on 'low crunching ability' mobile devices like phones, tablets, etc.  The data being analysed comes from Arecibo and is radio telescope data.  The tasks are quite different to the tasks from the Large Area Telescope (LAT) on the Fermi spacecraft which is picking up gamma-ray emissions from pulsars that don't emit at radio frequencies.  These gamma-ray emissions are the basis of the current FGRP5 CPU tasks.

Originally BRP4 (Arecibo) tasks, when being analysed on discrete GPUs, were 16 times larger than the current mini-tasks.  The current FGRP CPU tasks have (very roughly) approximately 12 times the 'work content' of the current BRP4 Intel GPU tasks.  So if yours were taking about 15 mins, you could argue that a CPU core doing an FGRP5 task in around 3 hours was performing very roughly the same amount of work as the internal GPU.  There, 12 consecutive tasks would take the same amount of time.

I have a couple of hosts with i5_3570k processors doing FGRP5 on CPU cores.  The crunch time is just over 5 hours so you could argue that you do a bit more 'work' if using the internal GPU instead of crunching FGRP5.  I prefer to do the FGRP5 tasks anyway since I find the prospect of finding the more exotic gamma-ray emitting objects is a bit more appealing than finding extra radio pulsars.

You should choose what most interests you :-).

 

Cheers,
Gary.

Peter Hucker
Peter Hucker
Joined: 12 Aug 06
Posts: 40
Credit: 41,915,550
RAC: 43,877

I'm equally interested in

I'm equally interested in both, so I was trying to go for whichever core was doing the most calculations.  Let's say that nobody used Intel GPUs, those tasks would have to go somewhere else - like onto CPUs.  So if I get more calculations done on the Intel GPU than one of my CPU cores had that project been shifted onto CPUs, then I'm being more helpful.

 

You mention phones are doing processing - doesn't that use the battery up very very quickly?  Or do people leave them on charge to process when the phone is at home?  Just how powerful are these phones anyway?  Aren't they like 1000 times less powerful than a desktop PC?  Just going by power consumption, a CPU is about 100 watts on a desktop, a phone CPU must be about 1 watt?  Otherwise the phone would become uncomfortably hot.

Peter Hucker
Peter Hucker
Joined: 12 Aug 06
Posts: 40
Credit: 41,915,550
RAC: 43,877

I've checked into it some

I've checked into it some more, analysing my 5 computers with different CPUs.  It seems some built in graphics are more powerful compared to their CPU cores, and some aren't.  And some slow down more than others when the CPU cores are fully used.  And some use the CPU core at the same time for some reason (different instruction sets for newer GPU and CPU cores causing the WU to run on both at once? - For example Asteroids on my i5-3570K with Einstein on it's GPU uses no CPU time for the GPU task.  Do the same on my i5-8600K, and the GPU task uses a whole CPU core at the same time.)  So it depends a lot on the chip.  I've tried different projects on each without much difference, the main thing appears to be the design of the CPU.  So I've set them all up differently to get the most work out of them :-)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.