Terrible increase in cpu utilization.

The new 2.0 client uses anywhere from 10 to 20 times as much cpu processing as the previous client. It has now gone from a nearly invisible thread to something generating a noticeable number of context switches.

Are there plans to decrease this?

(should clarify this is on the OSX client)

The 2.0.1a client should fix this.

Any improvements we see a long the way, we do, so it should get a smaller footprint in time.

The biggest resource hog will be the networking statistics, so if you don’t find the current state acceptable; disabling networking stats will save you a lot of cpu cycles.

I have network and uptime completely disabled in the settings currently.

After going back and forth with the settings and watching closely, there is very little difference in amount of cpu utilization from enabling or disabling these two options. With just the keyboard information the new client still takes at a minimum 10 times the cpu cycles of the old one. Typically it spikes between 10-20.

I realize this is a pretty small thing to most people. But to me, the things I am willing to leave running all the time and doing some sort of computing must be finely tuned. That little bump up in utilization from the new client is just enough to get my laptop fan to spin up sometimes. /overlyspecificenvironment

Just installed the latest update. Whatpulse is now using even more CPU than

I do see to have the same behaviour of the client on either of my 3 Linux machines where the lowest (!!!) amount of CPU taken, according to top is 3% and that is then the machine is idling and doing absolutely nothing of interest.

Furthermore, the memory-reservation for WhatPulse is really what worries me; please have a look at both attachments where more than 1.5GB of RAM is mapped to/on the WhatPulse-client.

It appears that, depending on system setup, WP is indeed a resource hog. If you have a system with less RAM on a PC (<4 GB), WP would be a memory hog. If you have a single core CPU (<3 GHz), WP would eat that for dinner. We’ll have to set up min and recommended requirements to run the client. It goes without saying that every system is different. Here is my system:

Six core AMD FX-6120 3.3 GHz (20 GHz total)

In other words, a powerful machine that has no trouble running multiple windows. I don’t mean to brag, but my machine is close to top of the line. Most systems are 6-12 GHz (Dual core or Quad core) processor and 4-8 GB RAM.

Processor speed does not work like that. You can’t just take ‘cores * speed = total’. If I had a processor that was a single core that ran at 20 GHz, I’d be laughing all the way to the bank - seriously.

The speed you read is the single threaded speed, and the amount of cores gives reasoning for multithreaded scenarios. Not all software is written to take advantage of multithreading, and some software cannot - it requires code that is parallel (i.e. independent) from each other to be run at the same time. Then you get the speed up. If you can’t parallelise your code, then you don’t get the benefit. Hence why processor manufacturers do not quote a ‘total speed’ like you have done.

Since 2.0.1a, I’ve not seen an increase in CPU usage like I did with 2.0, on any of my systems (from netbook to 8-core dual processor E5-2690 2.9 GHz with hyperthreading… or 92.8 GHz by WOWVern’s math :slight_smile:

My Windows 7 is configured that way, I think. My CPU usage never goes above 15%. Every configuration is different.

FWIW, I’ve noticed after an automatic pulse, WP’s CPU usage goes to 25% and stays there until I restart it. i7-2600k @ 4.2 w/ 16 GB

There’s absolutely no reason for this level of memory and cpu usage given what the software is doing. I only have 2 cores and 8 gig on this laptop, but when I see the WhatPulse client pull more cpu than a typical freebsd vmware instance there is something seriously broken.

We are aware that the CPU usage is too high in the Linux and Mac version, we are working on it.

I have had Whatpulse 2.0.1 stuck on 100% CPU usage twice now. Running windows 7 pro 64bit. It has happened around the same time mircstats has been running so I’m not sure if it is linked to that. And it doesn’t happen every time, it’s totally random.