Need Help Proving New Work Computers Aren't Powerful Enough

[Daffodilio]

Background Info:
I am a fairly new Systems Admin to a company of about 500 people. I was brought into an existing project to replace some old computers. We replaced Pentium 3 and Pentium D computers with new FoxConn nettops that only had a 1.5 GHz dual core Atom processor and 2 GB of RAM. It wasn't until we were well into the project that I finally got all of the details and noticed that the computers were pretty slow. Now that we have replaced about 70 computers, the users are complaining like crazy about various performance issues. I told my boss that I don't think the new computers are good enough to be a person's primary work computer and we need to investigate this. Now it's my job to formally research and test these to give my opinion on whether they are ok to keep, upgrade, or if they need to be replaced. This decision could cost us $100,000 so I'm pretty nervous. I'm not very strong with hardware issues, my specialty is software.

Question:
What suggestions do you have for testing? My suspicions are that one or all of these things aren't good enough: 1) Processor 2) Video 3) RAM
Is there software that I could run that could generate a report that would give me conclusive results? Is there any real world testing that I can do to prove what the true problem is, and if it can be fixed/upgraded? I'll include a link to the machine itself, as well as a link to the video decoder that we installed in them. Thanks in advance for your help!!!

http://www.foxconnchannel.com/product/Barebones/NT330i/index.html#

http://www.broadcom.com/products/Consumer-Electronics/Netbook-and-Nettop-Solutions/BCM70012

 

[sadams04]

Dual core Atoms are a step in the right direction from a single core, but they are not going to work for a lot of business setups. If only doing work in MS Office, Internet Explorer and a few light apps, then they may be fine. For what you are after, you will need to capture performance stats to support your recommendation (if there is one). Perfmon is your friend... If you can have a few people reproduce what they are doing to see poor performance, you can set up perfmon to capture stats on the CPU, memory and disk to see what may be maxing out. LOTS of options in perfmon to captuer data so a Google search is probably in order to find an example of what, and how, to log.

 

[coilaman]

Background Info:
I am a fairly new Systems Admin to a company of about 500 people. I was brought into an existing project to replace some old computers. We replaced Pentium 3 and Pentium D computers with new FoxConn nettops that only had a 1.5 GHz dual core Atom processor and 2 GB of RAM. It wasn't until we were well into the project that I finally got all of the details and noticed that the computers were pretty slow. Now that we have replaced about 70 computers, the users are complaining like crazy about various performance issues. I told my boss that I don't think the new computers are good enough to be a person's primary work computer and we need to investigate this. Now it's my job to formally research and test these to give my opinion on whether they are ok to keep, upgrade, or if they need to be replaced. This decision could cost us $100,000 so I'm pretty nervous. I'm not very strong with hardware issues, my specialty is software.

You guys made a big mistake getting those Atom based machines. 1.5GHz dual-core Atom is the same performance level as the 3.4GHz Pentium 4. The problem is that Atom performance is for 2 cores whereas Pentium 4 is a single core machine.

A lot of software is not optimized for dual-core machines, so every time somebody uses a web browser, for example, they will be utilizing only one core in the Atom processor and then you get the performance level of a low-end Pentium 4 or even a higher-end Pentium 3. So it seems that you guys didn't upgrade at all. Your new machines are slower than some of the Pentium D based machines you just replaced. Stay away from Atom processors and get yourself at least machines with new and current Dual-Core Pentium E (5000 series) or even Celeron E (3000 series) processors.

Those will have at least 3 times better performance benchmarks than your Atom processor. The current Celeron E (3000 series) processors are very good, affordable and true workhorses. Celerons are way better these days than they used to be.

 

[Daffodilio]

You guys made a big mistake getting those Atom based machines. 1.5GHz dual-core Atom is the same performance level as the 3.4GHz Pentium 4. The problem is that Atom performance is for 2 cores whereas Pentium 4 is a single core machine.

A lot of software is not optimized for dual-core machines, so every time somebody uses a web browser, for example, they will be utilizing only one core in the Atom processor and then you get the performance level of a low-end Pentium 4 or even a higher-end Pentium 3. So it seems that you guys didn't upgrade at all. Your new machines are slower than some of the Pentium D based machines you just replaced. Stay away from Atom processors and get yourself at least machines with new and current Dual-Core Pentium E (5000 series) or even Celeron E (3000 series) processors.

Those will have at least 3 times better performance benchmarks than your Atom processor. The current Celeron E (3000 series) processors are very good, affordable and true workhorses. Celerons are way better these days than they used to be.

I kind of had that feeling about the big mistake. I don't know much about hardware, but I know that they put Atom processors in netbooks. I certainly wouldn't give a user a netbook as their primary machine! Thanks for your great explanation! It really makes sense now why users complain that they are a downgrade, and also why some things (IE for example) run more slowly than others. Thanks again for the helpful info!

 

[Daffodilio]

Dual core Atoms are a step in the right direction from a single core, but they are not going to work for a lot of business setups. If only doing work in MS Office, Internet Explorer and a few light apps, then they may be fine. For what you are after, you will need to capture performance stats to support your recommendation (if there is one). Perfmon is your friend... If you can have a few people reproduce what they are doing to see poor performance, you can set up perfmon to capture stats on the CPU, memory and disk to see what may be maxing out. LOTS of options in perfmon to captuer data so a Google search is probably in order to find an example of what, and how, to log.

I had thought about Perfmon, but grazed by it since I wasn't familiar with it. I'll do some additional reading. Thanks!

 

[Daffodilio]

Best answer selected by Daffodilio.