Geekbench Your Computer!

[FanIn80]

Originally Posted by photon

He's running it on the same PC though, it's scoring faster inside the VM on OSX than it's scoring in Windows 7 on the bare metal.

Unless it's some kind of 32-bit 64-bit thing, maybe the virtual machine even though it's running the 32-bit version of geekbench the VM software is running 64-bit and taking advantage of some kind of optimization that the 32-bit version of geekbench in Windows 7 isn't.

It could just be a difference in the way OS is handling things. I would tend to believe the difference is on the OS side, rather than a problem with the tool itself.

[FanIn80]

Originally Posted by Jedi Ninja

Exactly. And the OS/X VM is 32 bit also, FYI.

You have to be really careful when constructing tests like this as you can run the same source code through two different compilers and end up with different machine code. It's possible that OS/X takes better advantage of certain processor features, but I'd take some convincing. I did try checking out the web site to see if there was more nuts-and-bolts information about the benchmarks, but the information is sparse, other than a blog page that appears to be exclusively Mac oriented. You really want a product like this to be open source.

Besides, this benchmark is so narrow in scope, I'm not sure how it translates to real-world end-user differences in speed.

Unfortunately, there aren't many cross-platform benchmarking tools out there.

The good news about GeekBench is that it's been out for a couple years and is updated regularly. Also, a look at their overall top results shows rather equal PC and Mac representation: http://browse.geekbench.ca/geekbench2/top

Edit: Here's their Twitter account, which has some interesting tweets. There's even one from a Microsoft SQL Server MVP who says that it's very useful for preparing for transaction processing workloads in Windows Server.

Edit: Whoops, forgot the link: http://twitter.com/primatelabs

[Hanni]

I'm not up to speed on sandy bridge, now that I've done some checking it makes sense how the new architecture can account in the difference as geekbench is testing speed and not overall system specs. The amount of memory you have for example has no bearing on the test.

Now I have to resist the urge to do a bunch of costly upgrading.

[Hack&Lube]

You're not going to find a great universal benchmark. The only real solution for Apples to Apples is for people to run the same OS and the same benchmark but I have a feeling Geekbench isn't properly making use of the 256-bit advanced vector extensions in Win7 SP1 which will greatly affect FPU performance.

VM ware will bring in its own complications that will trip up any benchmark, they may make use of processor virtualization features or optimizations or shortcuts because it's a virtual machine running in resident memory and not directly off the hardware, with overhead and another abstraction layer.

Ultimately canned benchmarks are still canned benchmarks and there is no real accounting for the accuracy and fairness of any canned benchmarking system. Manufacturers and developers can optimize their products for doing well on specific benchmarks that are deceptive when it comes to real world results.

It's just something to do for fun and there's nothing wrong with that. Just don't take it too seriously or get bent up about needing to upgrade your system because someone gets a few more points in a canned benchmark than you. More often than not, it does not translate to real world needs and performance.

Hanni, unless you are doing a lot of complex calculations or video processing or really want the integrated GPU for some reason, there's no need to switch to SB from your first gen i7. At a higher clock, your CPU will be just as good or better.

[Hanni]

I'm not actually considering upgrading anything. There is just that small part of you that thinks oh great now I'm obsolete. It's like getting passed on the highway, it shouldn't matter but somehow you feel like this person is questioning your manhood, how dare they pass you.

For me I would look at an SSD long before upgrading motherboard and cpu. I think an SSD is about the only thing that would show a really noticeable improvement in overall performance.

[FanIn80]

Originally Posted by Hanni

I just had a little issue with a certain comment (that has been edited out I see) that seemed to have no other purpose than to entice an OS argument.

You said one thing bothered you, but I had no idea what it was, so I took two things out of my post. One was a joke making fun of myself, and the other was a comment about *nix OSes being used for more processor-driven tasks like web servers and video editing.

You're suggesting that I was intentionally trying to pick a fight, but I think the fact that I immediately tried to remove whatever it was that could have possibly offended you is proof that I wasn't.

Trust me dude, these aren't the droids you're looking for.

[Hanni]

nm, pmed instead.

[FanIn80]

Originally Posted by Hack&Lube

there's no need to switch to SB from your first gen i7. At a higher clock, your CPU will be just as good or better.

From a "real world" standpoint, I agree there's little reason to upgrade, but the new arcitecture really does make a huge difference even with a lower clock speed. Look at last year's 17" i7 compared to this year's 13" i5.

http://www.primatelabs.ca/blog/2011/...rks-early-2011