Excel speed test comparison against an FX-8350

[Hail The Brain Slug]

I think there's probably a huge discrepancy based on version of Excel being used. Some 3570k scored 21 seconds, now my 4.5 GHz 5960x scores 35 seconds - But I'm using the 64 bit version of Excel 365, and he probably used Excel 2010.

 

[John Tauwhare]

Agreed, consistency of OS and Excel version is an issue with this. The model was originally written in Excel 2003. I was using Win7 and Excel 2007 when I started the thread. Now I'm using Win10 and Excel 2010. Did you turn off hyper-threading in the BIOS or Excel Options (HT is a big drag on the macro)?

 

[Hail The Brain Slug]

Agreed, consistency of OS and Excel version is an issue with this. The model was originally written in Excel 2003. I was using Win7 and Excel 2007 when I started the thread. Now I'm using Win10 and Excel 2010. Did you turn off hyper-threading in the BIOS or Excel Options (HT is a big drag on the macro)?

Interesting, I'll have to play with that on my 4770k system and see what combination scores the best.

 

[John Tauwhare]

I got this with my 4670k

 

[Hail The Brain Slug]

I got this with my 4670k

Interesting - I got 21.99 with my 4.3 GHz 4770k with HT off in W10 CU with Excel 365 64 bit.

 

[John Tauwhare]

Nice. Consistent with my 22.6 @4.3. Nothing wrong with Office 365 - I've had good results with it on laptops.

 

[Zstream]

Agreed, consistency of OS and Excel version is an issue with this. The model was originally written in Excel 2003. I was using Win7 and Excel 2007 when I started the thread. Now I'm using Win10 and Excel 2010. Did you turn off hyper-threading in the BIOS or Excel Options (HT is a big drag on the macro)?

No, it's a badly written benchmark.

 

[cyrusfox]

No, it's a badly written benchmark.

You must be unfamiliar with VBA macros, as they also are all unoptimized. This benchmark is perfect as it represents exactly what is mostly ran on excel(Which is code made to just work, only sometimes optimized for speed). Someone who spends most their time working with these custom excel subroutines/functions appreciates how accurately this test reflects their true workload and is then able to avoid processors whose performance is sometimes 4 times slower than a comparable chip. I am really excited to see how Ryzen has stepped up its IPC and made AMD competitive, have yet to see how it performs on this benchmark.

Show me a better Excel benchmark that reflects true macro performance? Until then, this is my go to for comparisons to show coworkers and clients on what is slowing down there work.

 

[Zstream]

You must be unfamiliar with VBA macros, as they also are all unoptimized. This benchmark is perfect as it represents exactly what is mostly ran on excel(Which is code made to just work, only sometimes optimized for speed). Someone who spends most their time working with these custom excel subroutines/functions appreciates how accurately this test reflects their true workload and is then able to avoid processors whose performance is sometimes 4 times slower than a comparable chip. I am really excited to see how Ryzen has stepped up its IPC and made AMD competitive, have yet to see how it performs on this benchmark.

Show me a better Excel benchmark that reflects true macro performance? Until then, this is my go to for comparisons to show coworkers and clients on what is slowing down there work.

No, what you're showing is what process works best in a mainly single threaded workload. In fact, it's doing a single sheet of calculations and moving on. If you want to know more, look at: http://analystcave.com/excel-vba-multithreading-tool/

 

[John Tauwhare]

No, what you're showing is what process works best in a mainly single threaded workload. In fact, it's doing a single sheet of calculations and moving on.

Which is exactly what's intended..
It doesn't have to be great Excel, in fact it shouldn't be; the objective is to assess CPU performance in a typical highly single threaded Excel workload, and for that it has proven to be fit for purpose.

 

[Zstream]

Which is exactly what's intended..
It doesn't have to be great Excel, in fact it shouldn't be; the objective is to assess CPU performance in a typical highly single threaded Excel workload, and for that it has proven to be fit for purpose.

Images moved from Photobucket to postimg.org today on this thread (yawn).

So you agree it's a poorly written benchmark.. gotcha. It's not a well thought out excel benchmark.