How can I do a test to stretch out my two PC's CPU performance for comparisons?

[Jack the Stripper]

I want to do a test that will stretch out both of my PC's - HP SFF i5-4570 using iGPU with 12GB of ram and i7-920 using GTX-650 with 12GB of ram, in a real world test, just CPU performance.

Basically the stuff I do every single day on the computer.

 

Internet - I need to know what I should do in terms of internet. How can I max out my CPU performance. Should I just have multiple windows open with multiple tabs of flash and youtube or is there something else that I can do to stretch out the CPU performance.

 

Gaming - Well I dont play a lot of games. The games I do play I tend to like indie games. I dont really know what indie games means, but I like cheesy, cheap looking games where the graphics aren't where its at. The only other game I play is World of Tanks, I play that one a lot. But in order to stretch out my systems, I am willing to buy a game worth $10 or $15 on Steam. I dont really want to test out the graphics beacuse my HP i5-4570 uses iGPU while my i7-920 uses a GTX-650. So maybe using gaming to stretch out CPU performance is not the key. I dont know if games are CPU intensive.

 

Any advice or suggestions on what I can do here. Or should I just run certain benchmarks. I have never run any benchmarks before so I hope it does not cost money to run benchmarks.

[Majestic]

What is the purpose of these tests? And you mean 'stress' not 'stretch'.

[Jack the Stripper]

Yes I meant stress, sorry.

 

I want to compare both of my PC's to see how they perform in a stress test. Basically I want to have a regular work load, then add stress to it and see the differences. Then when I upgrade my CPU on the older one, I can do the same test again and see if it was worthwhile or not.

[jj9987]

If you want to just stress test the system, AIDA64 is a good way to go.

For graphic stress test, I recommend FurMark or Heaven Benchmark.

 

EDIT: Saw your second post now. What you want to do, is benchmark the system, if you want to compare the differences.

Video transcoding would be one way to go, using ffmpeg or handbrake for example.

Or if you want an all-in-one kit, ASUS RealBench. It's free.

[Jack the Stripper]

Well I think graphics will be a no go because I can put a graphics card in either. I have seen LTT do the AIDA64 test, if  that is recommended for a stress test to test out my CPU's then I will look into that. Would you have a link for me to download AIDA64 and give it a go. Like I said I have never done any benchmarking before.

 

This it?

https://www.aida64.com/products/aida64-extreme

[Majestic]

3 minutes ago, Jack the Stripper said:

Yes I meant stress, sorry.

 

I want to compare both of my PC's to see how they perform in a stress test. Basically I want to have a regular work load, then add stress to it and see the differences. Then when I upgrade my CPU on the older one, I can do the same test again and see if it was worthwhile or not.

You can just look at benchmarks on the internet. 

 

http://www.anandtech.com/bench/product/47?vs=837

 

The 4570 is a tad slower than the selected 4670 (4570 isn't in the list), but it should give you a rough idea.

[Jack the Stripper]

I use CPUsuerbenchmark - http://cpu.userbenchmark.com/

For anandtech what benchmark(s) am I looking at? The thing is I have no clue whats close and whats miles away or whats overseas with all the #'s.

 

 

Picture - Which Xeon is a closer fit to X56yy?

 

 

Below - This would be a closer fit I think.

Bench Results

  Intel Core i7 920 - 2.66GHz - 1MB L2 - 8MB L3

  Intel Core i5 4570S (65W) - 4C/4T, 2.9 GHz, 1MB L2, 6MB L3

 

Product Ratings & Comparisons

3D Particle Movement: Single Threaded
Score (Higher is Better)

90.93

119.35

 

3D Particle Movement: MultiThreaded
Score (Higher is Better)

484.41

383

 

Cinebench R10 - Single Threaded Benchmark
Score in CBMarks - Higher is Better

3846

6994

 

Cinebench R10 - Multi-Threaded Benchmark
Score in CBMarks - Higher is Better

16211

25067

 

x264 HD Benchmark - 1st pass - v3.03
Frames per Second - Higher is Better

65.1

114.34

 

x264 HD Benchmark - 2nd pass - v3.03
Frames per Second - Higher is Better

25.8

37.02

 

7-zip Benchmark
32MB Dictionary - Total MIPS - Higher is Better

16283

16799

[Majestic]

4 minutes ago, Jack the Stripper said:

I use CPUsuerbenchmark - http://cpu.userbenchmark.com/

For anandtech what benchmark(s) am I looking at? The thing is I have no clue whats close and whats miles away or whats overseas with all the #'s.

It says whether it's single or multithreaded, and if higher or lower is better. Maybe take a bit longer than 5 seconds to study it.

If you're having difficulties understanding benchmark results, you have even less of a chance creating valid test results of your own.

 

Your single threaded performance is usually benificial for gaming, office and webbrowsers. Multithreaded is usually more useful in production (video editing, zipping, etc). A very simplified way of looking at things, but let's not overcomplicate it.

[Jack the Stripper]

Those websites would be listing stock i7-920 not even overclocked compared to a stock i5-4570.

I wonder how the benchmarks would be when the i7-920 is overclocked.

This is why I want a real world test.

 

Yes it says on every benchmark, but whats relative compared to the spread?

3D Particle Movement: MultiThreaded
Score (Higher is Better)

484.41

383

^difference of 100 pts, so its what? 20% = (100/500) difference, this is a linear correlation to perforance then, that is what you are saying?

 

 

Cinebench R10 - Single Threaded Benchmark
Score in CBMarks - Higher is Better

3846

6994

^42%

[Majestic]

They are just absolute numbers, not sure why you're bringing statistics into this. You run a benchmark, out rolls a score.

The higher the number, the better they are relative to each other.

 

They are different platforms, and you're comparing an 8 threaded CPU to a 4 threaded one. There will be scenario's where the i7-920 holds it's own, but in single threaded scenario's the Haswell is in a differen league. 

 

But Haswell is pretty old at this point though (2013). What sort of "upgrade" are we talking about?

[Energycore]

Passmark as a benchmark for just about everything - daily usage, content creation, games, etc

[Jack the Stripper]

i7-920 to a Xeon X56yy  either X5672,67,47,40 for $60 or X5677,87 for $100 all 4 core. Havent priced out 6 cores yet.

6 cores X5680 $150, 79,75 $125, 70 is $100 60 is $80.

 

So I would want to do tests on the i7-920, sell it, buy Xeon 4 core, test, sell it, Xeon 6 core, test, keep or sell and buy whats best. Do tests the mimic MY real world usage.

 

[Majestic]

2 minutes ago, Energycore said:

Passmark as a benchmark for just about everything - daily usage, content creation, games, etc

Passmark is built up of user samples, with high variances on the baseline. It's a last-resort benchmark, and only to be used to make very, very rough estimates.

[Energycore]

1 minute ago, Majestic said:

Passmark is built up of user samples, with high variances on the baseline. It's a last-resort benchmark, and only to be used to make very, very rough estimates.

Ah ok. There's one that's good though, one that Anandtech uses for CPUs

[Majestic]

1 minute ago, Energycore said:

Ah ok. There's one that's good though, one that Anandtech uses for CPUs

I meant the results page. Not the test itself. I thought you were referring to that.

[Energycore]

1 minute ago, Majestic said:

I meant the results page. Not the test itself. I thought you were referring to that.

I found it! It's PCMark, by the same people that do 3DMark

[Jack the Stripper]

So I use these for comparisons

- CPUuserbench mark

- anandtech

- PCMark

- AIDA64 software

[Energycore]

3 minutes ago, Jack the Stripper said:

So I use these for comparisons

- CPUuserbench mark

- anandtech

- PCMark

- AIDA64 software

Out of those, I would't ever trust userbenchmark, only PCMark and anandtech. Go ahead and run PCMark on the CPU