Tom's Hardware Wants You: CPU Benchmarks 2011/12

[paranoidmage]

When testing games, make sure to test at decently high resolutions and quality settings. 1680x1050 or 1920x1080 and mid to high quality.

It's more important to see actual differences and real performance than exaggerated differences just for the sake of finding a big change in FPS.

The biggest problem in the past was the lack of updates to the charts. They're useless unless they are updated with the new parts that we want to compare.

 

[Guest]

Super Pi
This test we show the comparative single core architecture performance

 

[snowluck2345]

I vote for GNU R, I used to use matlab, but I converted so the students wouldnt have to buy it. Ilike it more now. R presses all 80 of my cores to xeon 8850 cores to the max.

 

[Guest]

Benchmark at the same time: World of Warcraft + 1080p File Playback + 15 tab open Browser + a 3DS Max render.

Something to really show off how the cpu handles a heavy multitasking situation; and I have actually done all 4 of those at the same time on multiple occasions.

 

[wutname1]

Starcraft 2 and Battlefield 3 is a must for any test of any component

 

[Guest]

A "Pro Tools" professional audio benchmark listing how many instances of software synths and/or effects that can run stably would be helpful to me.

 

[agnickolov]

A Visual C++ code compilation benchmark please. I saw one performed this year in a review, but don't remember the details. I've been asking for this for years...

 

[Guest]

It may sound silly but I have to try this one: can you use The Sims 3 to benchmark low to low mid-range GPUs, including Intel HD 2000/3000 and AMD Llano? I'd post the results on TS3 forums and hopefully see less "can this run TS3" threads. Too many "simmers" buy low-end laptops with old IGPs and wonder why they experience problems ... Thank you

 

[Guest]

SCII supports two threads, as does crysis 1.

 

[maddy143ded]

itunes and cinebench suite are all good, but we should have some more trans-coding tests , using just the CPU suites. RIPBOT is one of the best free trans-coder out there , and there are many other such tools. choose one of the freely available softwares also for a test bench

 

[tomskent]

I would like to see Overclocked cpu's on the drop down list. The vast majority of people who buy a 2500k/2600k cpu's overclock them.

 

[ianj14]

A single, static SETI block for running in BOINC. Static in that it would be one block that doesn't change ever, and used for all tests (current and future) so that a fair comparison of basic processing power over time could be given.

 

[Guest]

Shogun 2 Total War benchmark please sir

 

[yyk71200]

Pov-Ray. Version 3.7 supports SMP. Currently at RC-3 stage. Anandtech uses it.

 

[Tijok]

I would like to cast yet another vote for Matlab and Solidwork!

Athough Autodesk is already represented, and that may cover the Solidworks request in part, this engineering specific software could benefit a number of professionals and engineers of various types.

Also, Linux and open source alternatives would be great for all of these, including existing programs, so I'd also cast a vote for those.

But a big vote for Matlab here! I've seen this program eat up hours, and would love to know the best way to save time with it!

 

[lobo521]

I stream while plaing starcraft 2. I can be good multitasking benchmark.

 

[bombat1994]

wouldnt mind seeing left 4 dead. that game is amazingly cpu depended

 

[arbreen]

MATLAB would be a good benchmark for CPU's and for GPU's, though you'd need a better set of programs than the built-in on for benchmarking.

ANSYS would be great to see, especially because you can tell it how many cores you want it to use and because it spits out performance in MFLOPS at the end.

 

[Axonn]

Visual Studio 2010 with a large project!

 

[crisan_tiberiu]

1. Starcraft 2
2. Diablo 3 (when its launched)
3. World of Warcraft (64 bit) (even its not as demanding for the CPU its still one of the most played games)
4. Video encoding (w/e program, but dont favorize intel -^)
5. Archiving
6. Office performance.

 

[goatsetung]

[citation][nom]Cuecuemore[/nom]Yep, this is a great test for single-core strength along with the original Crysis.[/citation]
I thought this changed in some patch and sc2 uses all cores now?

 

[razor512]

Instead of focusing on threading, focus more on applications that don't show a CPU bias.

The current list is good but I recommend adding maya unlimited and a render using the mental ray engine (it doesn't seem to have a CPU bias and it is highly threaded and scales extremely well. It can be used to measure both single threaded and multithreaded.



Also bring back the charts from a few years ago where you could keep the same horizontal bar graph but filter everything but checked cards allowing you to create a custom visual graph comparing just the CPU's we choose.

Now I know you wanted free but while maya is expensive, you can get it for free if any of your staff has a college email and can get the student version.

A working game benchmark is difficult to find as it is very common for games to be bias towards certain brands of CPU, because of this, they are not recommended and having any kind of bias benchmark will completely invalidate having any kind of overall score.

 

[Alsone]

Battlefield 3 - replace one of the older game tests with this as the Bf3 game is going to really push the boundaries by the looks of it:http://www.youtube.com/watch?v=2zw8SmsovJc

Sony Vegas Pro - popular video editor that's more affordable so more popular with home users than Premiere

Kaspersky AV - Scan

MalwareBytes - Scan

 

[Guest]

Another vote for:
Civ V or Shogun 2 - time it takes to end turn (make a save later in the game with lots of units)
Excel - big spreadsheet with complex macro, I have one which takes 40s on i5 2500K@4GHz but cant share it 🙁
some CPU bound games instead of GPU bound ones

 

[Guest]

The list of current benchmarks is a good gauge to test the performance of CPU's in different aspects, and I think that it is more than enough in that area. Practically it doesn't matter what type of games are included as long as there is several games in the benchmark that gauges the performance of CPU's in different environments - my point is that nobody buys multiple CPU's to play different games at the highest performance level.

I would suggest more practical comparisons - boot-up times, time to open common apps like MS Office, performance when playing HD and SD movies, and overall power consumption of the system. And if possible to have these practical benchmarks of older systems up to 2-years back for comparison. This is important because those are the systems that are primarily due for upgrade.

I've been a regular reader of Tom's hardware way back when it was more known among the techies as "sysdoc.pair.com", and I've made decisions in the past (more recently a few months ago) to upgrade a PC component due to compelling benchmark results, only to find out that it has no impact on the performance of my PC in terms of user experience. As an example, I recently upgraded from Phenom II X3 720 to Phenom II X4 970 - on reviews, you can see significant difference in performance, on actual usage I can see no difference on regular apps - browsing, playing movies, MS Office, and others. This is the same thing if you increase the memory of your PC - from 4GB to 8GB, you cannot notice performance difference on regular usage.

These practical benchmarks can also be applied on other component upgrades such as video cards and hard disks. In my personal experience of being a PC enthusiast for more than 20 years, the upgrades that are really noticeable are the the hard disk upgrades and the cpu cooler upgrades. An upgrade from 5400 rpm HDD to 7200 rpm provides instant performance boost that is noticeable, and an upgrade from HDD to SSD provides the best performance boost in almost all areas of regular computing. An upgrade to a good after-market CPU cooler does not give you the performance boost without overclocking, but significantly lower cpu operating temperature that definitely improves reliability of the system.