• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

techPowerUp 2011 CPU benchmark suite

Great stuff, keep it coming :)
 
Oh, and what about an IntelBurnTest's GFLOP rating? Most modern processors can blow through 5 runs in under 5 minutes.
 
I think it's better to stay impartial and use benchmarks that don't include Intel or AMD in their names :)
 
what about the new lga 1155 for sandy bridge, gonna need a board for those chips
 
I think it's better to stay impartial and use benchmarks that don't include Intel or AMD in their names :)

The test is impartial, it doesn't have bias towards one or the other, but I can see how that might look bad to the average person reading the review.:toast:
 
Wprime will create more heat and power draw than Prime95 ever could...set that baby to 1024 and watch the electricity fly out the window..
 
I'll give it a try, app. that results in more power draw will stay.
 
Since you'll be doing an experiment, why not test OCCT and Linpack as well? I'm curious as to which is the most stressful.
 
I wonder if there's any point in searching for best synthetics stress test that will result in absolute maximum power consumption. That will never happen in normal day to day use, so maybe it's better to measure power draw with some "almost perfect" utility? It would give a more realistic point of view to user on what kind of power supply is needed.
 
I wonder if there's any point in searching for best synthetics stress test that will result in absolute maximum power consumption. That will never happen in normal day to day use, so maybe it's better to measure power draw with some "almost perfect" utility? It would give a more realistic point of view to user on what kind of power supply is needed.
I like to see the absolute worst case scenario. Even if it won't happen in daily usage. If you make accommodation for the worst case scenerio, you'll always be covered. In reality, power consumption will be lower than calculated. I don't mind having over estimated numbers, especially when I'm aware that they are overestimates.

Also, I would like to see benchmarks of the CPU at a 'typical overclock'. Ideally, you could run all the benchmarks at stock and 'typical overclock' settings, but I understand that time is an issue here. So my suggestion is to have more than one or two overclocked benchmarks. A large percentage of the people reading are going to be running the CPU at a 'typical overclock' anyway so stock benchmarks are only so relevant. While on the topic of overclocking, power consumption while overclocked is also a very useful thing to measure.

Also, this is probably unrealistic, but could you develop a way to measure power consumption of the CPU only? Very very few sites do this and this information would be extremely valuable in comparing CPUs. System power consumption is an ok predictor but there are too many variables in order to make an accurate comparison. I know W1zzard has a way to measure GPU power only. The only other site I know of that does this is X-Bit Labs, but unfortunately, they only use it on GPU reviews. I know this is a huge thing to ask and it's probably not going to happen, but at least I mentioned it.
 
measuring cpu only power is almost impossible nowadays and requires extensive motherboard modification
 
Try aoTuV encoder in the SSE3 mode?
 
I guess that most people would not be interested in this, but here goes.
How about compile time for some (big) open source application (like Libre Office aka OpenOffice.org, or VideoLAN or something... maybe even Linux kernel)
 
I don't see that one happening, sorry :)
 
compile time might not be a bad idea, it's probably disk limited though on fast systems. also need a fast multi-threaded compiler.

what is popular open source software that you guys use?
 
That would require to go with SSD. I seriously doubt the consistency of results for that kind of benchmark + classic hard drive.
 
what is popular open source software that you guys use?
Can't talk for everyone, but a list of open source software I use daily:
There's probably a lot more, those are the ones that I really use a lot so I remembered them instantly :laugh:


@Omega & W1zzard:
How would that Seagate hybrid disk (HDD with some flash memory or what was it) do for benchmarking?
 
I see no reason to experiment with hybrid hard drives.
 
No need to experiment, I was more interested in your educated guess, as it might not have the shortcomings of SSDs, while still being faster than classical HDDs... so maybe, just maybe, better suited for benchmark rig.

But, I'm quite clueless as to how it would do after a year or two of usage. I just read some reviews and it seemed interesting (for example: Tom's Hardware review)
 
Benchmark Suite 2010/2011 - ***incomplete***​

  • SYNTHETIC BENCHMARKS:
  • Everest Ultimate - Read/Write/Copy/Latency
  • PC Mark Vantage 64-bit
  • 3D Mark Vantage Performance (CPU and Total score) / 3D Mark 11
  • wPrime v2.4 1024M
  • SuperPI 1M

  • AUDIO ENCODING:
  • unknown
  • unknown

  • VIDEO ENCODING:
  • x264 benchmark HD V3.0
  • unknown

  • RENDERING/IMAGE PROCESSING:
  • Photoshop CS5 Retouch Artists Speed Test
  • Cinebench R11 64-bit
  • POV Ray 3.7b 64-bit

  • ARCHIVING/ENCRYPTION:
  • 7zip 64-bit - 32 MB mutithreading
  • WinRAR Benchmark 64-bit
  • WinRAR/7zip real life compression ~ 1 GB folder
  • Windows 7 performance index encryption/compression

  • GAMING:
  • Civilization V
  • Mafia II
  • Formula 1 2010
  • Batman AA
  • unknown (FPS genre reserved)

Benchmarks currently undergoing testing:
- WCG CPU benchmark test
- TrueCrypt

As seen above, some optimizations are made on the future setlist. Two benchmarks for audio and two for video encoding, rendering/image processing is reduced from four to three benchmarks, as well as the number of games from six to five titles.

WCG and TrueCrypt are under testing and will probably find they're place among other benchmarks. What I need is still two reliable audio encoders with multithreading support and one more popular video encoding utility, and one more gaming title place is open.
 
How about SUPER?

It can encode practically any media file to any other media file, although of course it lacks a "benchmark" function so the only meaningful data (time and CPU usage?) would be gathered using something else.
 
Back
Top