Linux Benchmarking... Even Faster & A Very Interesting February

[phoronix]

Phoronix: Linux Benchmarking... Even Faster & A Very Interesting February

While benchmarked the most this month on Phoronix was the new ThinkPad X1 Carbon with Broadwell CPU given its the latest-generation Intel microarchitecture, February on Phoronix will be much more interesting if you're at all interested in servers or workstation hardware... Or just seeing what's possible if you happen to have a ton of system memory and disks...

http://www.phoronix.com/scan.php?page=news_item&px=Many-DIMMs-Xeon-Disks

[mufasa72]

Software vs hardware RAID across the different file systems would peak my interest.

[spstarr]

Michael something is broken with benchmarking..redo of benchmarks against X1 Carbon..

Here is my latest Comparison with the Lenovo X1 Carbon vs my Dell Precision M6800

http://openbenchmarking.org/result/1501315-DE-1501254DE86

Some notes:

- kernel 3.19-rc6
- Removed ccache as this is not detected and it will create a bogus result
- DPM is set to balanced / auto for performance
- GPU HyperZ is enabled
- LLVM 3.6 / Mesa is same build for i386/x86_64
- I disable LVDS/eDP and redirect via VGA to the Samsung panel display

* Disk I/O is crap this is a 5400 RPM 1TB hybrid disk, and it's poor, I'll swap it eventually for a real SSD 1TB when they become cheap.

The following tests failed to properly run:

- pts/fio-1.7.1: Type: Random Read - IO Engine: POSIX AIO - Buffered: No - Block Size: 4KB - Disk Target: Default Test Directory - Result: MB/s
- pts/unigine-tropics-1.6.0: Resolution: 1280 x 1024 - Mode: Fullscreen

The latter, seems to be how you execute script, it runs as a window, have to hack the script to get it to display to full screen.

Those numbers should be as following:

http://openbenchmarking.org/result/1501314-DE-REDO2991862:

Unigine Tropics 1.3:
pts/unigine-tropics-1.6.0 [Resolution: 1280 x 1024 - Mode: Fullscreen]
Test 1 of 1
Estimated Trial Run Count: 3
Started Run 1 @ 00:15:23
Started Run 2 @ 00:18:55 [Std. Dev: 1.67%]
Started Run 3 @ 00:22:26 [Std. Dev: 1.28%]

Test Results:
158.351
162.138
158.899

Average: 159.80 Frames Per Second

And at the display my Samsung panel uses:

Unigine Tropics 1.3:
pts/unigine-tropics-1.6.0 [Resolution: 1920 x 1200 - Mode: Fullscreen]
Test 1 of 1
Estimated Trial Run Count: 3
Estimated Time To Completion: 6 Minutes
Started Run 1 @ 00:43:14
Started Run 2 @ 00:46:44
Started Run 3 @ 00:50:15 [Std. Dev: 0.05%]

Test Results:
63.8257
63.8656
63.8049

Average: 63.83 Frames Per Second

[LinuxID10T]

Software vs hardware RAID across the different file systems would peak my interest.

Huh, that is funny, but it would PIQUE my interest too...

[darkbasic]

Huh, that is funny, but it would PIQUE my interest too...

Me too.

[Jeroen]

It would be very interesting to see some virtualization benchmarks on this kind of hardware (KVM,XEN,VMware,Virtualbox), especially across different raid setups.

[darkbasic]

It would be very interesting to see some virtualization benchmarks on this kind of hardware (KVM,XEN,VMware,Virtualbox), especially across different raid setups.

Agree, Xen PV vs Xen PVH vs KVM benchmarks please.

[ferry]

I'd like to see some disk bench marks based on database transactions (postgres?), knowing that btrfs will probable be a bit slow. I would be nice to track progress in this area over time.

[drSeehas]

I'd like to see ZFS on Solaris and/or FreeBSD.

[A Laggy Grunt]

On the server side, "crazy fast" takes a backseat to "utterly dependable" (somene was talking about a Tyan build with 10+ years of uptime!) and "totally perfect" (we don't want it to get a single-bit error in something which may involve, say, money or someone's life).