Recommended CPU for timetabling

[Volker Dirr]

thank you. i added that result.

so windows xp is faster windows 7?
(because:
3min 52s - Intel Core2 Duo E6550 (2.33 GHz - precompiled FET with Windows XP - metered by alfaromeo)
3min 55s - Intel Core 2 Duo E6600 (2.40 GHz - precompiled FET 5.14.3 with Windows 7 32bit - metered by y)
)
or it was just a bit luck, because results are always +/- a few seconds.

[y]

Maybe it has not anything to do with windows version. According to these specs sheets,
E6550: http://ark.intel.com/products/30783/Intel-Core2-Duo-Processor-E6550-4M-Cache-2_33-GHz-1333-MHz-FSB
and
E6600: http://ark.intel.com/products/27250
both have the same cache, E6600 is a bit older chip and it has 3% faster frequency, but E6550 has quite higher (25%) FSB speed. Maybe the latter gives that marginal better result to E6550.

I measured 3 times the E6600, giving 3:56, 3:54 and 3:55! I reported the middle value

[Christian Kemmer]

1min34sec precompiled FET 5.14.3 on Windows 7, 64-Bit, Intel i7-2700K@3.50GHz

[Volker Dirr]

Thank you for the report. I added your result in the first post.

[lalloso]

2 min 04 secs on an Intel Core i5-2520M CPU@2.50 Ghz 2.50Ghz

[Volker Dirr]

Thank you for the report. I added your result in the first post.

[Volker Dirr]

i also added a new result in the first post:
2 min 07 secs on an Intel Core i5-4200U 1.60GHz (of course clock speed is much higher during generating with FET)

[Volker Dirr]

Maybe someone wonder that an Core i5-4200U 1.60GHz is nearly as fast as an AMD x2 270 3.4 GHz.
Of course the i5 is much newer but on the other hand it is also not an desktop CPU.
But there is still a big difference between that CPUs.
If you generate 2 timetables simoultanously then both cores of the AMD CPU still stay at 3.4 GHz, so both timetables are solved in around 2 minutes.
But the Intel CPU can't hold the turbo clock (2.5 GHz) and slow down to 1.6 GHz. So generating 2 timetables simoultanously is much slower with that Intel CPU.

So what do you think? Do we need to modify the benchmark rules? As you can see: The current rules are only correct if you generate a single timetable. So it isn't very realistic. On the other side it will be very hard to do a more "realistic" benchmark.

I also like to know some other results. Anybody out there who can benchmark a new Intel i5-4xxx deskop CPU and/or an AMD A10-7xxx CPU?

[Volker Dirr]

results from Raspberry Pi (ARM CPU) added.
1h 11min 25s - Broadcom BCM2835 - ARM1176JZF-S (0.7 GHz - Raspberry Pi  - metered by Volker Dirr)

[Volker Dirr]

results from Raspberry Pi 2 (ARM CPU) added.
17min 15s - Broadcom BCM2836 - ARM Cortex-A7 (0.9 GHz - Raspberry Pi  - metered by Volker Dirr)

[Volker Dirr]

results from Commodore Amiga 500 added.
3d 14h 1min 59s - Motorola 68000 (7 MHz - AROS - metered by Volker Dirr)

[Liviu Lalescu]

results from Commodore Amiga 500 added.
3d 14h 1min 59s - Motorola 68000 (7 MHz - AROS - metered by Volker Dirr)

Wow!

You forgot to add this in the first post.

Is this comparable to a Sinclair Spectrum Z80 with 64 k memory which I was using in the 90's? You added more memory to it? Also, it was an 8-bit processor? How is this magic possible?

[Volker Dirr]

Amiga has got a 16 bit CPU. I think it is impossible with an 8 bit CPU to run FET.
The default memory of an Amiga is also 512 kb (not 64 kb) and it can be upgraded with 2 slots. So you can add 2 MB. Of course it isn't enough memory to run the GUI, i benchmarked by using fet-cl (command line version).

[y]

1min 57s - Intel Celeron G1840 (2.80 GHz - precompiled FET 5.14.3 with Windows 7 64bit - metered by y)

[Volker Dirr]

Thank you for report. I added your result in the first post.