[cfe-dev] Hardware for building LLVM+Clang in 10-20 minutes

[Kaelyn Uhrain]

On Fri, Mar 29, 2013 at 8:43 AM, Anton Yartsev <anton.yartsev at gmail.com>wrote:

> On 29.03.2013 19:16, John Criswell wrote:
>
>> On 3/29/13 9:48 AM, Anton Yartsev wrote:
>>
>>> Hi all!
>>>
>>> I have finally decided to upgrade from my old Core 2 Duo E8500, 6GB DDR2
>>> 400 MHz.
>>> Currently the full building+testing of llvm lasts for hours (slightly
>>> faster with VS2008, slightly slower with MinGW) and makes everything lag.
>>>
>>
>> By "test," do you mean the lit tests in llvm/test or a full run of the
>> test suite in the test-suite project?
>>
> I mean lit tests.
>
>
>
>>  Ideally would be happy to perform build+test for about 10-20 minutes, as
>>> some fast builders do, and have no lags with other useful applications,
>>> such as browsers and Acrobat, during building/testing.
>>> Please can anybody advise me an appropriate hardware? Or just tell, what
>>> hardware the fast buildbots are running (specifically
>>> clang-native-mingw64-win7, clang-x86_64-ubuntu-gdb-75 and
>>> clang-x86_64-debian-fast)?
>>>
>>>
>> Just to throw out some rough numbers, we have a 128 GB, 32 core machine.
>>  The processors are Intel(R) Xeon(R) CPU E7- 8837  @ 2.67GHz.
>> Doing a make -j20 on LLVM+Clang from scratch takes about 4 minutes of
>> wall time and about 50 minutes of user+system CPU time.  I can't seem to
>> run the tests in llvm/test, so I can't time that.
>>
> Cool! :)
> Then I can forget about 10 minutes..
> looking at something like Intel Core i7-3770K or Core i7-3930K in the best
> case.
>

Out of curiosity I decided to see how long my home system takes to build,
and I think 10 min shouldn't be too hard to reach with consumer hardware.
On my Core i7-2600K (Sandy Bridge) @ 4.8GHz with 16GB of DDR3-2133 RAM and
Linux running on a 2TB 7200 RPM SATA-III hard drive with 64MB cache, in a
new build directory with the current SVN llvm+clang+compiler-rt configured
by running "cmake -DCMAKE_BUILD_TYPE=Debug -Wno-dev .." and using gcc 4.7.2
as the compiler:

$ time { make -j9 && make -j9 check-all ; }
(lots of build output followed by a successful run of the tests)

real    10m43.235s
user    69m20.537s
sys     4m7.796s

FYI: Doing all of the above but running "check" instead of "check-all"
reduced wall-time by about 3 1/2 minutes:

$ time { make -j9 && make -j9 check ; }
real 7m15.333s
user 46m33.499s
sys 2m32.316s

Cheers,
Kaelyn


> Thanx for your replay!
>
>
>> -- John T.
>>
>>
> --
> Anton
>
>
> ______________________________**_________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/**mailman/listinfo/cfe-dev<http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20130329/94ec361e/attachment.html>