[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

lynx-dev Compiling lynx.exe for the VIA C3 CPU

From: vtailor
Subject: lynx-dev Compiling lynx.exe for the VIA C3 CPU
Date: Wed, 24 Apr 2002 14:43:31 -0400

Anyone out there who owns a VIA/C3 system and uses makelynx.bat to compile
lynx might want to use different optimization flags than the ones 
specified in makelynx.bat:

For the VIA/C3 (mine is an 866 Mhz.), the flags are

                -O -m486

The reason for these flags is that I tested my system using the nbench
Linux benchmark compiled for Windows using MingW32 gcc, and the fastest
version of nbench.exe was the one with the above flags.

I attribute the favoring of -O optimization to the fact that some engineer
at VIA decided to make the L1 cache 128 KB and the L2 cache 64 KB, against
any "popular" wisdom to the contrary.  Since the L2 cache gets kinda
crowded in operation, using optimization that saves space in the program
squeezes more program into the L2 cache at one time.

Oh yes, the prime benefits of the C3 are that comparable C3s run at about
80% of the speed of the newest Celerons for typical Windows programming
tasks (see web site version) and only consume about 3
watts or so on their tiny 54 square mm. dies.  (Compiling lynx is
noticeably faster with the C3/866 than with my old K62/500.) This
discrepancy in speed is attributed by VIA to advanced microprogramming
tricks encoded in Celeron silicon, and missing in the more frugal C3,
but I think the real problem is the L1 to L2 cache size ratio.

Oh, if this sounds like a commercial announcement for the VIA C3, it is.
I have observed that, when I advertise something, it generally goes out
of business.

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]