Is FreeBSD amd64 the right choice for my netbook?

I have installed FreeBSD before on my netbook. I did not read the documentation available and simply installed the i386 version on account of the fact that this is what I always use.

The system seemed to work but I could not configure Xwindows. The screen would just blank and the netbook would freeze. I think now that this may be down to the graphics driver, but I am not sure. Due to these problems I had to revert to Linux.

I am now wishing to try again at installing FreeBSD on my netbook. These are the specifications:
--------------
Model: HP Mini 110-3500
Processor: Intel Atom CPU N455 @ 1.66GHZ
Processor Speed: 1666 MHz
Memory size: 1024 MB
VGA compatible controller: Intel Corporation N10 Family Integrated Graphics Controller
00:02.1 Display controller: Intel Corporation N10 Family Integrated Graphics Controller
Network controller: Broadcom Corporation BCM4313 802.11b/g/n Wireless LAN Controller (rev 01)
03:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8101E/RTL8102E PCI Express Fast Ethernet controller (rev 05)
---------------

My questions are:
1. Should I be installing the amd64 version?
2. Are there any 'known issues' in FreeBSD running X windows with my hardware?

I am aware that there is not currently a driver for my wireless controller, but that is not a primary concern for me.

Many thanks
 
neilms said:
1. Should I be installing the amd64 version?
Not really, you only have 1GB of memory. The amd64 version is really only interesting if you have 4GB or more.

2. Are there any 'known issues' in FreeBSD running X windows with my hardware?
The graphicscard may be an issue. Be sure to use the latest FreeBSD and compile Xorg with KMS.
 
wblock@ said:
64-bit might run slightly faster.

Really? My experience has been that 9-stable i386 is generally faster than amd64 on the same hardware. I realize that this is probably application sensitive, etc.
 
Multiple people have told me that amd64 should be faster, but I have not benchmarked it or seen anyone else do it. If you have numbers for both on the same hardware, please post them.

My general benchmark, buildworld, doesn't work because amd64 has a lot more to build than i386.
 
I don't have good numbers at my fingertips, and might have been confused by amd64 buildworld making more than i386. If I have time to do a decent benchmark on it you'll hear about it.
 
By definition AMD64 is not faster than i386. Both types of instructions are executed directly in the processor's core, there's no 'translation' required. That means an ADD instruction for example takes the same amount of clock cycles in 64 bit or 32 bit mode. Hence they run at the same speed.
 
SirDice said:
...takes the same amount of clock cycles in 64 bit or 32 bit mode. Hence they run at the same speed.

Of course, this ignores memory bandwidth and cache effects, which may vary due to the changes in operand and instruction size. These effects will naturally be sensitive to the details of cache implementation and program behavior.
 
Uniballer said:
Of course, this ignores memory bandwidth and cache effects, which may vary due to the changes in operand and instruction size. These effects will naturally be sensitive to the details of cache implementation and program behavior.

True but the Pentium (and higher) processor has a 64 bit external databus. So it takes just as much effort to latch in a 32 bit instruction as it does a 64 bit one. If anything the 32 bit may even be faster as it's able to latch in 2 instructions (2 32 bit words) at the same time.

On the other hand AMD64 does have more registers so in theory that could speed up things (less access to slow 'external' memory). But the compiler does need to use them and I'm not entirely sure that's the case. I don't know enough about compilers to make that judgment.

Floating point math definitely should be faster on AMD64 as it uses SSE for that compared to the 'old' way of handing it off to the slow FPU.

In short, you can't simply state AMD64 is faster than i386. In some specific cases, yes, it may be faster. For most real world applications, the difference will be negligible.
 
Back
Top