Need recommendations for new Super System

aleric

Distinguished
Jun 21, 2004
12
0
18,510
Good day all.

I post this in the motherboard forum, but really I am in the process
of deciding which components to use for my new system.

Normally, I only buy a new PC when the new one is at LEAST four
times as fast at everything as the old one. However, I always AIM
for 10 times as fast ;). My current system is a pentium-4 1.7 Ghz.

When I say faster, I mean at compiling C++ code on linux.
Unfortunately, there does not seem to exist benchmarks that I can use
to see how fast a new PC is in this regard :(

Normally I am biased towards AMD cpu's because they excell at
compile speed. Take for example the following benchmark into account
which I performed myself:

--------------------------------------------------------------------------------
Benchmark:
- Compile time (time make) of libcwd-0.99.45 after a ./configure --enable-maintainer-mode -disable-pch
compiler: gcc version 4.1.2 20061115 (prerelease) (Debian 4.1.1-21)
(OS: debian 'testing' (Lenny) at Apr 26, 2007).

System 1:
- model name : AMD Athlon(tm) Processor
cpu MHz : 908.119
cache size : 256 KB
bogomips : 1818.08
MemTotal: 906592 kB
Diskspeed:
Timing cached reads: 272 MB in 2.00 seconds = 135.85 MB/sec
Timing buffered disk reads: 148 MB in 3.02 seconds = 49.04 MB/sec

System 2:
- model name : Intel(R) Pentium(R) 4 CPU 1.70GHz
cpu MHz : 1708.705
cache size : 256 KB
bogomips : 3420.70
MemTotal: 1036664 kB
Diskspeed:
Timing cached reads: 628 MB in 2.00 seconds = 313.39 MB/sec
Timing buffered disk reads: 182 MB in 3.00 seconds = 60.61 MB/sec

'vmstat 1' shows that during compilation 100% cpu is being used,
and both, id(le) and wa(it for IO), are constantly 0. Hence, we are
not measuring diskspeed here - but cpu speed.

Results:

The Althon 900 compiles libcwd in 2 minutes and 5 seconds.
The Pentium-4 1.7 GHz does the same job in 1 minute 53 seconds.

Conclusion: the pentium is only 1.12 times faster, despite that it's nearly
double clock frequency.
--------------------------------------------------------------------------------

HOWEVER --

In order to get something 10 times as fast, I will need four cores
at high frequencies. AMD's only option would be their dual socket F
motherboard with two Athlon 64 FX-72 cpu's for example. My problem
with that is that it eats too much power (over 400 Watts idle).

Therefore, purely because I am not really interested in getting more
speed at the cost of a lot of electricity (then I might as well build me
a grid), my decision is that only 65nm or better based chips will do;
which at present time brings us to intel.

So - the cpu I was thinking of now is the Intel core 2 Extreme QX6700
(the price/performance of the QX6800 is a bit too much money for what
you get extra imo).

I am normally not into overclocking (I think about speed gains in steps
of a factor 2 -- I don't think that watercooling, and months of tweaking
something that I don't really have any experience with, risking to blow up my PC, is worth the 20% gain I might get).

I am also normally not into gaming, that is: I want my PC to compile fast.
Whatever the resulting speed with gaming is will be fine with me.

HOWEVER -- I intended to start to use triple head (currently using
dual-head) and use three times a 1680x1050 20" TFT for that. And
having that I certainly insist that I can play 3D games that use all three
monitors. This means that I will need two identical video cards (I think)
that together can drive a 'screen' as big as 5040x1050. I have read
that the nvidia's (It have to be nvidia chipsets) 7xxx can only do 4096x4096 windows with 3D. One source told me that the 8xxx series
should not have this limitation, but that he had different experiences :(

I am not sure what kind of demands this puts on the motherboard.
I can't afford ... um no, I am not interested in paying for two 8800's
that together cost more than a QX6700, as I am not really a gamer.
I only want that it can DO 5040x1050, not that it is very fast or something.

Given the above - what would be a good motherboard for me?
I don't really care about the price; if what I get back is a mobo that
will give me all the compile performance that I can get with an QF6700.

Thanks for the help,
Aleric
 

SuperFly03

Distinguished
Dec 2, 2004
2,514
0
20,790
Good grief. This is outside my league for the most part. TheGreatGrapeApe and Cleeve are the resident video experts, along with some others who are far more knowledgeable than I. I am not qualified to speak to whether or not video cards can drive 3 monitors at that resolution.

As for a CPU, the QX6700 will be a monster and it is easy enough to overclock (and I mean real easy, 2 or 3 settings could get you some decent clocks) or leave it at stock and still get great performance compared to your previous rig. You will also want to buy an aftermarket cooler for your QX6700. While the stock one foes the job, it gets pretty hot during load and excellent air coolers can be had for about $60 and aren't difficult to install.

Moving to motherboards, in general terms of performance the 680i/965/RD600/975x are all relatively close in performance, what usually tends a person to pick one over the other is the graphics card, or their desire for overclocking. In the case of graphics cards it only matters if you end up running two identical cards in parallel, but if you run the separately then it doesn't matter.

Frankly the clock speed war is dead nowadays, it has become more about performance/watt and performance/clock rather than ratcheting the clock speed as high as it will go.

I wish I had a specific recommendation for you, but if you don't overclock the difference between the chipsets is 2-3% usually (not in every case). I think you should tend towards the nVidia 680i chipset if you intend to run dual graphics cards because if you start gaming at that ridiculous resolution (compared to normal :wink:) you will need the combined power of those two graphics cards in parallel (SLi feature).

Like I said the setup you have in mind is a bit out of my league, so I could be in error about the graphics card stuff.
 

dsidious

Distinguished
Dec 9, 2006
285
0
18,780
I think you should tend towards the nVidia 680i chipset if you intend to run dual graphics cards because if you start gaming at that ridiculous resolution (compared to normal :wink:) you will need the combined power of those two graphics cards in parallel (SLi feature).

The 680i chipset has some special features that improve video performance a bit when used with recent nVidia cards such as the 8800 GTX. I forgot the details, sorry, I read about that 2 months ago. A motherboard like Asus Striker Extreme or P5N32-E SLI are probably your best bet.

Don't know if your compiler can actually take advantage of multiple cores. Of course, you can always try splitting your makefile in pieces and writing a shell script to run the pieces in parallel. That would be a fun project :p