Prev: Typography: Choosing Century Expanded?
Next: Considering the forces of computation: questions for the group, I have three:
From: Tim Little on 2 Dec 2009 23:35
On 2009-12-02, factorboy13 <temp15(a)atelierweb.com> wrote:
> Well, I run factorization tests both on it and on an old Core2 2600
> (2 cores/2threads) and the I7 (4cores/8threads) is more than 3x
> faster. Also ran factorizations with a dual CPU Xeon 3.0 that does
> hiperthreading (then 4 threads), but this one is much slower than
> the Core2, probably the sole reason is that it running 32-bit
> binaries (cannot support 64-bit OS). I can only be happy with the i7

I'm not trying to imply that the i7 is poor, just that use of its
hyperthreading capability has mixed performance impact depending upon
type of workload and fine details of algorithm implementation. The
appropriate comparison is to see how the same i7 performs with
hyperthreading turned on, vs turned off.


- Tim
From: factorboy13 on 4 Dec 2009 07:00
I understand what you mean, there are many cases, particularly games, where it appears that HT actually reduces performance.
I have not made any test with HT disabled.
With the HT enabled what I see is that CPU utilization is proporcional to the number of threads in use. For example if I run a single-threaded console application CPU utilization is about 12.5% and for every instance I add CPU utilization increases by another 12.5%. When 8 applications like that are running at full spead CPU utilization is near 100%.
I really don't know if each application was going to run at more than double speed if I was going to run with HT disabled. Probably i will have to make a test sometime.
First  |  Prev  | 
Pages: 1 2 3 4 5 6 7 8
Prev: Typography: Choosing Century Expanded?
Next: Considering the forces of computation: questions for the group, I have three: