| 1 |
Zhang Weiwu wrote: |
| 2 |
> Andrew Gaffney wrote: |
| 3 |
> |
| 4 |
>> Zhang Weiwu wrote: |
| 5 |
>> |
| 6 |
>>> Hello. After I get an old U5 box, I decide to replace my desktop |
| 7 |
>>> computer with another UltraSparc. My desktop computer is Pentium 1G / |
| 8 |
>>> 256MB memory / ATA100 IDE box. |
| 9 |
>>> |
| 10 |
>>> What UltraSparc (with how many of what CPU) has similiar average |
| 11 |
>>> desktop performance as my Pentium box? I usually only browsing |
| 12 |
>>> around, use openoffice and gimp a little, vim as editor and doesn't |
| 13 |
>>> use huge IDE like eclipse. What's your suggestion? |
| 14 |
>> |
| 15 |
>> This is just a guess, but an Ultra 2 with dual 300s or 400s is |
| 16 |
>> probably going to be pretty close in performance. |
| 17 |
>> |
| 18 |
> A small test showed very disapointing result. (This is not a desktop |
| 19 |
> performace test, only I play with it) |
| 20 |
> |
| 21 |
> I tried to compare my Ultra 5 ( 333MHz single CPU / ATA66 IDE / 256 MB ) |
| 22 |
> with Celeron 600 MHz box ( ATA 66, 256MB memory). My test is to |
| 23 |
> calculate the complecity of Linux kernel code. |
| 24 |
> (command line: # sloccount /usr/src/linux-sparc-2.4.27/ ) |
| 25 |
> |
| 26 |
> On U5 it takes325 seconds, on Celeron it takes 257 seconds. They both |
| 27 |
> take almost same amount of kernel time, but U5 takes 210 seconds user |
| 28 |
> time, while Celeron only takes 76 seconds. |
| 29 |
> |
| 30 |
> To perform this task, Celeron used 79% of total time compare to U5. So I |
| 31 |
> think perhaps U5 does not perform better than Celeron 500MHz. This way, |
| 32 |
> I can hardly expect dual 300MHz would overperform Pentium 1G, especially |
| 33 |
> dual CPU (IMHO) does not help much with desktop thanks to that only a |
| 34 |
> few desktop applications use more than one process... |
| 35 |
> |
| 36 |
> Perhaps I am wrong in this small test? |
| 37 |
|
| 38 |
Yes, a U2 is *not* a U5. The U5 and U10 were probably the crappiest Ultra |
| 39 |
machines that Sun put out. Also, the test was a little one-sided since the |
| 40 |
Celeron had almost twice the speed and an IDE chipset with working DMA :) A |
| 41 |
2x300 U2 would "mop the floor" with the Celeron in that test. |
| 42 |
|
| 43 |
-- |
| 44 |
Andrew Gaffney |
| 45 |
Gentoo Linux Developer |
| 46 |
Installer Project |
| 47 |
|
| 48 |
-- |
| 49 |
gentoo-sparc@g.o mailing list |
Archives