![]() |
![]() |
#1 |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
10001110001112 Posts |
![]()
I currently have 2 Pentium 4 PC's factoring in the 43M range.
An IBM 2.8Ghz with 512 Mb RAM takes 605 seconds to do 1% of 68 bits when Task Manager shows Prime95 using 99%. An AVRO 2.4Ghz with 512 Mb RAM takes 570 seconds to do 1% of 68 bits when Task Manager shows Prime95 using 99%. Purely based on processor speed and RAM I would have expected the relative speeds to be reversed. I suppose the obvious answer is that there is a lot more to raw number crunching than Mhz and RAM? |
![]() |
![]() |
![]() |
#2 |
Account Deleted
"Tim Sorbera"
Aug 2006
San Antonio, TX USA
17×251 Posts |
![]() |
![]() |
![]() |
![]() |
#3 |
"Jason Goatcher"
Mar 2005
5×701 Posts |
![]()
Silly question time: Clock for clock, which does more? An Athlon processor or a Core 2 Duo?
|
![]() |
![]() |
![]() |
#4 |
Apr 2006
Down Under
89 Posts |
![]()
You're right it is a silly question as the answer is "it depends".
However for 'most' performance tests a Core 2 duo will outperform an Athlon X2 at the same clock speed. As is the way in technology once Phenom comes out the tables may well be turned again. Check out Tom's for a detailed comparison. |
![]() |
![]() |
![]() |
#5 |
Feb 2006
AR, US
24×32 Posts |
![]()
I would say the Core 2 Duo, and I can see this by comparing my X2 4400+ with an E6400. Iteration times for the E6400 are over 50% less than 4400+ (with similar clock speeds).
Also, the Core 2 Duo executes 4 instructions per clock cycle as opposed to the Pentium cpus based on the NetBurst architecture which execute 2 instructions per clock cycle. |
![]() |
![]() |
![]() |
#6 | |
Aug 2002
North San Diego County
68210 Posts |
![]() Quote:
Might try running both instances in safe mode to see if there are odd processes stealing cycles. Last fiddled with by sdbardwick on 2007-07-18 at 21:53 |
|
![]() |
![]() |
![]() |
#7 |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
3·37·41 Posts |
![]()
They are both Intel (not Athlon). I don't know how to tell if they are Northwood or Prescott.
A couple other things: 1. The 2.4Mhz is running XP Home the 2.8Mhz is running XP Professional 2. When I bring up task manager the 2.4 never shows more than 49% going to Prime95 (I asked last year and was told this was normal --- that I still was getting the entire CPU ... but if I run another CPU intensive program I still show 49% Prime95 and up to 49% for the other). The 2.8 does show 99% going to Prime95. |
![]() |
![]() |
![]() |
#8 |
(loop (#_fork))
Feb 2006
Cambridge, England
2·3,191 Posts |
![]()
OK, the 2.4GHz machine is hyperthreaded (it pretends to have two processors, one running in the idle cycles of the other, and prime95 uses all of one of them, efficiently enough that there are few idle cycles), and the 2.8GHz machine isn't.
This means that the 2.4GHz machine is a late-model Northwood on an 800MHz front-side bus (because only those had hyperthreading), and the 2.8GHz machine is a Prescott-based Celeron on a 533MHz bus; so it accesses memory more slowly. |
![]() |
![]() |
![]() |
#9 | |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
455110 Posts |
![]() Quote:
Knowing this, should I be doing anything different or changing any settings? And now that you know so many of my deepest secrets I'm going to end up losing my CIA Level 7 certification . ![]() |
|
![]() |
![]() |
![]() |
#10 | |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
455110 Posts |
![]() Quote:
![]() The IBM 2.8Ghz is processing an exponent in the 39M range and taking .077 seconds per iteration. The AVRO 2.4Ghz is processing an exponent in the 35M range and taking .081 seconds per iteration. This is closer to what I would have expected based purely on Processor speed and RAM Comments?? Fivemack?? |
|
![]() |
![]() |
![]() |
Thread Tools | |
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
Budget PC Throughput | Rodrigo | Hardware | 14 | 2011-09-26 10:16 |
PrimeNet throughput over the years | Brain | PrimeNet | 5 | 2010-12-08 00:53 |
how is the throughput calculated? | ixfd64 | PrimeNet | 5 | 2008-05-21 13:39 |
increasing P-1 throughput with hyperthreading? | hhh | Software | 17 | 2005-06-09 02:56 |
Fake throughput drop | Lumly | Lounge | 12 | 2002-09-05 20:00 |