Navigate:

Advice
Beginners
BIOS Guide
CPUs
Links
Mainboards
Memory
Network
Storage
Video/Sound Cards

Contact
Forum
SiteMap
Sponsors
WebNews
Home
. .

Prices:

Mainboards

ABIT
ASUS
Chaintech
Shuttle
Soyo
Tyan

CPU
Intel
P4 2.4C-800
P4 2.6C-800
P4 2.8C-800
P4 3.0-800
P4 3.2-800

AMD
AthlonXP
XP 1700+
XP 2000+
XP 2400+
XP 2500+
XP 2700+
XP 3000+
XP 3200+

Athlon64
Athlon64 3200+
Athlon64 FX-51

Opteron
Opteron 240
Opteron 242
Opteron 244
Opteron 246

Memory

Corsair
Crucial
Kingston
Mushkin
OCZ

Search Prices:


























































































































LOSTCIRCUITS

SHORTCUTS:
New Performance Panoramas
A Memory Address Space Crash Course
System Configuration
Disk Caching?
System Memory Density Impact
WoW, WinXP and WinXP-64
Conclusion
Discuss this article here:

 Panorama Factory and the $64-bit Question
What's in a Benchmark...?
(Review by MS, September 2, 2004)
OCZ PC3200 DUAL-CHANNEL EL DDR 512MB(256X2)
400MHz DDR CAS2 - PLATINUM

The Benchmarks

We were looking at a total of four different (albeit related) benchmarks that were used by Panorama Factory. The first two benchmarks were a rather Gargantuan stitching of 22 pictures to generate a 360 degree panorama. Each picture used was approximately 190 MB in file size, meaning that the entire amount of data to be processed was in the order of ~4.2GB. However, since there was some overlap between the different images, the total size of the “blended” file was about 3.1 GB – coincidentally just a hare’s ass above the limitations of Windows2000 Server.


     

Left: Partial view of the stitched and rendered scene (bottom) with two tumbnails of the entire 360 degree surround view on top. Note the picture dimensions of 41262 x 4191 pixels which, in .png format occupy 3.11 GB of memory space (shown as page file size in the right picture).

The individual benchmarks are scripted to run without user intervention and the first two options will attempt to run the 22 picture behemoth crafted to show the insufficiencies of the 32-bit memory addressing with the result that the 32-bit application running either in a WoW (32-bit Windows on 64-bit Windows) or in native WindowsXP will fail and exit with the predictable “out of memory” message. According to schedule the out of memory error occurs at the point where the pagefile usage jumps from 1.6 GB above the 2GB limit. Using the 64-bit application, the benchmark finishes on schedule, however, it is necessary to note that the system will temporarily become completely unresponsive to keyboard and mouse, and even the“Performance” window in Windows Task Manager will freeze completely. The total recall of all system resources to manage the more than 3 GB of data being processed is, however, only temporary and the system will recover from catatonia after a few minutes.

Athlon 64 3000+
(Newcastle core) At:

           

Left: stitch obtained from four individual images. Center: we monitored up to 1.6GB of virtual memory usage in the 32-bit application which was no problem at all, however, as soon as the point was reached where the 64-bit version of Panorama Factory showed a jump to 2 GB of virtual memory usage, the system faulted and generated an "Out Of Memory" error message as predicted (right picture). Note that even after Panorama Factory was terminated by a fault handler execution, the page file usage remained at about 1.5 GB.

Tests 3 and 4 execute a reduced workload consisting of four pictures only that are stitched together. It is still a formidable amount of memory that is commanded in this case, moreover, some of the data appear to stay resident within the virtual memory on the hard disk since consecutive runs of the same task will see some substantial increase in performance, at least as long as only 1 MB of memory is used. Increasing the total amount of system memory decreases the variation between runs to no more than about 5%, regardless of whether the system is rebooted in between or not. Since we are interested in the difference between 64-bit computing mode rather than in the impact of different caching strategies, we ran the main bulk of benchmarks with 2 GB of memory. However, we also show some data reflecting the increase in performance stemming from higher system memory density as well as some numbers reflecting the variability between results with 1 GB of memory only.

next page:      => Test Configuration =>

All advice and educational articles on LostCircuits are free, but if you feel you can, please make a small donation to us!
Thank you!

General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site. All names and trademarks mentioned in this review are the exclusive property of the respective parent companies.
All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.
Copyright 2002 - 2008 LostCircuits