p.17 #3 · Let us testing Photoshop speed in our system
taob wrote:
Yet we also have reports of a dual 3.2GHz Xeon with 4 GB running the operation in 18.7 seconds, and a quad 3.0GHz Xeon with 8 GB clocking in at 9.8 seconds. Something is not being reported correctly here.
/geek mode ON
Prolly incorrect implimentation of hyper threading or poor multithreading implementation.
Adobe is HT and SMP aware. other questions to be raised about the 3 xeon setups are:
1.> Are they hyper-threading xeons?
2.> Are they running on a single shared memory bank or does each processor have a dedicated memory bank.
3.>Are they running on an OS that properly identifies the processors for what they are?
4.> are they 512k,1mb, or 2mb cache xeons?
Windows 2000 server or 2003 server sees those quad xeons as one of the following 4 configurations.
Windows 2000 server:
A> 8 individual processors if they are hyper-threading capable. even though they are four logical and four physical ones. This is a BAD idea. infact they usually perform better if HT is turned OFF at this point as far as processing data goes, especially if SSE2 computations are involved. Adobe uses SSE2 heavily if it's availible.
This is a "non-issue" to microsoft, they have NO plans to make windows 2000 hyperthreading processor aware. In this case you have to have an 8 Processor license vs a 4 processor <2k3 "understands" HT processors as a single processor with an availible logical processor> license with windows 2003 server. Capitalism at it's finest.
B> 4 Physical Processors. Ideal in ANY case HT capable or not.
Windows 2003 Server:
A> Four Processors with HT support
B> Four Processors with out HT support. < if they are not hyper threading processors, or you forgot to enable the HT support in the bios.>
And this is just the quad xeon. Cut my numbers in half in the above situations for the dual xeons. Also what chipset are they running? are they using ddr2 or ddr. 266, 333, 400, or 533 speed memory? I could further ramble on that while adobe is HT and multithreaded aware, using any program that relies on SSE2 for calculations takes a SEVERE hit to the logical of the HT enabled processor, infact completely nullifing it's existance.
/geek mode OFF
As someone said before me, this is not a very scientific setup. There are too many uncontrolled/unreported variables involved. Take the thread for what it's worth, My computer is bigger/badder/faster/better than yours.
That being said,
22.6 seconds
3.8 ghz p4 HT < overclocked from a 3.2>
2 gig of ddr2
ATI x800 XT 256mb PCI-Express
2 36.7gb Raptors raid 0
2 160gb Seagate SATA raid 0
2 250 gb Seagate SATA for backup porposes.
I took the liberty of installing adobe cs on my home server running windows 2k3 SBS as well
29.4 sec
2 2800MP Athlon MP
Nvidia 5200FX 128mb PCI
1 gb ECC pc3200
120 gig seagate IDE drive
11*200gb seagate SATA raid-5 for a total of 2 terrabytes of storage.
200gb seagate sata as a hot spare.
3ware 9500S-12 Raid controller with 128mb of cache.
Done with 2 instances of Folding @ home going cause I'm too lasy to disable then renable the services after the testing.
Ironically setting the adobe and windows scratch drives to the same volume yeilded better results due to the initial slow repsonse of a raid-5 array.
Edited by Rincewind on Jan 18, 2005 at 01:02 PM GMT (Reason: cleaned up some mis spellings.)
p.17 #6 · Let us testing Photoshop speed in our system
here's the table of geeky goodness upto page 14.
It's sorted by speed, then memory then Time completed.
Essentially Machines get BETTER as you go down the list of the processors by manufacturer. I'll reorder the intel section later. and I'll find that amd 64 fx as well.
Assumptions, and other misc you need to know this stuff.
1.>Athlon, Athlon 64, Athlon MP are listed by their model/PR numbers not their true MHZ as I was WAY to lasy to look them up.
2.> All g5's were assumed as SATA hdd unless stated otherwise.Apple standard spec
3.>All g4's were assumed as IDE hdd unless stated otherwise. Apple standard spec
4.>All Laptops were assumed as IDE and Integrated GFX unless stated otherwise.
5.> All multi processors are listed @ their combined frequncy or model number. A dual 3.2 Xeon is listed as a 6400mhz processor. a Dual 2600MP+ is listed as a 5200, a dual opteron 244 (1.8ghz) is listed as a 3600
6.> Pentium M and Centrino are grouped together as they are the same chip. What makes a Centrino vs a Pentium M is what mobo it's installed on. Donthan or Banias on an Intel approved motherboard makes it a Centrino, otherwise its sold as a Pentium M.
I left out a couple of incomplete ones or exotics. IE 14" G4 power Book and the dual Mobile Athlon.
If you stated that you used anything other than PS 7 or PS CS I left you out.
**edit** It's only done to page 14 so far. and yes my spelling stinks. it's 3am I'm all tired and stuff.**/edit**
p.17 #15 · Let us testing Photoshop speed in our system
After switching out my P4 2.6 Northwoold (running at 2.8) for a P4 3.2 Northwood (running at 3.45) my time dropped from 60 to 48 seconds. That seems a little strange, because the 25% performance increase was actually slightly larger than the 23% clock speed increase... maybe I'm just not too good with a stopwatch.
p.17 #20 · Let us testing Photoshop speed in our system
50 sec , checked twice
3.4 GHz P4, 2 Gb RAM, some other stuff running but probably didn't affect it.
OTH, I have had this computer since last May. Normally it is faster than I am.
oh and the 30" display doesn't suck either heheheh
