Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Interesting Adobe Speed Test  (Read 4225 times)
dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« on: September 23, 2005, 03:04:21 AM »
ReplyReply

There is a thread about what computer to use, and here is a real life test you can all use to decide what you want to buy or build.

===============================================
Using Photoshop CS or CS2 (or PS 7)
==================================================

1.) Download Image from http://www.quicklance.com/test.jpg
2.) Save it to computer and then open it up in Photoshop
3.) From there please apply a 'radial blur' with the settings at:

Amount = 100
Blur Method = Spin
Quality = Best

Use the PS timer to see how long it takes to apply this filter. Run the test three times by closing the image each time and rerunning the radial blur test after opening the image fresh. Then average the three runs.

Now, here's the rub for those who wish to run another application while you work in Photoshop: Grab about 1GB of data and start the compression program Win RAR compressing the files (you can just select your entire image collection if you wish). I say use Win RAR because I know it pegs your CPU at 100% and must share the clock cycles with other apps running.

So start Win RAR compressing a bunch of files large enough to run the radial blur test three times, and then average it.
You can get a traial verison of Win RAR here: http://www.rarlab.com/ I don't think it ever expires, just nag screens you on start. I like it much better than Win Zip.

My score:
No applications running except Photoshop, firewall, virus protection, etc., and things like that.

System:
AMD X2 64 4800 and an AMD FX 57 (I don't have the FX 57 any longer, but it ran with the same hardware below:
Mother Board: Asus A8N Premium SLI
RAM: 2 GB Corsair Extreme Pro (1 GB x2) Dual Channel mode.
Video Card: BFG Tech 7800 GTXOC

Only photoshop running:
X2 system: 35 seconds
FX 57 64 seconds

With Win RAR compressing files:
X2 system: 40 seconds
FX 57: 1:48

FX57 2.8 Ghz--the fastest single core desktop processor in the world at this writing: 64 seconds.
X2 4800 2.4Ghz Dual Core CPU: 36 seconds

All your regular programs running, virus scanners, firewalls, etc, plus Win RAR working on a 1GB file. Open Photoshop, and runt eh test:
FX57 1:49 secs
X2 4800 40 secs.

And there you have it. This also goes for other applications that you may want to run while you work in Photoshop, like Win Amp, Windows media payer, both to play music, a video conversion program, etc.

Speed comes at a rpice however, but do not be detered. Just buy one of the lower speed X2 Dual Cores and enjoy about the same speed for hundreds less.

AMD X2 4800 CPU: about 850.00
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #1 on: September 23, 2005, 10:55:29 AM »
ReplyReply

Too lazy to run the compression in the background, but got the following on the spin test with normal apps open and running...

System = dual Xeon 3.4's, XP-32, 4G ram, 1.7G to CS, 3 @ 15K SCSI drives (one OS and primary programs, one CS scratch and one OS paging secondary programs), 2 @ 250G SATA for data:

29.2 seconds.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #2 on: September 23, 2005, 06:54:34 PM »
ReplyReply

Quote
Too lazy to run the compression in the background, but got the following on the spin test with normal apps open and running...

System = dual Xeon 3.4's, XP-32, 4G ram, 1.7G to CS, 3 @ 15K SCSI drives (one OS and primary programs, one CS scratch and one OS paging secondary programs), 2 @ 250G SATA for data:

29.2 seconds.
That's about right for that clock speed and an Itel CPU:

Another Test:

Applications not only running, but working.

1) AVG anti virus running a scan on all files
2) SpyBot Search and Destroy running a full scan
3) RAR compressing a 4.5 GB file
4) Seagate Firewall
5) Spybot Tea Timer
6) ASUS Probe running
7) FireFox--this window
9) All the other processes that run in the backround for Win XP Pro., such as LAN and Onboard sound drivers.

Ran the Blur test three times and got 68, 48, 67.
___
AMD 64 X2 4800
ASUS A8N SLI Premium
BFG 7800GTXOC
Corsair Extreme Pro 2 GB Matched Pair
ANTEC P160
ANTEC True Power EPS 550
Plextor 740A
WD 160GB SATA 150
------------------------

You should be able to do aas weel if not better since you ahve dual CPUs. Remeber though that the AMD is running at it's stock 2.4Ghz speed compared to 3.4 Ghz. But you should have no problem with RAR running.
Logged

61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #3 on: September 23, 2005, 07:11:28 PM »
ReplyReply

My Dual 2.3G5 w/1.5GB ram did the spin in 40 sec. (I'll test again when I get my extra memory in Monday).

This is about what I expected.

Basing performance on just one filter is a bit limiting. Using multiple filters (esp. ones we'd use for Photography editing) would be more telling. Someone with some javascript skills should put together a benchmark that involves several photo-related testes including raw file developing.
Logged
dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #4 on: September 24, 2005, 04:35:15 PM »
ReplyReply

Quote
My Dual 2.3G5 w/1.5GB ram did the spin in 40 sec. (I'll test again when I get my extra memory in Monday).

This is about what I expected.

Basing performance on just one filter is a bit limiting. Using multiple filters (esp. ones we'd use for Photography editing) would be more telling. Someone with some javascript skills should put together a benchmark that involves several photo-related testes including raw file developing.
It's the test everyone uses tho, and I think it is accurate as to how long it will take to do other CPU intensive jobs.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #5 on: September 24, 2005, 04:38:10 PM »
ReplyReply

For those of you testing with Dual CPUs, remember that the AMD Dual Core is a Single CPU, and probably much cheaper than two full CPUs for the same relative speed. I'd like to see what the Dual CPU's can do running the RAR compression with the Photoshop test. That would be interesting.
Logged

davidr805
Full Member
***
Offline Offline

Posts: 102


« Reply #6 on: September 24, 2005, 09:51:57 PM »
ReplyReply

I don't find the PS timer??? any helpp!!!!Huh
Logged
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #7 on: September 25, 2005, 10:40:48 AM »
ReplyReply

Quote
I'd like to see what the Dual CPU's can do running the RAR compression with the Photoshop test. That would be interesting.
Didn't bother with your rar or AV, but I did run the spin test while RSE was batch converting a bunch of raw files, IE was open, Outlook was open and Yahoo! IM was open and running, AVG anti-virus running but not scanning.

I got 32.7 seconds.

I should point out 1) that I also have my dual Xeons set to hyperthread, so it looks like 4 processors on my system manager; 2) that I have the CS "large tiles" switch activated; 3) that I ran this test three or four times and the times improved about 7% over the series of runs to the stated times -- not sure why that should be the case...
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #8 on: September 25, 2005, 10:44:12 AM »
ReplyReply

Quote
I don't find the PS timer??? any helpp!!!!Huh
Click on the little triangle down at the bottom of the open image frame...
Logged

davidr805
Full Member
***
Offline Offline

Posts: 102


« Reply #9 on: September 25, 2005, 12:22:23 PM »
ReplyReply

First test 188 seconds ... thats bad .

 planing on buying the power book 1.67 ghz   how fast should it  be?


Hardware Overview:

  Machine Name:   PowerBook G4 15"
  Machine Model:   PowerBook5,2
  CPU Type:   PowerPC G4  (1.1)
  Number Of CPUs:   1
  CPU Speed:   1 GHz
  L2 Cache (per CPU):   512 KB
  Memory:   768 MB
  Bus Speed:   167 MHz
Logged
61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #10 on: September 25, 2005, 01:15:34 PM »
ReplyReply

Quote
It's the test everyone uses tho, and I think it is accurate as to how long it will take to do other CPU intensive jobs.

That's not my point, and not really...

Different filters use the CPU differently. Some are multi-threaded, some aren't. Not only that but there are different types of CPU useage. One CPU test could tell a very different story than a different CPU test.

This is why a variety of filters would be much more telling and usefull.

Quote
1) that I also have my dual Xeons set to hyperthread, so it looks like 4 processors on my system manager;
Just for the sake of general information: It should be noted that HT only provides a 5-15% boost in performance vs no hyperthreading.

Quote
that I ran this test three or four times and the times improved about 7% over the series of runs to the stated times -- not sure why that should be the case...

I'd imagine It's Photoshop taking processor priority over the other apps. The more you used PS, te more the system realized that should take priotity. I you run the test with no other apps running you should find that the speed stabalizes more quickly.

Quote
For those of you testing with Dual CPUs, remember that the AMD Dual Core is a Single CPU, and probably much cheaper than two full CPUs for the same relative speed. I'd like to see what the Dual CPU's can do running the RAR compression with the Photoshop test. That would be interesting.

Although the Dual-core system is a single CPU that isn't what's important. It's the core that does the processing. Both a Dual-CPU system and a single-dual-core system have two cores. One just has the cores on a single dye. (*Tech people don't knit-pick please. i'm generaliseing so a full essay doesn't have to be written, thanks)

A dual CPU system would be fairly similar to a Dual Core system if not a little faster (due to the ability to buy faster CPUs)
Logged
dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #11 on: September 26, 2005, 01:09:40 AM »
ReplyReply

Quote
First test 188 seconds ... thats bad .

 planing on buying the power book 1.67 ghz   how fast should it  be?


Hardware Overview:

  Machine Name:   PowerBook G4 15"
  Machine Model:   PowerBook5,2
  CPU Type:   PowerPC G4  (1.1)
  Number Of CPUs:   1
  CPU Speed:   1 GHz
  L2 Cache (per CPU):   512 KB
  Memory:   768 MB
  Bus Speed:   167 MHz
If you work with more than one app at a time, I would advise sual core CPUs, and Itel has them now too. Although the Intel chips are not 64 bit, and AMDs are.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #12 on: September 26, 2005, 01:30:47 AM »
ReplyReply

Quote
Quote
It's the test everyone uses tho, and I think it is accurate as to how long it will take to do other CPU intensive jobs.

That's not my point, and not really...

Different filters use the CPU differently. Some are multi-threaded, some aren't. Not only that but there are different types of CPU useage. One CPU test could tell a very different story than a different CPU test.

This is why a variety of filters would be much more telling and usefull.

Quote
1) that I also have my dual Xeons set to hyperthread, so it looks like 4 processors on my system manager;
Just for the sake of general information: It should be noted that HT only provides a 5-15% boost in performance vs no hyperthreading.

Quote
that I ran this test three or four times and the times improved about 7% over the series of runs to the stated times -- not sure why that should be the case...

I'd imagine It's Photoshop taking processor priority over the other apps. The more you used PS, te more the system realized that should take priotity. I you run the test with no other apps running you should find that the speed stabalizes more quickly.

Quote
For those of you testing with Dual CPUs, remember that the AMD Dual Core is a Single CPU, and probably much cheaper than two full CPUs for the same relative speed. I'd like to see what the Dual CPU's can do running the RAR compression with the Photoshop test. That would be interesting.

Although the Dual-core system is a single CPU that isn't what's important. It's the core that does the processing. Both a Dual-CPU system and a single-dual-core system have two cores. One just has the cores on a single dye. (*Tech people don't knit-pick please. i'm generaliseing so a full essay doesn't have to be written, thanks)

A dual CPU system would be fairly similar to a Dual Core system if not a little faster (due to the ability to buy faster CPUs)
What I meant is that in the MAC forum, they are using the same test, and in the HardO PC forum, the same test. And I think I read somewhere that this is same test and image Adobe uses for marketing, but not sure on that so don't quote me. Also, CPU usage is CPU usage as far as we're concerned. You probably won't find a CPU intensive task in PS that will see an advantage with one chip over another--in other words, if both chips are multithread capable, and your chip is faster than mine with the radial blur test, you will probably see the same advantage with every other test in PS we could cook up--but please prove me wrong here. That would be interesting. One way I think you are right, however, is with a dual core CPU adn a single core "multithreaded" CPU, in which case the true dual core will have an anvanage at some point.

Well, I think hyperthreading will provide more than 15% boost when two or more applications are running full bore, say a C++ programming language compiling and running a video game at the same time, like my friend does since he programs for Mythic. In fact, he moved from his Intel single core non multithreaded CPU to an Intel MT CPU for the gain in performance while running two or more CPU intensive applications. He then moved recently to the AMD dual core, and said the difference is night and day--as you can attest to with your true dual CPU set up. But yes, 5-15% IF only one program is being used and IF the application utilizes MTing. More than 15% then is more than one app is running?

I ddn't mean that the Dual Core was a single CPU, but like you say two on teh same dye. It's two K8 cores running at 2.4 Ghz each. What I was saying is that you can buy one CPU  that fits into a single CPU MB much cheaper than buying two full CPUS. And you get about the same performance--with AMD's bus system, you actually get tons more performance than even two full Intels running more than 1Ghz faster each than the AMD X2 4800. And with proper cooling you can OC the X2 4800 to 2.8 with stability and longevity.

I thik if you had two CPUs with two seperate busses for each, in theory you could be lots faster, since the Dual Core shares the bus, and that is a potential bottleneck; but I don't think that bottleneck, given the AMDs great bus arcehtecture, has been reached yet. So yeah you're right I think about that too. I ahve no idea about Intels new dual cores.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #13 on: September 26, 2005, 01:33:01 AM »
ReplyReply

I wish some of you would run teh RAR/Blur test. Come on, it's easy to set up and use.
Logged

davidr805
Full Member
***
Offline Offline

Posts: 102


« Reply #14 on: September 26, 2005, 10:07:39 AM »
ReplyReply

Its apple using sual core CPUs ? on powerbooks
Logged
61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #15 on: September 26, 2005, 10:36:53 AM »
ReplyReply

Quote
Also, CPU usage is CPU usage as far as we're concerned.
That's not how computers work. Different types of tasks use different types of math and thusly use the CPU differently. If you want proof of that, then just look up any number of countless CPU speed tests.

The best and most telling tests will combine different types of simulations along with real-world testing. (just like cameras!)

Testing just the radial blur tells us how fast radial blur will work. It does not tell us any real-world information such as how quickly a 6mb raw file will be developed.

Don't get my wrong, this is a fun little test; a good diversion but it is far from telling of overall PS performance.

Quote
Well, I think hyperthreading will provide more than 15% boost when two or more applications are running full bore... But yes, 5-15% IF only one program is being used and IF the application utilizes MTing. More than 15% then is more than one app is running?
No offense but that doesn't make sense. We're talking about overall load here. The CPU can only handle so many threads. It doesn't matter if you are running one multi-threaded app on a HT system or multiple single-threaded apps, the gain in processing speed with HT on will be the same.

If anything, the situation with only one multi-threaded app running would probably perform better than multiple apps since it would have the CPU all to itself.

Quote
I ahve no idea about Intels new dual cores.
Intels current dual-cores are basically just two regular CPUs stuck together with some bailing wire and spit. AMD really caught them with their pants down. They have some new chips coming out next year that are true dual-cores (which macoids are hoping will make it into the upcoming Intel-Macs).

Quote
Its apple using sual core CPUs ? on powerbooks
No duals for powerbooks. They only have the pathetic G4.
Logged
dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #16 on: September 26, 2005, 07:24:37 PM »
ReplyReply

Quote
Its apple using sual core CPUs ? on powerbooks
I meant to say Dual Core--it just means that there are two CPUs on one dye, which looks like one CPU. I have no idea what MACs are doing. I've always been a PC person.
Logged

dwdallam
Sr. Member
****
Offline Offline

Posts: 2044



WWW
« Reply #17 on: September 26, 2005, 07:33:25 PM »
ReplyReply

Quote
Quote
Also, CPU usage is CPU usage as far as we're concerned.
That's not how computers work. Different types of tasks use different types of math and thusly use the CPU differently. If you want proof of that, then just look up any number of countless CPU speed tests.

The best and most telling tests will combine different types of simulations along with real-world testing. (just like cameras!)

Testing just the radial blur tells us how fast radial blur will work. It does not tell us any real-world information such as how quickly a 6mb raw file will be developed.

Don't get my wrong, this is a fun little test; a good diversion but it is far from telling of overall PS performance.

Quote
Well, I think hyperthreading will provide more than 15% boost when two or more applications are running full bore... But yes, 5-15% IF only one program is being used and IF the application utilizes MTing. More than 15% then is more than one app is running?
No offense but that doesn't make sense. We're talking about overall load here. The CPU can only handle so many threads. It doesn't matter if you are running one multi-threaded app on a HT system or multiple single-threaded apps, the gain in processing speed with HT on will be the same.

If anything, the situation with only one multi-threaded app running would probably perform better than multiple apps since it would have the CPU all to itself.

Quote
I ahve no idea about Intels new dual cores.
Intels current dual-cores are basically just two regular CPUs stuck together with some bailing wire and spit. AMD really caught them with their pants down. They have some new chips coming out next year that are true dual-cores (which macoids are hoping will make it into the upcoming Intel-Macs).

Quote
Its apple using sual core CPUs ? on powerbooks
No duals for powerbooks. They only have the pathetic G4.
Cook up some more CPU tests and well see if there is a great deal of differnece between our compared radial blur speeds. It should be easy algebra (to break down our scores in radial blur) to see if different CPU tasks make real world differences here. Technically, you're correct, but I think you're splitting hairs--maybe not. Let's try it.

We could probably just run Sandra Sisoft and convert those scores to our compared score in Radial Blur using a percentage for comparison, CPU to CPU. I'm game.

I'd be interested if you would just come up with a couple more tests in PS if you can think of any that you consider to use different CPU tasks. Then we could could convert and compare those.

However, we're fooling ourselves, since a true dual CPU ,may be faster than a Dual Core CPU simply becasue of that fact, not CPU tasks. What do you propose to give a more accurate speed test of the CPU?

Perhaps we could run Sisift Sandra, although that is not real world testing?
Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad