Guys this is amazing... I just changed the memory on my MBP and now I have 2 GB (which means dual channel etc). The time for this test is now... 1 min 20 sec !!! (from more than 3 min !)
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
'Real World' Photoshop Benchmark - Speed Test
- Thread starter BakedBeans
- Start date
- Sort by reaction score
Guys this is amazing... I just changed the memory on my MBP and now I have 2 GB (which means dual channel etc). The time for this test is now... 1 min 20 sec !!! (from more than 3 min !)
OOPS - tried again.
I re-tried it with history set to "1" - I thought that it was set the first time but it wasn't.
30 seconds.
Tried again, 33 seconds.
On a Mac Pro 1.1, 8GB iRam, 8800GT, Caviar Black, RAID 0 scratch drive, Photoshop CS4
Have Fun,
Keri
I re-tried it with history set to "1" - I thought that it was set the first time but it wasn't.
30 seconds.
Tried again, 33 seconds.
On a Mac Pro 1.1, 8GB iRam, 8800GT, Caviar Black, RAID 0 scratch drive, Photoshop CS4
Have Fun,
Keri
1st comparison:
64 sec PMG5, 2,3Ghz DP, 3,5GB RAM, Leopard, 7200rpm scratch, Radeon 9600 (CS2)
31 sec Q8200, 2,33 Ghz, 4 GB RAM, WinXP, 7200rpm scrarch, Radeon HD 3870 (CS4)
Funny - considering that the specs are in many ways comparable (and that the GPU should not matter) it's as if every core/processor would have as much punch...
2nd comparison (MacBook UNI(Al), 2,4 Ghz with CS4):
UPDATE (24.1.10): While "beefing up" my newest "toy" (bigger, faster HD, more RAM) i decided to log how much the test is influenced.
Due to possible measuring inconsistencies all test were run 5 times. Below are the averages.
1min45sec 2GB RAM, 250GB@5400rpm(original Toshiba)
1min42sec 2GB RAM, 500GB@7200rpm(Seagate momentus 7200.4)
46sec 4GB RAM, 500GB@7200rpm
44sec 4GB RAM, Intel X25-M G2 SSD
- Now I admit I did not expect any major performance increase thru the hard drive update in this test as I can't see it being HD-intensive, so those results are on par with expectations. Hard drive was updated first because after RAM increase HD significance would presumably have declined further.
- I did expect a significant improvement in the RAM extension - even so, the results absolutely blew me away. I had expected it to shave off at best some fifteen-twenty seconds - but to more than halve the time needed seemed so clearly preposterous that I had to redo that last test ten times (with fairly consistent results) before I had the courage to post them. (and yes, I did restart between tests). I'd welcome your views as to why the performance increase is so big... The only difference significant enough to be spotted with the bare eye is that the gaussian blur-phase took ages with 2 gigs, whereas the machine practically wheezed through it at 4 gigs... go figure.
EDIT: Later changed the 7200rpm for an SSD, again, a marginal improvement, which very much belittles the all-round "snappiness" improvement the SSD brought. As I noted in another (later) post, this benchmark is primarily processor and memory-critical, and does thus not give an all-round view of normal-usage Photoshop performance.
Pekka
P.S. First post...since extended a few times.
64 sec PMG5, 2,3Ghz DP, 3,5GB RAM, Leopard, 7200rpm scratch, Radeon 9600 (CS2)
31 sec Q8200, 2,33 Ghz, 4 GB RAM, WinXP, 7200rpm scrarch, Radeon HD 3870 (CS4)
Funny - considering that the specs are in many ways comparable (and that the GPU should not matter) it's as if every core/processor would have as much punch...
2nd comparison (MacBook UNI(Al), 2,4 Ghz with CS4):
UPDATE (24.1.10): While "beefing up" my newest "toy" (bigger, faster HD, more RAM) i decided to log how much the test is influenced.
Due to possible measuring inconsistencies all test were run 5 times. Below are the averages.
1min45sec 2GB RAM, 250GB@5400rpm(original Toshiba)
1min42sec 2GB RAM, 500GB@7200rpm(Seagate momentus 7200.4)
46sec 4GB RAM, 500GB@7200rpm
44sec 4GB RAM, Intel X25-M G2 SSD
- Now I admit I did not expect any major performance increase thru the hard drive update in this test as I can't see it being HD-intensive, so those results are on par with expectations. Hard drive was updated first because after RAM increase HD significance would presumably have declined further.
- I did expect a significant improvement in the RAM extension - even so, the results absolutely blew me away. I had expected it to shave off at best some fifteen-twenty seconds - but to more than halve the time needed seemed so clearly preposterous that I had to redo that last test ten times (with fairly consistent results) before I had the courage to post them. (and yes, I did restart between tests). I'd welcome your views as to why the performance increase is so big... The only difference significant enough to be spotted with the bare eye is that the gaussian blur-phase took ages with 2 gigs, whereas the machine practically wheezed through it at 4 gigs... go figure.
EDIT: Later changed the 7200rpm for an SSD, again, a marginal improvement, which very much belittles the all-round "snappiness" improvement the SSD brought. As I noted in another (later) post, this benchmark is primarily processor and memory-critical, and does thus not give an all-round view of normal-usage Photoshop performance.
Pekka
P.S. First post...since extended a few times.
23.4 seconds
CS4
cache levels: 4
history states: 1
Mac Pro 2008
2x 2.8 GHz
8GB RAM
8800GT 512MB
Booted into 64 bit kernel
on a side note, I did this test with the scratch disk set to my boot drive, is there really a large performance hit when using the boot drive as a scratch disk?
CS4
cache levels: 4
history states: 1
Mac Pro 2008
2x 2.8 GHz
8GB RAM
8800GT 512MB
Booted into 64 bit kernel
on a side note, I did this test with the scratch disk set to my boot drive, is there really a large performance hit when using the boot drive as a scratch disk?
iMac i5/i7 ?
I know this thread is in the Mac Pro section but I'm mystified why (as of yet) no-one has posted their results with the new iMac's.
I have my own vested interest here, because I'm planning on buying one for mainly graphic work (PS, LR, Aperture, Freehand etc.), and would very much like to know how the i5's and i7's compare with the higher-end C2D's...
RGDS,
Pekka
I know this thread is in the Mac Pro section but I'm mystified why (as of yet) no-one has posted their results with the new iMac's.
I have my own vested interest here, because I'm planning on buying one for mainly graphic work (PS, LR, Aperture, Freehand etc.), and would very much like to know how the i5's and i7's compare with the higher-end C2D's...
RGDS,
Pekka
2:21
Hello,
I've the old (from year 2005) Power Mac G5 Dual 2.0 Ghz, with memory inside maccina of 1 TB + 0.5 TB; outside 2 TB + 2 TB as back ups; 4 Gt DDR SDRAM and the basis ATI RADEON 9600, test was on CS4 Photoshop Extented, OSX 10.5.7 and 23" MAC display´s x2.
.. and the time for test was 2 min 21 sec, which is not the worse time here but if with new MAC Pro´s it could be approx 2 min´s less, there is a huge different.
(I did not clean any caches or anything else, was just interested of the normal speed in my maccina.)
While my actions are going, I can still get a nice cup of coffee and read some email´s
BW
Kalevala
ps. thanks for making this test (to give something to think about)
Hello,
I've the old (from year 2005) Power Mac G5 Dual 2.0 Ghz, with memory inside maccina of 1 TB + 0.5 TB; outside 2 TB + 2 TB as back ups; 4 Gt DDR SDRAM and the basis ATI RADEON 9600, test was on CS4 Photoshop Extented, OSX 10.5.7 and 23" MAC display´s x2.
.. and the time for test was 2 min 21 sec, which is not the worse time here but if with new MAC Pro´s it could be approx 2 min´s less, there is a huge different.
(I did not clean any caches or anything else, was just interested of the normal speed in my maccina.)
While my actions are going, I can still get a nice cup of coffee and read some email´s
BW
Kalevala
ps. thanks for making this test (to give something to think about)
CS4-19 seconds
10.6.2
A hackintosh I built for myself with a budet of around $1200.
4.2Ghz i5-750 Quad-Core
8GBs DDR3-1600MHz
64GB SSD Boot/App drive
2TB 7,200RPM Data Disk
9800GT 1GB DDR3 Video Card (basically the same as the 8800GT in the MP's from 2007-8)
This CPU does not support hyper-threading, so I wonder how much of a real world increase hyper-threading would give me with the same GHz rating? Guess I need to get an i7 and drop it in.
10.6.2
A hackintosh I built for myself with a budet of around $1200.
4.2Ghz i5-750 Quad-Core
8GBs DDR3-1600MHz
64GB SSD Boot/App drive
2TB 7,200RPM Data Disk
9800GT 1GB DDR3 Video Card (basically the same as the 8800GT in the MP's from 2007-8)
This CPU does not support hyper-threading, so I wonder how much of a real world increase hyper-threading would give me with the same GHz rating? Guess I need to get an i7 and drop it in.
I built a hack with an i7-920 OC'd to 4.2ghz with everything else just about identical, though 12gb ram at 1600 and a 4890 video card. The hyperthreading does make a difference--about 50% speedup:
8 bit: 8.9 sec
16 bit: 12.5 sec
32 bit: 13.8 sec
yeah, this would be an interesting data point. I'm curious if hyper-threading hurts or helps PS. Someone up-thread said it helped on a Hackintosh, but it would be great to see an iMac i5 to i7 comparison, with the processor speeds normalized to clearly see the benefits (or lack thereof) of HT. Though this post seems to say HT makes no difference.
Found this hyper-threading/Turbo Boost test for the i7 on iXBT. They ran lots of tests for both HT & TB. Quite enlightening.
Hate answering my own questions...
iMac 27" i7, 4GB RAM, 1TB HD (10.6.2)
Out of the box, CS4 installed and configured to test
26 sec.
I have two extra 1 gig memory modules sitting around. Might pop them in to test whether 6gigs (2+2+1+1) beats 4gigs (2+2), but I expect it either:
- will be a purely marginal improvement
- will be a marked slowdown (modules not of same size)
- will show no difference.
Anyone interested?
Pekka
Wow !!! not bad only 10 seconds slower then the Mac Pro above with 32gb of ram & the extra hard drives spinning @ 15,000 RPM & SSD well u get the point.>
>
>
Core i7 920 @ 3.8GHz
6GB DDR3 1600 RAM
Nvidia GTX295
Single 500GB Samsung Spinpoint F3 for scratch
Cost - £1100
Windows 7 64bit
CS4
13.1 seconds.
Also had Firefox running, and a few other apps.
How much did that Mac Pro cost?
I also have a core2duo based MacBook... that was 2 mins and 12 secs... is that right? Seems slow to me. I know it's not gonna be in i7 territory... but over 2 mins?
As I noted earlier, this benchmark is very memory critical:
I also remember not getting past 2m55s with my first gen MBP (2Ghz) running CS2 (rosetta).seveej said:
So if you have more than 2 gigs of ram, and it's not a 1st or 2nd gen MB (and aren't testing with PS CS<3) then that is a bad result. Otherwise I would not be worried...
Pekka
P.S. There have been very little noticeable improvements with memory over 4 gigs - seems current versions of PS do not know hot to use more. After that bottleneck (memory) has been cleared, the competition is very much decided by raw processing power. That's why hackintoshes and windows-pc's (overclocking ability) sometimes get so good figures here. I'd be interested in what Pookeyhead's machine would score at iMac i7 -equivalent clockspeeds (2,8Ghz).
But honestly people - when you start the thread with page 1, and consider how dated this benchmark is (The only step i could even observe on my iMaci7 was the Gaussian Blur) then you start to wonder about the "Field relevance" of even this "Real World" test. For most users the critical issue is whether their machine works "in real time" or whether they have to go for a coffee each time they press a button. I know my credo is: "May I be the bottleneck of this man-machine interface."
Just over 20 seconds @ 2.8GHz.
Triple channel DDR3-1600 awesomeness I was honestly expecting to be the wrong side of 25 secs at 2.8.
RAM timings don't make a massive amount of difference on X58 motherboards, but every little helps. Running my RAM at 1T command rate and CAS6makes the difference.
There's headroom in this rig yet. I can run 24/7 stable at 4.2GHz, but my temps are are a little high at that. Watercooling will cure that.. watch this space I've benched this chip at 4.4GHz, and with water, I'm hoping temps will alow a 24/7 overclock at that.
Triple channel DDR3-1600 awesomeness
I was honestly expecting to be the wrong side of 25 secs at 2.8.RAM timings don't make a massive amount of difference on X58 motherboards, but every little helps. Running my RAM at 1T command rate and CAS6makes the difference.
There's headroom in this rig yet. I can run 24/7 stable at 4.2GHz, but my temps are are a little high at that. Watercooling will cure that.. watch this space
I've benched this chip at 4.4GHz, and with water, I'm hoping temps will alow a 24/7 overclock at that.33 / 34 / 33 seconds
MP 2x 2.66 Dual Core (1.1)
9GB RAM
500GB Samsung single scratch
nVidia 7300GT
OS X 10.5.8
MP 2x 2.66 Dual Core (1.1)
9GB RAM
500GB Samsung single scratch
nVidia 7300GT
OS X 10.5.8
Again answering my own questions. Above mentioned machine, now with 8 gigs: 24 secs.
Purely marginal improvement. But helps with multitasking...
RGDS,
Pekka
Core i7 920 @ 3.8GHz
6GB DDR3 1600 RAM
Nvidia GTX295
Single 500GB Samsung Spinpoint F3 for scratch
Cost - £1100
Windows 7 64bit
CS4
13.1 seconds.
Also had Firefox running, and a few other apps.
How much did that Mac Pro cost?
I also have a core2duo based MacBook... that was 2 mins and 12 secs... is that right? Seems slow to me. I know it's not gonna be in i7 territory... but over 2 mins?
About $200... I upgraded from my last version and sold it for a 10% depreciate... What's your hakincrap worth to a PC user... $700 @ most? ButIdont mind spending money for something that has a 3 year warranty. What will it cost you there mr thrifty arrogant? Have fun with your depreciated windows box... I have a buyer right now willing to givens $2k for this Nehalem and Ill put out only $500 to get a new 6 core (including 8GB of ram) and you'll be trying to sell your PC there for peanuts and no warranty. Good try though.
It'll be interesting to see how CS5 deals with memory over 4 gigs.
So far as those OC'd machines... Fine, Great, you (PC'ers, not charchilp) beat me on a benchmark. Congratulations.
Have you tried living with an OC'd PC or Hackintosh? I tried it once.
No Thank You. I'll stick with my ancient, creaking slower-than-slow MP1.1 and come out ahead on time not spent tinkering.
Have Fun,
Keri
PS. I'm not intending to go with CS5 at least not for now.
Anyone care to give it a try?
So far as those OC'd machines... Fine, Great, you (PC'ers, not charchilp) beat me on a benchmark. Congratulations.
Have you tried living with an OC'd PC or Hackintosh? I tried it once.
No Thank You. I'll stick with my ancient, creaking slower-than-slow MP1.1 and come out ahead on time not spent tinkering.
Have Fun,
Keri
PS. I'm not intending to go with CS5 at least not for now.
Anyone care to give it a try?
yikes sparky calm down, no need to be a jerk.
I'm quite impressed that a machine that only cost £1100 was able to pull such a low number.
With CS4, not CS5 so I am limited to about 3GB RAM my results are...
42 secs: 27" iMac 3.06 C2D
50 secs: 13" MB 2.26 C2D w/4Gb RAM
42 secs: 27" iMac 3.06 C2D
50 secs: 13" MB 2.26 C2D w/4Gb RAM