Simple Photoshop Test to compare i5/395 verses i7/395X

[Macmamamac]

After much deliberating, i chickened out of the maxed out and got the i5/395 ( 256 SSD ) instead.

I have the latest Photoshop installed, and after running a simple speed test it left me a bit underwhelmed.

I am just wondering if there is anyone out there with an i7/395X ( 256/512 SSD ) who can do a quick Photoshop test for me so we can compare results? ( I am thinking of returning the i5 and getting a maxed out i7 instead )

Its a really simple test, here is the set up:

New Document > U.S. paper > 600 Resolution ( image size should say 96.3M )

Filter > Render > Difference Clouds

Once set up is done, speed test is:

Blur > Radial Blur > Amount: 100 > method: Zoom > Quality: Best

Here are my ( non scientific ) results in seconds for the Radial Blur:



48 seconds for a radial blur on a document that is only 96.3M doesn't seem that great. Is it the i5? the 395? or should i upgrade the RAM? ( Have standard 8GB installed at the moment. )

Any help would be much appreciated. ( Also, if anyone else has any other simple tests for photoshop please post them up. )

 

[h9826790]

https://forums.macrumors.com/threads/real-world-photoshop-benchmark-speed-test.200558/

Here is another long lasting photoshop real world test thread. It's inside the Mac Pro forum, but many other Mac participate. You may test your system and compare the results at there.

Anyway, in general, photoshop benefit from RAM and CPU single thread performance the most. Anything before 12G can greately increase photoshop performance. Once more than 16G, the performance increment won't be that significant.

Of course, for some particular filter, or very large image, etc. Better GPU, i7, and 32G RAM can still make significant contribution to the system.

I assume the setting in performance section is correct. the very 1st thing to do will be find out the bottleneck. At this moment I believe is the RAM, 8G sure is enough to run the software, but not good enough for further acceleration. You may install iStat and check the CPU, GPU, RAM usage during the test, and try to work out the real bottleneck then fix it.

 

[joema2]

...I am just wondering if there is anyone out there with an i7/395X ( 256/512 SSD ) who can do a quick Photoshop test for me so we can compare results? ( I am thinking of returning the i5 and getting a maxed out i7 instead )...48 seconds for a radial blur on a document that is only 96.3M doesn't seem that great. Is it the i5? the 395?...

I just did your test on my 2015 iMac 27 with 4Ghz i7, 1TB SSD, 32GB, M395X, Photoshop CC ver 14.2.1: it took 31.3 seconds. iStat Menus showed the operation was using both GPU and CPU, however it appears the CPU was saturated so I'd expect the CPU made a bigger difference than the GPU, memory or SSD.

I also ran the Photoshop test listed in the Mac Pro forum. It took 7.9 sec the 1st execution, and 2:15 the 2nd execution on the resultant image from the 1st pass. I don't know if that is an intended test mode.

 

[MacGizmo]

I have an iMac 27" i7 processor with 32GB RAM, 256GB SSD (not the stupid Fusion Drive) and the 4GB video card upgrade. The test took 22 seconds.

The thing to remember is that it's going to run differently for almost everyone unless everyone is on the exact same machine, with absolutely nothing installed or running but the OS and Photoshop.

I have around a dozen apps running in the background (things like Dropbox, Bartender, Yoink, Spectacle, etc.) and all of them affect the outcome of the test because they're using resources. To see a difference in how an app like Photoshop runs, simply uninstall all the extra fonts the system — keeping only the 10 or 12 the system itself needs to run. You'll be amazed at the speed increase when launching the app and working in menus and dialog boxes.

I suspect someone can get a better score if they run this test on a fresh machine with nothing but Photoshop installed. I just don't want to bother...

 

[Macmamamac]

https://forums.macrumors.com/threads/real-world-photoshop-benchmark-speed-test.200558/

Here is another long lasting photoshop real world test thread. It's inside the Mac Pro forum, but many other Mac participate. You may test your system and compare the results at there.

Anyway, in general, photoshop benefit from RAM and CPU single thread performance the most. Anything before 12G can greately increase photoshop performance. Once more than 16G, the performance increment won't be that significant.

Of course, for some particular filter, or very large image, etc. Better GPU, i7, and 32G RAM can still make significant contribution to the system.

I assume the setting in performance section is correct. the very 1st thing to do will be find out the bottleneck. At this moment I believe is the RAM, 8G sure is enough to run the software, but not good enough for further acceleration. You may install iStat and check the CPU, GPU, RAM usage during the test, and try to work out the real bottleneck then fix it.

Thanks for the reply, really helpful info.

i clicked on the Mac Pro thread but unfortunately the link to the test is not working anymore. Do you know the set up they used?

Also, i downloaded the istat menu you mentioned, during the radial blur test memory doesn't seem to be that stressed by photoshop, although on activity monitor memory pressure turns yellow; not sure if that means a bottleneck or not?

I've attached a few screens of whats happening with CPU and GPU when i run the radial blur test:

 

[Macmamamac]

I just did your test on my 2015 iMac 27 with 4Ghz i7, 1TB SSD, 32GB, M395X, Photoshop CC ver 14.2.1: it took 31.3 seconds. iStat Menus showed the operation was using both GPU and CPU, however it appears the CPU was saturated so I'd expect the CPU made a bigger difference than the GPU, memory or SSD.

I also ran the Photoshop test listed in the Mac Pro forum. It took 7.9 sec the 1st execution, and 2:15 the 2nd execution on the resultant image from the 1st pass. I don't know if that is an intended test mode.

Thanks for taking the time to do the test, its really helped me out because after closing down all apps i took the test again and i got a score of of 36 seconds, so its not too far off of your maxed out. Although, i am sure the gap would grow larger as file size went up, which is why i think the i5/395 might be the wrong choice.

Btw, i asked already but what is that test on the mac pro thread? i can't seem to locate it.

I have an iMac 27" i7 processor with 32GB RAM, 256GB SSD (not the stupid Fusion Drive) and the 4GB video card upgrade. The test took 22 seconds.

The thing to remember is that it's going to run differently for almost everyone unless everyone is on the exact same machine, with absolutely nothing installed or running but the OS and Photoshop.

I have around a dozen apps running in the background (things like Dropbox, Bartender, Yoink, Spectacle, etc.) and all of them affect the outcome of the test because they're using resources. To see a difference in how an app like Photoshop runs, simply uninstall all the extra fonts the system — keeping only the 10 or 12 the system itself needs to run. You'll be amazed at the speed increase when launching the app and working in menus and dialog boxes.

I suspect someone can get a better score if they run this test on a fresh machine with nothing but Photoshop installed. I just don't want to bother...

Thanks for doing the test. Did you have everything running in the background when you got the 22 seconds?

 

[h9826790]

Thanks for the reply, really helpful info.

i clicked on the Mac Pro thread but unfortunately the link to the test is not working anymore. Do you know the set up they used?

Also, i downloaded the istat menu you mentioned, during the radial blur test memory doesn't seem to be that stressed by photoshop, although on activity monitor memory pressure turns yellow; not sure if that means a bottleneck or not?

I've attached a few screens of whats happening with CPU and GPU when i run the radial blur test:

The renewed link is at the very last page

http://scriptsandbits.com/

Anyway, all your 8G RAM was fully utilised, which is good, but also means the system can't use more RAM to further accelerate. More RAM means more cache in current OSX and photoshop. Which may help a lot in some case.

If you look at the numbers, the overall RAM used is more than 8G (because some RAM was compressed, so the system consider you have more than 8G RAM). And the Swap is not zero, which means the system start to use SSD rather than RAM to store temporary data. Luckily the SSD nowadays is very fast and low latency for random access files. The swap also just 185MB, most likely it's just used to store some other idle data, but not really using the SSD to replace RAM for photoshop calculation.

Yellow also mean the system push all the RAM close to the limit (with little performance degradation).

According what Apple says, as long as it's not red, your system is fine. Of course, RAM pressure stay at green is better in terms of performance. Because yellow means there is no more spare RAM to act as cache to further accelerate.

And for i5, total CPU availability should be 400%, photoshop can use up to 332%, which is also very good. But once again, it means you already push your system to its limit. An extra 10% demand will saturate your system.

For GPU, OSX is very good on fully utilising VRAM (on the red's GPU, or it's a bug there VRAM won't be released until requested by another software), so the utilisation almost always high. It's very hard to determine if photoshop push all your VRAM to limit. Or most of the VRAM was actually used by the system and doing nothing at that moment. I personally believe VRAM is not a problem in your system even it looks like the utilisation is above 80% (it's definitely not at the 100% saturated state). And the GPU almost doing nothing. By considering current photoshop is more CPU bounded than GPU bounded. I wouldn't worry the GPU itself is the bottleneck.

If this blur test is more demanding than your normal workflow. Then congratulation, you get a prefect machine to fit your needs, not much wasting / idling resources, and only minor performance hit under full load.

However, if this is just a quick test, and your normal workflow may be more demanding, then sure you can benifit from the RAM and CPU upgrade.

 

[Macmamamac]

Thank you for such a detailed reply. Out of everything I have read online, this has been the most useful.

I'm going to upgrade I think after this. I just have to tell the wife that a guy

The renewed link is at the very last page

http://scriptsandbits.com/

Anyway, all your 8G RAM was fully utilised, which is good, but also means the system can't use more RAM to further accelerate. More RAM means more cache in current OSX and photoshop. Which may help a lot in some case.

If you look at the numbers, the overall RAM used is more than 8G (because some RAM was compressed, so the system consider you have more than 8G RAM). And the Swap is not zero, which means the system start to use SSD rather than RAM to store temporary data. Luckily the SSD nowadays is very fast and low latency for random access files. The swap also just 185MB, most likely it's just used to store some other idle data, but not really using the SSD to replace RAM for photoshop calculation.

Yellow also mean the system push all the RAM close to the limit (with little performance degradation).

According what Apple says, as long as it's not red, your system is fine. Of course, RAM pressure stay at green is better in terms of performance. Because yellow means there is no more spare RAM to act as cache to further accelerate.

And for i5, total CPU availability should be 400%, photoshop can use up to 332%, which is also very good. But once again, it means you already push your system to its limit. An extra 10% demand will saturate your system.

For GPU, OSX is very good on fully utilising VRAM (on the red's GPU, or it's a bug there VRAM won't be released until requested by another software), so the utilisation almost always high. It's very hard to determine if photoshop push all your VRAM to limit. Or most of the VRAM was actually used by the system and doing nothing at that moment. I personally believe VRAM is not a problem in your system even it looks like the utilisation is above 80% (it's definitely not at the 100% saturated state). And the GPU almost doing nothing. By considering current photoshop is more CPU bounded than GPU bounded. I wouldn't worry the GPU itself is the bottleneck.

If this blur test is more demanding than your normal workflow. Then congratulation, you get a prefect machine to fit your needs, not much wasting / idling resources, and only minor performance hit under full load.

However, if this is just a quick test, and your normal workflow may be more demanding, then sure you can benifit from the RAM and CPU upgrade.

Can't thank you enough for this post. The most informative thing I have read. I think because of your recommendations I will upgrade RAM and look into CPU further. Many thanks again for such a detailed post.

 

[Macmamamac]

Ok, so if anyone is interested i found another Photoshop test online. Its an action script, so no need to do any set ups yourself. Interesting to read the comments section with all the older mac speeds.

On my 15/395/8GB RAM with only photoshop running i am getting 55 seconds.

If anyone with an i7/395X setup has couple of minutes and wouldn't mind taking the test it would be much appreciated.

 

[iemcj]

35 seconds right on the head three times in a row. I'm running cs6 though so that might not be taking as much advantage of the cpu as the new cc would. Cuz holy crap 22 seconds seems like a typo lol
running that i7 4.0 and 395x. Hope it helps!

 

[Macmamamac]

35 seconds right on the head three times in a row. I'm running cs6 though so that might not be taking as much advantage of the cpu as the new cc would. Cuz holy crap 22 seconds seems like a typo lol
running that i7 4.0 and 395x. Hope it helps!

Many thanks for doing the test. I've done the test a few times now and its always in the mid 50's. Can't believe its so much slower!

Looks like i'm going to send back and get your set up. BTW, how much RAM do you have? Maybe the extreme difference is a RAM issue?

EDIT: Just did a few more tests and i'm now getting 45 seconds, even as low as 39 seconds, Its not consistent at all anymore.

 

[iemcj]

Did the test on my other two computers. My late 2011 macbook pro with the 2.4 ghz i5 8gb ram and intel hd graphics 3000 did the test in 2:15.

My late 2009 imac 3.06ghz intel core 2 duo with 12gb ram and radeon hd 4670 took a whole 2:54 to do the test. So that test certainly liked the multicores, I'd suspect that's why the i7 does better than the i5 even at a similar clockspeed, the i7 supports hyperthreading which effectively turns it into an 8 core when needed.

 

[eoren1]

I ran the test from scripts and bits on my late 2015 i7/512ssd/m395 with only stock 8 gigs of ram - 9.88 seconds

I checked the sample image and it was from a Canon 20D. Decided to give the iMac a real test and downloaded Image2 from this page which is a sample image from the Canon 5Ds - http://web.canon.jp/imaging/eosd/samples/eos5dsr/

That test took 58 seconds. I'm expecting better from the 32 gigs of RAM coming tomorrow.

On a more 'real world' note, I am coming from a stock middle tier 2011 iMac with an i5 processor, 32 gigs RAM and a 128GB SSD over Thunderbolt. I can tell you the difference is night and day.

I typically do initial adjustments in LR on 5DmkIII files and then edit in PS CC. When done with a few layers, I click Save and used to have to wait for 10 or more seconds for the image to save. Now it's instant.

Loading and scrolling through images is much faster even though the generated previews are literally twice as fast.

I had one image that took a while to load in photoshop but it was actually a 10-shot sequence that I opened as layers while having Safari open with 10 tabs, Chrome playing music via Pandora, Mail, Messages and a few other apps open as well. Oh and Photoshop was only set for using 4gigs of the available RAM. I'm sure the RAM will help greatly with those scenarios.

 

[HalfOnWhole]

35 seconds right on the head three times in a row. I'm running cs6 though so that might not be taking as much advantage of the cpu as the new cc would. Cuz holy crap 22 seconds seems like a typo lol
running that i7 4.0 and 395x. Hope it helps!

Are you talking about the original gaussian blur test or the second script?

On the first test (gaussian blur), i got 31 seconds with i7, 395, 16GB, 1TBSSD.

Yet..somehow lightroom CC is spinning up my fan all the time and not very fluid. HD acceleration is on.

 

[eoren1]

Are you talking about the original gaussian blur test or the second script?

On the first test (gaussian blur), i got 31 seconds with i7, 395, 16GB, 1TBSSD.

Yet..somehow lightroom CC is spinning up my fan all the time and not very fluid. HD acceleration is on.

I was talking about the one in post #8. I think that one is outdated mostly due to the image used which is why I looked for a 5Ds file.

I tried the one in #10 but got some warnings about fonts which threw it off...

I have been running LR for most of the day with occasional jumps to PS and I have not heard any noise (loving the SSD setup). Fans I think have stayed at 1200.

 

[iemcj]

Are you talking about the original gaussian blur test or the second script?

On the first test (gaussian blur), i got 31 seconds with i7, 395, 16GB, 1TBSSD.

Yet..somehow lightroom CC is spinning up my fan all the time and not very fluid. HD acceleration is on.

I was talking about the test the OP proposed with the gausian blur. The second script page just leads to a dead link so I didn't run that. Little jealous you're beating me by 4 seconds but I'm attributing that to me using the older cs6 program which might not be as optimized for the multithreading.

 

[HalfOnWhole]

I was talking about the test the OP proposed with the gausian blur. The second script page just leads to a dead link so I didn't run that. Little jealous you're beating me by 4 seconds but I'm attributing that to me using the older cs6 program which might not be as optimized for the multithreading.

Ha, well I'm jealous your LR is flying. I spend much more time in there myself than in PS. And as a sidenote: when I ran the PS blur test I closed every single other app and the only software I have installed on this new machine is LR and PS. Best testing conditions.

 

[iemcj]

Yeah I have a ton of other stuff installed now on here, I didn't do it from a clean restart so that mighta helped too. Either way I'm more than pleased at how fast this thing is!

 

[roadkill401]

35 seconds right on the head three times in a row. I'm running cs6 though so that might not be taking as much advantage of the cpu as the new cc would. Cuz holy crap 22 seconds seems like a typo lol
running that i7 4.0 and 395x. Hope it helps!

Funny that. I'm running a late 2014 i7 with just the 290x video and a 512ssd with CS6 and even with all my other apps running in the background (imessage, email, chrome, etc..) I get 35 seconds too. I bet the video card is playing little to no help at all as your video card is supposed to be at least 2.5x faster than mine.

 

[iemcj]

The 395x is nowhere near 2.5 times faster than your card dude lol. It's about a 25% improvement depending on task. Blur is almost completely cpu, if you have last years top of the line i7 it's only a 8-12 percent gain going to this skylake model. The biggest improvement with the new imacs is that they stay cool. So for photography this doesn't do much, you're likely not ever going to get your card hot doing anything with photos. Video, gaming, and doing high intensity things like blu ray rips and conversions will get the most good from the new system.

 

[Floris]

I am so jealous of everybody having the 5k iMac, pout.

 

[iemcj]

I am so jealous of everybody having the 5k iMac, pout.

Haha well, some of us just jumped in at a different time. My older imac is from 2009, it still runs, just getting too slow for my needs. I admit that yes it is faster but ultimately... it does the same thing. Yes retina screen is nice to look at and makes my jobs (photographer) easier but ultimately it's not that different. A few seconds faster here and there... The OPs original photoshop test is not something a normal user is ever going to do. I'm a professional photographer and for me this new computer is shaving a lot of time off my photoshop and lightroom work. But for most people it's just a fancier toy to play around on haha

It's kinda like if you have a minivan and someone else has a ferarri. There's going to be times when that 0-60 speed is going to be great but when going to the grocery store to pickup eggs, you're both going to be getting the exact same eggs when it's all said and done, he might make it home a few minutes faster but is it that big a deal? If your job is driving professionally then you very well may need the ferrari but most of the people I see buying these programs are basically casual users who dabble in the higher end stuff. So don't worry too much man, this computer is certainly amazing but there's nothing on this computer I can't do on my 6 year old computer... just slightly slower.

 

[Floris]

Haha well, some of us just jumped in at a different time. My older imac is from 2009, it still runs, just getting too slow for my needs. I admit that yes it is faster but ultimately... it does the same thing. Yes retina screen is nice to look at and makes my jobs (photographer) easier but ultimately it's not that different. A few seconds faster here and there... The OPs original photoshop test is not something a normal user is ever going to do. I'm a professional photographer and for me this new computer is shaving a lot of time off my photoshop and lightroom work. But for most people it's just a fancier toy to play around on haha

I can't link it here, but - I have a site where I hold myself publicly accountable for saving up money to get a new iMac. This way I can't go ahead and spend it on a new dishwasher or vacation or simply the electricity bill. My iMac is 24", from 2009, and showing wear and tear. I am ready for a new one. I have stuff going on that prevents me from having a proper job, so just doing some overtime or a few extra weekends to justify a 3KEuro investment .. it will take me two years to save up and get close. Just jelly that others can already play with awesome tech. Honestly just jealous, but.. I will get there, eventually, and I am hoping the next iMac will also last me a good 6 to 8 years.

 

[jazzer15]

Did the test on my other two computers. My late 2011 macbook pro with the 2.4 ghz i5 8gb ram and intel hd graphics 3000 did the test in 2:15.

My late 2009 imac 3.06ghz intel core 2 duo with 12gb ram and radeon hd 4670 took a whole 2:54 to do the test. So that test certainly liked the multicores, I'd suspect that's why the i7 does better than the i5 even at a similar clockspeed, the i7 supports hyperthreading which effectively turns it into an 8 core when needed.

Interesting. I just did this for laughs with my late 2009 2.8 ghz quad core with 12 GB Ram and the radeon hd 4850 (which I have been thinking of replacing) and it did the test at 1:03. Not so bad for a 6 year old machine, I guess.

 

[iemcj]

Yeah in your case your quad core (compared to my dual core) makes a big difference in that sorta thing. Slow clock speed but being able to do 4 calculations at a time is quite helpful on a blur action.