Need help benchmarking a image resizing tool (CPU and GPU)

24HZ

Member
May 25, 2013
52
0
0
I hate to ask this with a zero track record here. I used to post here a looooooong time ago but it was so long ago that I forgot the old credentials not that I really care for them.

So here's the backstory, I'm heading a project at work to process hundreds of thousands of images using Photozoom Pro and I have a budget to put together a few machines to run this on. I'm just looking for the best way to spend the money to get the best performance/$ so going all out top shelf might not be prudent.

The tool itself is quite niche so there's not much sample data out there across different CPUs and GPUs. It operates in 2 modes BTW, CPU or GPU+CPU with GPU+CPU being the fastest. I'm thinking of running 2 instances per machine so I can take advantage of all resources.

The trial should work as it has everything turned on except that it watermarks the output image. It even comes with a sample image as well which could serve for the test.

http://www.benvista.com/downloads

Here's what I'm getting with my really old personal machine:


CPU: Q6600 @ 2.83Ghz - 1m:17s (77 seconds)
GPU: GTX460 - 0m:17s (17 seconds)



Here's how I'm running it:

1)Start the app
2)Choose 'Later' for registration/serial entry
3)The sample image should be loaded automatically
4)On the left panel, set 'Width' to 400%. Height should automatically update to match
5)On the left panel, set 'Resize Method' to 'S-Spline Max'
6)On the left panel, set 'Preset' to 'Photo - Extra Detailed'
7)Hit the 'Save' button in the upper left
8)Choose 'No' on registration again
9)Choose TIFF as the format and use whatever filename
10)TIFF options I'm using are 'LZW' for compression and 'IBM PC' for byte order not that these should affect the performance
11)Hit 'OK' and the elapsed time shown should be the GPU score since the app uses GPU by default

12)Go to Options > Preferences > Processing, and uncheck 'Use GPU acceleration', hit OK
13)Hit the 'Save' button in the upper left again, the previous settings will be the same
14)Choose 'No' on registration again
15)Choose TIFF as the format and use whatever filename, overwrite is fine
16)TIFF options I'm using are 'LZW' for compression and 'IBM PC' for byte order
17)Hit 'OK' and the elapsed time shown will now be the CPU score

I would be really grateful if anyone here can test with a stock and OC CPUs as well as GTX680 or faster GPUs. I will consolidate the data here in the first post for anyone who posts results. Thanks!

edit: Correction on how the app works. It looks like with GPU acceleration on it still pegs the CPU pretty hard(I've seen 90% over 8 cores) so 2 machines with the same GPU but different CPUs should give different results.

edit: Results chart with data from 2 forums

 
Last edited:

Enigmoid

Platinum Member
Sep 27, 2012
2,907
31
91
On my laptop.

GPU not supported.

i7-3630qm at 3.2 ghz/660m (not supported--?)

Run 1 -- 30 sec
Run 2 -- 34 sec
Run 3 -- 32 sec
 

24HZ

Member
May 25, 2013
52
0
0
On my laptop.

GPU not supported.

i7-3630qm at 3.2 ghz/660m (not supported--?)

Run 1 -- 30 sec
Run 2 -- 34 sec
Run 3 -- 32 sec

Thanks! Not sure whats going on with the GPU but it is their first major version with GPU support. Did you 'GPU speed optimization' step?
 

24HZ

Member
May 25, 2013
52
0
0
Testing on a few machines at work:


Dual Xeon X5482 3.2Ghz(2x4 cores) - 33 seconds
Quadro FX5600 - 18 seconds
 

24HZ

Member
May 25, 2013
52
0
0
CPU: i7 990X(stock) - 22 secs
GPU: Geforce GTX Titan (stock) - 10 secs

Wow Titan, thanks!

Some more from workstations at work:

Dual Xeon X5570 (2x4 cores, HT off) - 28 seconds
Quadro FX4800 (GTX 260? Haha, why are you inside a z800???) - 17 seconds
 

24HZ

Member
May 25, 2013
52
0
0
Correction on how the app works. It looks like with GPU acceleration on it still pegs the CPU pretty hard(I've seen 90% over 8 cores) so 2 machines with the same GPU but different CPUs should give different results.
 

unixwizzard

Senior member
Jan 17, 2013
205
0
76
ok I'll bite:

CPU: i7 3770 - 24 secs
GPU: Radeon HD 7770 GHz Edition - 19 secs

The CPU and GPU are both running at stock speeds, haven't gotten around to any OC tweaking yet.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
3930k@ 4Ghz with 2x GeForce 680s

CPU time: 17 seconds 100% CPU usage
1 GPU time: 8 seconds 100% CPU usage
2 GPU time: 8 seconds 100% CPU usage


Enabling SLi didn't improve performance at all, it performed identically. Either its CPU limited or it only uses a single card via the openCL interface. Quite a few openCL based programs show similar CPU usage patterns with NVidia cards, often AMD uses a lot less CPU in openCL, depending on the program of course. But based on the speed up I suspect the program does not use Double Precision floating point for its calculations and hence its unhindered on a 680 which is crippled for compute on DP.
 
Last edited:

24HZ

Member
May 25, 2013
52
0
0
3630k@ 4Ghz with 2x GeForce 680s

CPU time: 17 seconds 100% CPU usage
1 GPU time: 8 seconds 100% CPU usage
2 GPU time: 8 seconds 100% CPU usage


Enabling SLi didn't improve performance at all, it performed identically. Either its CPU limited or it only uses a single card via the openCL interface. Quite a few openCL based programs show similar CPU usage patterns with NVidia cards, often AMD uses a lot less CPU in openCL, depending on the program of course. But based on the speed up I suspect the program does not use Double Precision floating point for its calculations and hence its unhindered on a 680 which is crippled for compute on DP.

Thanks! Are you sure your CPU model is correct? I couldn't find it online.

edit: Oh I see now in your sig. You mean 3930K.

edit2: Anyone with a GTX 780 ?
 
Last edited:

24HZ

Member
May 25, 2013
52
0
0
Actually, it would be cool to see how well this tool scales with an AMD FX-8 series CPU. Anyone game?
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
Wow Titan, thanks!

Some more from workstations at work:

Dual Xeon X5570 (2x4 cores, HT off) - 28 seconds
Quadro FX4800 (GTX 260? Haha, why are you inside a z800???) - 17 seconds

Well...it's OpenCL, not CUDA...so the Titan dosn't show it's full potentail
 

24HZ

Member
May 25, 2013
52
0
0
Well...it's OpenCL, not CUDA...so the Titan dosn't show it's full potentail

And yet its still doing much better with Nvidia GPUs vs AMD.

Here's all the data I've collected since starting this thread on this forum and another:

edit: removing old chart
 
Last edited:

24HZ

Member
May 25, 2013
52
0
0
New chart. Added a few 2600s and a 16 core Opteron.

edit: see op
 
Last edited:

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
Based on the data it looks like what you want is:

1) Fastest CPU you can get, probably 6 core Intel SB-E. If your overclocking then a 3930k will do it, otherwise you'll want to splash out on the 3970X or it wont perform much like mine did (3.2 instead of 4.0 Ghz on 6 cores).

2) Any 600 series NVidia card, a 660 ti looks like it would be sufficient without loosing any performance. The 460 result suggests that maybe you could even go this low without any major impact as that result is looking like its largely driven by the CPU performance.

A GPU definitely improves things, but you want to max out the CPU and lots of cores first, then the GPU is secondary, needs to be NVidia but mid range is fine.
 

24HZ

Member
May 25, 2013
52
0
0
Yeah I garnered that much. Thing is, this will be for work but I'm testing on personal hardware since nothing has been green lit at work yet but I wanted to show some data and get some extra brownie points.

Once things kick off at work, the CPUs will have to be Xeons as the only machines we buy are z820s. Shouldn't be a problem to get 2-3 16 core z820s once I show them this data. GPU might be trickier. We probably only have Quadros on stock but it looks like high end consumer would actually be the top tier like the newer Titans and 780s. Money wise those are cheaper than the Quadros we have but the studio isn't going to buy Titans and 780s when we have $2000+ Quadros on hand. But I might be able to talk them into at least 2x 780s if I'm charming enough.

Once I get a loaner test machine next week, I'll post the numbers, hopefully, from a 16 core sandy bridge Xeon with some flavor of stupidly expensive Quadro.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,361
136
AMD Athlon II 750K @ 4.2GHz = 1.04 mins
AMD Athlon II 750K @ 4.2GHz + HD7950 @ 1GHz, 1500MHz Memory = 18secs

 

Imouto

Golden Member
Jul 6, 2011
1,241
2
81
Q6600 (stock) + HD5850 (stock) = 27 sec
Q6600 (stock) = 92 sec (1:32)
 

Durvelle27

Diamond Member
Jun 3, 2012
4,102
0
0
CPU: AMD FX 8320 @ 4.4GHz = 29 secs





GPU: AMD Radeon HD 7870 @1200/1450 = 13 secs

 
Last edited:

24HZ

Member
May 25, 2013
52
0
0
CPU: AMD FX 8320 @ 4.4GHz = 29 secs





GPU: AMD Radeon HD 7870 @1200/1450 = 13 secs


Thanks! But I think your results might be slightly fast because you chose 'Photo - Detailed' and not 'Photo - Extra Detailed'. I tested the difference between them and it was ~10%. I will pad your results by 10% unless you want to run it again at the correct setting.
 

Durvelle27

Diamond Member
Jun 3, 2012
4,102
0
0
Thanks! But I think your results might be slightly fast because you chose 'Photo - Detailed' and not 'Photo - Extra Detailed'. I tested the difference between them and it was ~10%. I will pad your results by 10% unless you want to run it again at the correct setting.

CPU







GPU

 

24HZ

Member
May 25, 2013
52
0
0
Ok, this will probably be the last chart. I missed entering a few data points but they're there now and the chart is getting so big its hard to read.

 
Last edited:
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |