I was wondering if anyone has better than i7 920 computer running at 2.6Ghz or faster and can time D800 RAW file processed in ACR while PS is already opened. It takes 5 seconds on mine running at 4.2GHz but it does not matter because I can drop it to your speed for comparison purposes.
Maybe or maybe not better but probibly bigger and more expencive. i7 980 3.33GHz = 3.2sec. Other stuff that might or might not matter: WIN 7-64, ASUS Rampage III Black Edition (for i7 support and more PCI slots not for over clocking), 12G RAM, 2-SSD RAID-0 for OS & programs, 2-SSD RAID-0 for Swap files, 6-SATA RAID-10 for dynamic files (NEF, DOCs, WEB site files, Program downloads, etc.), 6-SATA RAID-10 for system backups, fonts, clipart, music, videos and other things that don't change often, 1-SATA with a clean OS (unconnected), NVIDIA graphics card that supports OpenGL, and a bloody big case with lots of fans. I built this last year and added an internal USB-3 multi-card reader and a 4-port USB-3 PCI card after experiencing D800 NEF download times.
Interestingly the NIK filters open and regen back to PS much slower than any image modifications done in PS. The OpenGL use by Adobe? The exception being the JPG image generation during "Save to Web . . ."
Question. Did you open a single file or a batch of 5 or 10 for example? If I open a single file I get 3.2 seconds but if I do batch I get 5 seconds. And it is not HD because I tried to open a batch of files from SSD with the same results. I also have 12GB of RAM. I just don't have as many drives as you have. Single SSD (Vertex 3) for OS and software. Raid Zero (WD Black 2tb) for data. Water cooling with external radiator. ATI 6870 for Video. I did a lot of experimenting and setting scratch file to SSD or HD does not make any difference. SSD in Raid does not make any difference. Changing speed makes a lot of difference and HT makes a difference. Thinking going with 3770K and getting close to 5Ghz but Ivy Bridge is difficult to cool due to smaller cooling surface and Intel's screw up with not soldering heat spreader to the chip. Maybe I should wait for the next generation
Sorry to take so long to reply - got busy with stuff. My reported time was the time to read one D800 file from ACR into PS. I havent experimented with water cooling yet. Mostly I try to get a gamers MB figuring if it will take overclocking it might be reliable and fast enough for 4-5 years. I recently installed 24G @ 2400MHz of RAM with a cooler as recommended by ASUS if all of the RAM slots were filled. That was an increase from 12G @ 1600MHz. I sense an increase in performance (physiological?) but I have not run any tests.
I am using a R-0 with 2 SSDs to double the capacity as much as for increased speed. The OS drives and the R-10 data drives are controlled by a LSI sub-system with 1G of RAM.
I am very happy to see there other people here that are also crazy enough to take gear over the top.
Mon 19-Nov-12 01:45 AM | edited Mon 19-Nov-12 01:54 AM by DAJolley
My recently built i5-3570K (3.4GHhz) overclocked to 4.08GHz running Win 7 Pro 64bit opens my D800E NEFs (14 bit lossless compressed) files in just under 4 secs. NEF file size might impact your comparison; I have NEFs from 38-56MB in size depending on subject matter. Dave Jolley
David Jolley Pickerington, Ohio Please visit my Website
I have tried to group small files and large files and process them separately. No difference at all. Your CPU lacking HT that makes a difference. Turning HT off on my setup slows down by 20%. How much memory do you have?
Currently I have 16GB of memory, I am adding another 16GB later this week. My OS and scratch disk are on a 256GB Samsung 830 SSD. I considered the i7-3770K but the $80 price difference didn't seem worth the minor performance boost. Dave
David Jolley Pickerington, Ohio Please visit my Website
>Currently I have 16GB of memory, I am adding another 16GB >later this week. My OS and scratch disk are on a 256GB Samsung >830 SSD. I considered the i7-3770K but the $80 price >difference didn't seem worth the minor performance boost. >Dave
You may need 32GB SDRAM for a particular reason. However, it could actually slow the system down depending on the speed of the memory and sometimes more importantly, the latency. The more sticks of memory, the harder it is to maintain a low latency. Be careful to get matched memory or you may not achieve their specs. If you go from 16GB of matched high speed, low latency memory to 32GB of the same speed, same latency, matched memory, that $80 delta to go to i7-3770K will seem like chump change.
I just built a I7-3770k system, not OC'd with 16G of 1600mhz memory.
My timing comparisons were primarily D800 files and time to build 1:1 previews. This setup got them down to 5-6 seconds each when doing a series.
I just tried this experiment as follows: opened CS6, chose 5 files in the open dialog, and let them open (applying the XMP file from lightroom) - 47 seconds. So a LOT slower than what you had. These were FX frame D800 shots with lossless compression, so were about 45-50MB.
That's a LOT slower than you are getting, comparing 4.2GHz vs. my 3.5GHz can only account for about a second of it.
You are opening to 16 bit?
I have a very conservative disk setup - Raid-1 for SSD's for OS, Raid-1 with two 2-TB for the files, temp and ACR cache on one SSD (non-raid). Windows caching turned off for drives. But I saw very little disk activity while doing the test, I think it is processor limited, so I am now puzzled over the speed difference.
I have hyper-threading turned off, with 4 cores and single user I thought that might be better. when I get some time will turn it on and see what I get.