Actually in this case I just saved to the desktop. So Windows pagefile yes (C drive), but PS scratch disk no. That has its own drive.
I have never touched the plugin. Compatibility on (but agree It shouldn't matter here)
I must be counting too s l l o o o w l y
D Fosse wrote:
on paper your system should be much faster than mine. So I do think I'm coming out of this with my head high after all. At the moment I only have 8GB RAM and 512 MB VRAM, and my quad-core CPU is slower than each of your two. I have DDR3, though, if that matters.
You have a newer machine, and apparently quite well tuned for Photoshop throughput. Excellent! And yes, higher memory bandwidth could clearly matter.
Regarding the save speed, you're clearly seeing your write caching at work, because few systems today can write a gigabyte per second. But that's good - it says you're using your hard drive well, and that you have enough RAM space for such caching.
It's stuff like this that I was hoping we'd turn up in this thread. Combos that work particularly well for Photoshop.
Could you please list more of the details of your system and configuration?
Noel Carboni wrote:
you're clearly seeing your write caching at work, because few systems today can write a gigabyte per second.
Yes, that makes perfect sense.
I'll dig up some more details in a minute. I built it myself, so I'm hoping this is payback for the way I did it: I went for "old" components and technologies that had been on the market for at least a year or so. No cutting edge. This is the explanation for the i5 750: I wanted the 1156 socket, not the newer 1366 required by the "big" i7's.
Minute's up D.Fosse.
I did a bit of cleanup, including turning off buffer flushing for the key HD's.
It seems that simply rebooting helps the Opn/closing time for the test file. Open lost 1/2 sec to open in 1.5 sec, Close is practically instantaneous.
The rest changed very little. But it is encouraging that my worst case at 2500 pix is only 2x the time on Noel's machine His 5 sec, mine 10 sec.)
D Fosse, it doesn't surprise me that your i5 750 performs as well as it does. The only difference I see between the i7 920 and ther i5 750 (without actually having one here, on an Intel board!) is the thread count. The CPU speed is 2.66GHz, exactly that of the 920, and that little puppy was screamingly fast! Of course, that was back in 08/09. Today it's a slug!
Be aware that that LGA socket and the associated cpu's are being discontinued some time this year, if not already.
A final note on stability. Overclocking is a fine way to check it, because if the unit will overclock stably to the level of the next faster version, it is, FAPP, that version (assuming they both are either locked or unlocked) My AMD will not overclock, so I regard it as but marginally stable and certanly not in the next higher speed class, or even at 5% faster. I'll probably replace it with a Phenom. So far, it hasn't crashed at stock +, the plus being I have it set for max performance with all 4 cores active at all times.
So, what you get when you buy the lowest speed at the attractive price point is what is left after gleaning the better units to meet the demand. Therefore, some folks who bought the i7 920 may in fact have a higher speed unit.
I noticed that my computer's score in Passmark had an unusual score for Memory Mark, in that it was equal to or better than the units to which I compared, where, in some instances, they were much better. I also note here that my Open?Close times do as well as Noel's and Fosse.
Minute's up D.Fosse.
Yeah, sorry about that, got sidetracked...and I can't find the documentation without opening the case.
It's a Gigabyte motherboard, I think it's called GA P55 (and that probably refers to the Intel chipset), with four 2GB DDR3 modules. This was a bit of a hassle because the mobo required 1.65 volt modules (IIRC) but most on the market were 1.8 volt. But I found a Kingston matched set, running at 1333 MHz.
BIOS settings are auto whenever possible. I tried to leave it alone.
Four internal HDs, all Samsung, totalling 3 TB. The mobo does RAID, but they all run as single volumes. System 750GB with Windows pagefile. 2 x 1 TB Documents. 250GB PS scratch disk.
Win 7 pro 64 bits. I made a very conscious decision to not fiddle with the Windows install, I just let it do its thing.
Did I forget something?
About the save time: It is of course entirely possible that stuff happens after the cursor stops spinning. That's all I looked at - hit save and watch the cursor.
Yeah, it's the P55 chipset, which means the i5 750 is not directly related to the i7 9xx series.
I'm surprised that the higher voltage RAM could not be accomodated. Probably, unless you were to go into BIOS to make that chage, it wouldn't do it. On the Asus for AMD, BIOS accomodates 1.5 to 2.205 on Auto. Perhaps it's one of the things that is sacrificed for lower cost.
By the way, THANKS for the "Flate" fix, Pierre. I timed a save of an image of a big cat I was working on, with only 4 or 5 layers, and was waiting 10 seconds to get a 140 MB file on disk. With the fix a 220 MB file was created in less than 1 second.
Wow! A 63% increase in file size.
I'll wait for the save to finish!
Glad to be of help, Noel.
Lawrence, that is the idea, if one wants very fast saves of some files, but can afford disk space, it works wonderfully.
At age 74, I have more time than....hmmm, at age 74, maybe I don't have more time!
I am wondering if part of this test could be recorded in an action (a path that could be stroked, with different settings (using tool presets for the different settings) to make a more repeatable benchmark