[walterfog]

hello,

i’ve got a very hard problem using MR v3 on my new pc workstation i7 4930k 6 cores 12 thread with an intel 180 GIGA SSD disc and windows 7 pro. I try to explain:
the problem is that rendering scene at high resolution as 5680 x 4016 px the pc workstation when the SL is updating it took a very long long time to write the MXI files.. writing a 3,2 GIGA of MXI for an image of 5680 x 4016 px MR take about 6 minute and 53 second. The same scene, in an old 2008 MAC PRO 2,8 GHZ 8 cores, with the same SSD disc, writing the same MXI of the same scene, it take 1 minute and 48 second. And the same thing happen for both MR 2.7 and MR ultimate v3 version.
The problem is not the PC hardware or the SSD disc.. i try several test, and using the disc for coping and writing from a folder to another, the speed is the same from PC to Mac Pro… even if i use other 3D software.. the problem is only using MR...
This is a very great trouble for me… for high resolution the PC workstation bench drops dramatically… in fact if i try the benchwell for small resolution with small MXI file, the bench of the PC is more than double of the Mac Pro… but if we think at the image describe above, the difference of the bench from pc to Mac Pro is only 30 points…INCREDIBLE!!
I don’t know really what to do.. i try to unistall and re install MR… starting MR as administrator… the problem still remain !!! I install a SSD disc on pc workstation to maximize the speed of MR … and instead in writing MXI it is slower than a normal SATA disc.. what’s happen?
For me is not acceptable a situation like this, cause that mean that for 3 update in a hour, i spend about 20 minute in MXI writing operation…. that’s crazy thing!
Please help me as soon as possibile..
best regards
walter

[Bubbaloo]

How much free space does your 180 gb SSD have left?

[walterfog]

How much free space does your 180 gb SSD have left?

the free space is 140 giga ...... inside there is only windows 7... maxwell render... cinema 4d... nothing else!!!

[Mihai]

Just to check if Maxwell or copying large files to that disk in general is a problem, can you try just moving the MXI to another disk, then copying it to that SSD? So without using Maxwell at all.

I really don't think an SSD or not should have an influence on how fast Maxwell writes this file to it....and you say the same disk works fine on OSX.

Another option you have is to limit the nr of MXI writes by going to the Maxwell preferences, File>Preferences>Min time for saving to disk. Set it to 60min for example.

[walterfog]

Just to check if Maxwell or copying large files to that disk in general is a problem, can you try just moving the MXI to another disk, then copying it to that SSD? So without using Maxwell at all.

I really don't think an SSD or not should have an influence on how fast Maxwell writes this file to it....and you say the same disk works fine on OSX.

Another option you have is to limit the nr of MXI writes by going to the Maxwell preferences, File>Preferences>Min time for saving to disk. Set it to 60min for example.

for the first option, the computer have only one disk.. the SSD disk...

the difference between an SSD disk and a normal disk on maxwell update MXI is really important just because in a SSD disk, the read and write operation are faster than a normal disk... to better understand what i'm talking about, i show you 2 picture, of the resources monitor, that clearly show the different speed on maxwell write and windows write...



as you could see the problem is in how maxwell communicate whit windows or the disk, cause normally the disk is ultra fast in windows routine...

[Mihai]

I can't say I've experienced such low write speeds for an SSD with Maxwell....can anybody else with an SSD on Windows comment please?

[eric nixon]

Well if the pc only has one hd.(why just one hd?). Its probably accesing the page file which is on that disk, maybe maxwell is using the page file when refreshing the render? or maybe intel ssd's are just funky, I dont know..

[mashium123]

I can't say I've experienced such low write speeds for an SSD with Maxwell....can anybody else with an SSD on Windows comment please?

did a quick test with win7 pro, ssd 512gb samsung 840 pro on an i7 4930K oc@4,2ghz, 32gb ram:

- generated a bunch of copies of one 600mb mxi-file (so i had around 15gb-20gb of data-volume to have time to read the i/o-rate) and copied this bunch of files:
i/o at it's peak: ~400mb/s
- opened the simball-scene in studio (latest v3-beta), activated multilight at cranked the resolution up to 5K*5K to get a fairly big file (~600mb while rendering, ~1.5gb when being written finally, after stopping the render), wiated until SL4, stopped:
i/o at it's peak: ~100mb/s

no active swap file on ssd (using the 2nd drive (no ssd) for swap-file purposes).

[walterfog]

Well if the pc only has one hd.(why just one hd?). Its probably accesing the page file which is on that disk, maybe maxwell is using the page file when refreshing the render? or maybe intel ssd's are just funky, I dont know..

the PC has only one HD cause it is a pc that i use only for maxwell... nothing else... why does a second HD could get better performance in this machine?
sorry i don't understand well what you would tell me...

[walterfog]

I can't say I've experienced such low write speeds for an SSD with Maxwell....can anybody else with an SSD on Windows comment please?

did a quick test with win7 pro, ssd 512gb samsung 840 pro on an i7 4930K oc@4,2ghz, 32gb ram:

- generated a bunch of copies of one 600mb mxi-file (so i had around 15gb-20gb of data-volume to have time to read the i/o-rate) and copied this bunch of files:
i/o at it's peak: ~400mb/s
- opened the simball-scene in studio (latest v3-beta), activated multilight at cranked the resolution up to 5K*5K to get a fairly big file (~600mb while rendering, ~1.5gb when being written finally, after stopping the render), wiated until SL4, stopped:
i/o at it's peak: ~100mb/s

no active swap file on ssd (using the 2nd drive (no ssd) for swap-file purposes).

it seems that the performance of your HD go down using maxwell.. i don't understand your last words: what mean:

no active swap file on ssd (using the 2nd drive (no ssd) for swap-file purposes)

please... explain me better.... what is swap file?

[mashium123]

...

it seems that the performance of your HD go down using maxwell.. i don't understand your last words: what mean:

no active swap file on ssd (using the 2nd drive (no ssd) for swap-file purposes)

please... explain me better.... what is swap file?

hm... yes.
i guess, there's quite a bunch of parameters that do influence the max. speed of data-processing on an ssd: how big a file is, how many files there are etc... I am in no way sufficiently educated - neither in maxwell nor in ssd-technology in general - to decide if a peak of 100mb/s (and that's what it was, just the peak - there were also lower values than that) is an unusual value in this context.
but it seems to me that the 2mb/s that your screenshot was showing is far from my experience from this one quick test.

sorry for confusing the terms.
with "swap-file" I was referring to eric nixon's "page file" comment, to make clear, that the i/o-values i posted are not disturbed by any possible pagefile read/write-actions, that could slow down the maxwell-write process somehow (which, i believe, was the hint he was trying to give concerning possible reasons for your low i/o-value of 2mb/s): there is no page file activated on the ssd drive i used for this test.

[walterfog]

...

it seems that the performance of your HD go down using maxwell.. i don't understand your last words: what mean:

no active swap file on ssd (using the 2nd drive (no ssd) for swap-file purposes)

please... explain me better.... what is swap file?

hm... yes.
i guess, there's quite a bunch of parameters that do influence the max. speed of data-processing on an ssd: how big a file is, how many files there are etc... I am in no way sufficiently educated - neither in maxwell nor in ssd-technology in general - to decide if a peak of 100mb/s (and that's what it was, just the peak - there were also lower values than that) is an unusual value in this context.
but it seems to me that the 2mb/s that your screenshot was showing is far from my experience from this one quick test.

sorry for confusing the terms.
with "swap-file" I was referring to eric nixon's "page file" comment, to make clear, that the i/o-values i posted are not disturbed by any possible pagefile read/write-actions, that could slow down the maxwell-write process somehow (which, i believe, was the hint he was trying to give concerning possible reasons for your low i/o-value of 2mb/s): there is no page file activated on the ssd drive i used for this test.

thanks for your reply.... another test that you could do, more simply, it is to clock manually the time that maxwell take to write or update the MXI file on the disc... if you could, try to make a scene of about 5680 x 4016 resolution, with multi light active and 4 or 5 emitter in the scene.. this could take a MXI of about 3,8 giga.... it is very interesting to see how many time maxwell take to update or write the MXI in your SSD..
in my SSD it take about 7 minute the MXI update / writing .....

[mashium123]

...
thanks for your reply.... another test that you could do, more simply, it is to clock manually the time that maxwell take to write or update the MXI file on the disc... if you could, try to make a scene of about 5680 x 4016 resolution, with multi light active and 4 or 5 emitter in the scene.. this could take a MXI of about 3,8 giga.... it is very interesting to see how many time maxwell take to update or write the MXI in your SSD..
in my SSD it take about 7 minute the MXI update / writing .....

well, i did not really take the time manually but: around 20 sec. should have been it. not very much longer.

but the test i did is basically quite similar to the setup you suggested.
as written, i took the simball-scene, activated multilight in "intensity"-mode (don't remember how many ml-sliders were active in that scene - guess 3-5), pushed the resolution up to 5000px*5000px (which is slightly bigger than your suggestion) - the final file was ~1.6GB. that's smaller than your suggested file size but i doubt, that this makes that much of a difference as 7min and ~20 sec does.

given that the simball-scene is part of the installation, it's something everyone, read: you, could try too - this way comparability would be provided at it's max.

[dariolanza]

Hello Walter,

In order to know if there is any disk swapping process (that would make the disk response too slow) we would need to know how many Ram memory does your system have and how many is in use during the MXI writing.
Could you send me an screenshot of your Task Manager window showing the memory usage and all processes in use during the slow writing to disk?

Let us know.

Dario Lanza

[walterfog]

Hello Walter,

In order to know if there is any disk swapping process (that would make the disk response too slow) we would need to know how many Ram memory does your system have and how many is in use during the MXI writing.
Could you send me an screenshot of your Task Manager window showing the memory usage and all processes in use during the slow writing to disk?

Let us know.

Dario Lanza

just net you an email with the two screenshot requested....
let me know please
thanks
walter