MP 7,1 4 VS 6 channels RAM in Real-world Lightroom TEST! It is 39.2% FASTER?! [UPDATE RAMDISK TEST]

[zhpenn]

I know when I upgrade from 4 channels to 6 channels will have about 10% Geekbench scores increase!
BUT will it make a difference in real life? In short YES, and HUGE!

Geekbench 5 scores:

CPU:
28-Cores 3275M

RAM:
4x32g + 4x8g = 17638
4x32g ONLY = 19360 (increase by 9.76%)
6x32g ONLY = 20979 (increase by 8.36% more)

Before adding 2 extra 32G DIMM to make it from 4 Channels to 6 Channels.
I have done this small test, but this really surprise me!

Export 309 x Canon 5D Mark4 RAW files to 8 bit TIFF using Latest Lightroom Classic CC.
Each test I export twice and take the average.

4x32g RAM
It takes 6:34 = 394s

add 2 extra 32G DIMM

6x32g RAM

It takes 4:43 = 283s


??????????????? WHAT ????????????????

I can not believe my eye!

It is 39.2% FASTER?! Such a HUGE difference!

That means, all the export-time tests I have done before when compared to Mac Pro 6,1 will have even a lot bigger gaps between them.

2019 Mac Pro 7,1 Lightroom and Photoshop configuration and Test [UPDATE: Radeon VII, 2013 MAC PRO TEST]

I use Lightroom a lot, almost every day for editing and export 200-3000 RAW images. I will also use Photoshop. I have just got MacPro 7,1 base model with 1TB SSD, then upgrade 28 Cores 3275M CPU and 128GB RAM(will have 192GB later). Old Mac MacPro Late 2013 2.7GHZ 12 Cores E5 64GB RAM AMD...

forums.macrumors.com

UPDATE:
I REDO THE SAME TEST WITH MONITORING SCREENSHOT
ALSO, COMARE TO MacPro 6,1 doing the SAME TASK
Trash Can Mac Pro 6,1 (12Cores, 64GB RAM)

MP7,1 - RAW TO TIF 4:46 = 286s (2.72x faster MP6,1)

MP6,1 - RAW TO TIF 13:00 = 780s

MP7,1 - RAW TO JPG 1:56 = 116s (2.7 x faster MP6,1)

MP6,1 - RAW TO JPG 5:14 = 314s

When Compare Mac Pro 2019 7,1(28 Cores) to Trash Can MacPro 6,1(12 Cores)

All tasks have 2.7 times faster performance. with 6 channel RAM on MAC PRO 7,1

BUT, have just 2 times faster performance. with 4 channel RAM on MAC PRO 7,1

39.2% performance increase from 4x32g RAM to 6x32g RAM on MAC PRO 7,1 (Base on the iStat Menu APP and No Swap Memory, the RAM should not fully filled even on 4x32g)

But is the increase of the performance, is it only because the 6 channel or LR actually make use of the extra 64GB of RAM is a question.
I will try to limit the Ram to 128GB but in 6 channel and try again to make it more clear.


UPDATE on Fab 2, 2020,
To further experiment of whether is just benefits from the extra 2 channel or it also benefits from an extra 64GB of RAM.
I have done the following test.
I have use this command line to Create a 64GB of RAMDISK,

diskutil erasevolume HFS+ “RamDisk64GB” $(hdiutil attach -nomount ram://131072000)

And Fill it up with about 50GB of files
To be more "TOUGH" to the new 6 Chanel of the test.
I create another 64GB of RAMDISK II
And fill it up with a similar 50GB of file. which "takes 100GB of RAM away"

Then the same test of the RAW to TIF export AGAIN

I can feel a little bit laggy using the mouse than before during exporting.

BUT, the result is almost the same!

4x32g RAM (with 128GB RAM available)
It takes 6:34 = 394s

6x32g RAM (with 192GB RAM avalible)
It takes 4:43 = 283s

6x32g RAM (with about just 92GB RAM available)
It takes 4:49 = 290s

It just 7 seconds more than before, and with
Even less RAM than the 4x32GB, THE 6 CHANEL STILL HAS a 39% percent performance increase.

And BTW, I have also given the RAMDISK a speed test

4700MB/S Read
4550MB/S Write

I know the test may not very professional, but it really reflects some real-world performance increase.


Another Update:
Just Saw this thread there, there a memory bandwidth document shows different DIMMs setup differences

Best RAM configuration for the 2019 Mac Pro

Yes Definitely... 4xDimms is actually only 66% Performance ( bit more in reality ) you have 6 channels of memory access and you are Currently using 4. it's like 95% performance with 6 sticks and 100% with all 12 ( so long as they are all the same ) From page 18...

forums.macrumors.com

From page 18

https://lenovopress.com/lp1089.pdf

(actually that whole section explained it all... balanced and near balanced etc)

6channels 95% performance
4channels 66% performance ( bit more in reality )
That is a 43.9% increase, which is similar to the test results.

Another Update:
There is a similar test, which has a similar conclusion

Mac Pro review - A photographer and Videographers' perspective

I am going to update my post but here is a snippet with RAM change to a Single capacity modules as opposed to mixed. As confirmed in other threads, there is a massive boost.

forums.macrumors.com

ALL IN ALL, MAKE SURE YOU INSTALL 6 or 12 DIMMs with the same size on MAC PRO 7,1
and DO NOT MIX DIMMs size! (eg. 4x64gb + 2x8gb etc.)

 

[TTD]

Thanks for posting the test results.

I currently have 4 x 32gm installed and want to add 2 x 32gb more. Can I check what slots you added them into?

And do you think increasing to a total of 8 x 32gb ram would cause a reduction in speed compared to 6?

 

[defjam]

Are you sure your comparison is not measuring the additional 64GB of memory that the 6 x 32GB configuration provides? I have a difficult time believing that memory bandwidth would have such an increase on performance. Especially a task which is disk based. I suspect you're seeing the effects of caching and not memory bandwidth. Any chance you can run the test with equal amounts of memory?

 

[zhpenn]

Thanks for posting the test results.

I currently have 4 x 32gm installed and want to add 2 x 32gb more. Can I check what slots you added them into?

And do you think increasing to a total of 8 x 32gb ram would cause a reduction in speed compared to 6?

as to slots, I follow the sticker on the back of the DIMM Cover by apple.

I think 8 x 32 will reduce speed.

Just 6 or 12 DIMMs will be the best option. I did not expect such a big difference until I test it.
[automerge]1580313107[/automerge]

Are you sure your comparison is not measuring the additional 64GB of memory that the 6 x 32GB configuration provides? I have a difficult time believing that memory bandwidth would have such an increase on performance. Especially a task which is disk based. I suspect you're seeing the effects of caching and not memory bandwidth. Any chance you can run the test with equal amounts of memory?

I think this might possibly be one of the reasons, but exporting files should rely more on CPU.
but because I don't have RAM config to test 6 channels with equal amounts of memory.

 

[defjam]

I think this might possibly be one of the reasons, but exporting files should rely more on CPU.
but because I don't have RAM config to test 6 channels with equal amounts of memory.

RAM and CPU are significantly faster than even the fastest SSDs. I assume your export involves:

• A read from the SSD of the RAW file
• Conversion to TIFF by the CPU
• A write to the SSD of the TIFF file

The write to the SSD is, IMO, the slowest operation in that process. Significantly more so than whatever bandwidth the increase in memory bandwidth achieves. A 32% increase in real world application work seems too good to be true. If it's not caching which is in play then I may suggest that maybe the 4 channel configuration is really poor, which I would have a difficult time believing.

How many cores does Lightroom use while performing this test?

 

[zhpenn]

RAM and CPU are significantly faster than even the fastest SSDs. I assume your export involves:

• A read from the SSD of the RAW file
• Conversion to TIFF by the CPU
• A write to the SSD of the TIFF file

The write to the SSD is, IMO, the slowest operation in that process. Significantly more so than whatever bandwidth the increase in memory bandwidth achieves. A 32% increase in real world application work seems too good to be true. If it's not caching which is in play then I may suggest that maybe the 4 channel configuration is really poor, which I would have a difficult time believing.

How many cores does Lightroom use while performing this test?

28cores fully used

 

[fhturner]

Are you sure your comparison is not measuring the additional 64GB of memory that the 6 x 32GB configuration provides? I have a difficult time believing that memory bandwidth would have such an increase on performance. Especially a task which is disk based. I suspect you're seeing the effects of caching and not memory bandwidth. Any chance you can run the test with equal amounts of memory?

Yes, definitely this. You need to observe memory usage in Activity Monitor and make sure the operation is not getting into virtual memory paging, which will start to significantly slow you down. Hard to believe that 128GB would be a bottleneck, but I'd definitely make sure you're not exhausting it w/ that much while not w/ 192GB.

 

[libertyranger10]

Wow, some impressive results! do you find this across the board with multiple tests?

 

[zhpenn]

Wow, some impressive results! do you find this across the board with multiple tests?

I only dis test with this task only. because i did not expect any difference?
[automerge]1580345051[/automerge]

Yes, definitely this. You need to observe memory usage in Activity Monitor and make sure the operation is not getting into virtual memory paging, which will start to significantly slow you down. Hard to believe that 128GB would be a bottleneck, but I'd definitely make sure you're not exhausting it w/ that much while not w/ 192GB.

I will try using ram disks to take the extra 64 gb ram out and do test again

 

[AidenShaw]

I only dis test with this task only. because i did not expect any difference?
[automerge]1580345051[/automerge]

I will try using ram disks to take the extra 64 gb ram out and do test again

Most systems have boot flags to disable some RAM.

Most also have ways to flush filesystem caches. In Linux, the (root) command:

# sync; echo 1 > /proc/sys/vm/drop_caches​

will flush the caches. It won't disable them, but at least you get a clean start between runs.

 

[thexash]

For what is worth it mirrors my experience, even 6x8 + 6x32 is significantly slower than just 6x32. Imo dont mix sizes and use 6 or 12 dimms only...

 

[zhpenn]

More test done
When Compare Mac Pro 2019 7,1 to Trash Can MacPro 6,1
All tasks have 2.7 times faster performance. with 6 channel RAM on MAC PRO 7,1
BUT have just 2 times faster performance. with 4 channel RAM on MAC PRO 7,1

MAKE SURE YOU INSTALL 6 DIMM with the same size on MAC PRO 7,1
and DO NOT MIX DIMMs size! (eg. 4x64gb + 2x8gb etc.)

 

[defjam]

More test done
When Compare Mac Pro 2019 7,1 to Trash Can MacPro 6,1
All tasks have 2.7 times faster performance. with 6 channel RAM on MAC PRO 7,1
BUT have just 2 times faster performance. with 4 channel RAM on MAC PRO 7,1

MAKE SURE YOU INSTALL 6 DIMM with the same size on MAC PRO 7,1
and DO NOT MIX DIMMs size! (eg. 4x64gb + 2x8gb etc.)

I'm sorry but I cannot consider these tests valid when comparing four versus six channel memory configurations. You have not held everything but the memory channel configuration constant. Specifically the six channel memory configuration provides 64GB more memory than the four channel configuration. You really need to compare the same memory capacity with the two different memory channel configurations before you can state the six channel memory configuration is x% faster in real world tasks.

 

[Snow Tiger]

Yes, definitely this. You need to observe memory usage in Activity Monitor and make sure the operation is not getting into virtual memory paging, which will start to significantly slow you down. Hard to believe that 128GB would be a bottleneck, but I'd definitely make sure you're not exhausting it w/ that much while not w/ 192GB.

He has activity monitor screens open for all tests on both systems . There is no swapping ( paging ) going on . No virtual memory is being used at all anywhere .

 

[defjam]

He has activity monitor screens open for all tests on both systems . There is no swapping ( paging ) going on . No virtual memory is being used at all anywhere .

Why would you expect increased paging when you have more memory available?

 

[zhpenn]

I'm sorry but I cannot consider these tests valid when comparing four versus six channel memory configurations. You have not held everything but the memory channel configuration constant. Specifically the six channel memory configuration provides 64GB more memory than the four channel configuration. You really need to compare the same memory capacity with the two different memory channel configurations before you can state the six channel memory configuration is x% faster in real world tasks.

is there any good ramdisk app or one-time command line that I can take out 64GB of RAM to use as RAM disk that works on Catalina? I found an app that only works on Lion.

 

[Snow Tiger]

Why would you expect increased paging when you have more memory available?

never said I did ; I was just replying to another poster's concern about the OP possibly utilizing virtual memory in his tests . Which , he isn't .

 

[defjam]

is there any good ramdisk app or one-time command line that I can take out 64GB of RAM to use as RAM disk that works on Catalina? I found an app that only works on Lion.

I'm not aware of any but I have never investigated either one. If there was a means to disable some memory we'd need to know how it is disabling memory (i.e. per module, across all modules, etc.). Perhaps you can view cached files and disk I/O in Activity Monitor to see if that gives you an idea of what is occurring. Even then I'm not sure that would tell us the whole story.
[automerge]1580392092[/automerge]

never said I did ; I was just replying to another poster's concern about the OP possibly utilizing virtual memory in his tests . Which , he isn't .

Ah, my apologies.

 

[zhpenn]

I'm not aware of any but I have never investigated either one. If there was a means to disable some memory we'd need to know how it is disabling memory (i.e. per module, across all modules, etc.). Perhaps you can view cached files and disk I/O in Activity Monitor to see if that gives you an idea of what is occurring. Even then I'm not sure that would tell us the whole story.
[automerge]1580392092[/automerge]

hare are more screenshots when doing exporting tif

 

[Coyote2006]

would be interesting to see the speeds on the 8-core using 4, 8 and 12 dimms

 

[AidenShaw]

is there any good ramdisk app or one-time command line that I can take out 64GB of RAM to use as RAM disk that works on Catalina? I found an app that only works on Lion.

How to limit RAM in an OSX device?

Related to my other question, I'm about to buy a new Mac Mini. I would like to reduce RAM to 8GB in order to see if 8GB is enough to my daily tasks in my MBP retina 15 early 2013 16GB RAM. The ide...

apple.stackexchange.com

 

[defjam]

hare are more screenshots when doing exporting tif

I'm not sure what this information is telling me without any context. I don't know what each screen image represents. By this I mean what were the specific configurations for each image? What specific task was being performed? Etc.

 

[bxs]

is there any good ramdisk app or one-time command line that I can take out 64GB of RAM to use as RAM disk that works on Catalina? I found an app that only works on Lion.

Maybe use this command to adjust the amount of RAM the system can use....

Example... set RAM size to 8 GB
sudo nvram boot-args="maxmem=8192"
and then reboot

To reset RAM to what is physically available then....
sudo nvram -d boot-args
and then reboot

However, I'm unsure how setting "maxmem" to less than what is physically installed influences how much of each DIMM is used to accomplish this.

 

[high heaven]

I think the test is meaningless cause it's Mac Pro 2019 VS 2013, not 4 channel vs 6 channel. Mac Pro 2019 has 28 cores while Mac Pro 2013 has 12 cores? Seriously?

Do you have test results with Mac Pro 2019 by using 4 and 6 RAMs? Otherwise, it doesn't represent anything at all.

 

[zhpenn]

I think the test is meaningless cause it's Mac Pro 2019 VS 2013, not 4 channel vs 6 channel. Mac Pro 2019 has 28 cores while Mac Pro 2013 has 12 cores? Seriously?

Do you have test results with Mac Pro 2019 by using 4 and 6 RAMs? Otherwise, it doesn't represent anything at all.

Read the context, the last part is just an extra.

in the beginning, already saids on 7,1

4x32g RAM
It takes 6:34 = 394s

add 2 extra 32G DIMM

6x32g RAM
It takes 4:43 = 283s

It is 39.2% performance increase from 4x32g RAM to 6x32g RAM on MAC PRO 7,1 (Base on the iStat Menu APP and No Swap Memory, the RAM should not fully filled even on 4x32g)