Re: Fw: dd oflag=direct performance

[Michael J. Baars]

Hi Leslie,

It's integrated MMC memory storage. The size of the diskfile is now at 64 
blocks of 1048576 bytes (see the makefile), that should have made dd run at 
5663.7168 mb/s. Instead it is running at approximately 24.5 mb/s.

Internally we read 1 gb and we write 1 gb. The file is therefore copied exactly 
16 times from source to destination. That will get our measurement pretty 
accurate.

Your are talking about fstrim, while you should be talking about the 
fragmentation of the file instead, I suppose. The performance of the test is 
indeed affected by fragmentation of the file. In theory the test can run at 
even higher speeds, you are right about that.

Regards,
Mischa.

On Thu, 2020-11-12 at 20:05 +0000, Leslie S Satenstein wrote:
> 
> Hi Mike
> 
> You did not indicate the size of the disk file. Was it an m.2 NVME drive or 
> an SSD.
> I did not see in the code where you ran a fstrim to reset the NVME or SSD 
> before the next run.
> Those are small things that can change the results of your benchmarks. 
> 
> Thank you for your presentations.
> 
> 
> Regards  
>  Leslie
> 
>  Leslie Satenstein
> Montréal Québec, Canada
> 
> 
>         
> 
> 
>         
>         
> 
>             
>                 ----- Forwarded Message -----
>                 From: Michael J. Baars 
> <mjbaars1977.coreutils@cyberfiber.eu>To: "coreutils@gnu.org" 
> <coreutils@gnu.org>Sent: Thursday, November 12, 2020,
> 8:49:16 a.m. ESTSubject: dd oflag=direct performance
> 
>                 Hi,
> 
> I needed to zero out my hard drive because one of my nodes has become 
> unstable. To this purpose I used coreutils dd with the following command line 
> arguments
> 
> dd if=/dev/zero of=/dev/mmc... status=progress
> 
> and I noticed how slow this program is in doing the job. So I tried a couple 
> of different settings, like
> 
> bs=1048576 oflag=direct
> 
> but without significant improvement. The results are always the same... 
> around 25 mb/s.
> 
> Then I remembered this little benchmark I write not so long ago, please do 
> have a look at it, it won't destroy your drive.
> 
> I included the results obtained by running the benchmark on the computer I'm 
> currently working on, so you can compare them to your own.
> 
> Hope that one of you feels inspired enough you to pick up the dd source codes 
> and finish the job.
> 
> Best regards,
> Mischa Baars.
> 
>             
>