1. ## Re: System->Admin->Disk Utility->Benchmark results

Originally Posted by Onesimus
(i.e at 50% of the time)
But it isn't 50% of the time - it's 50% of the disk capacity.

2. Spilled the Beans
Join Date
Nov 2007
Beans
16

My results using a 30 GB vertex SSD
min: 122.0 MB/s
max: 249.7 MB/s
avg: 197.5 MB/s
avg access: 0.5 ms

3. Tea Glorious Tea!
Join Date
Apr 2007
Beans
307
Distro
Ubuntu 12.04 Precise Pangolin

Originally Posted by andrewabc
I didn't notice any major problems. The average would be average for the whole test, not just an average between min/max. Although a 200mb datapoint even though nothing else is above 100mb/s does throw off avg a bit. The way to solve this would be to use statistical analysis to eliminate irregular occurrences (forget proper term) from calculating the averages.
What I am suggesting is a statistical analysis. There are three ways to calculate the average. The mean, the median, and the mode. The average that I am suggesting should be determined is the median (i.e. the 50 percentile - it occurs for 50% of the time). This would be a better average because it wouldn't be affected by large data points that don't occur very often. Whereas the mean average would be affected.

Originally Posted by MacUntu
But it isn't 50% of the time - it's 50% of the disk capacity.
Then if that is the case, how does it determine values for 0% of the disk capacity - does it empty the disk ? Or for 100% does it fill it ?

Until the data is clearly presented and understood how they have been dervied, then there seems little point in posting values on the forum, because it is not entirely clear what the data values actually mean.

4. ## Re: System->Admin->Disk Utility->Benchmark results

Originally Posted by Onesimus
W
Then if that is the case, how does it determine values for 0% of the disk capacity - does it empty the disk ? Or for 100% does it fill it ?
.
You dont understand. A regular harddisk has different performance characteristics at the inside tracks as on the outside. The outside turns faster (well, the speed of the platter is higher relative to the head) -> higher performance. Nothing to do with how full it is, the graph shows performance from the outside to the inside of the hdd platter.

5. Tea Glorious Tea!
Join Date
Apr 2007
Beans
307
Distro
Ubuntu 12.04 Precise Pangolin

Originally Posted by P4man
You dont understand. A regular harddisk has different performance characteristics at the inside tracks as on the outside. The outside turns faster (well, the speed of the platter is higher relative to the head) -> higher performance. Nothing to do with how full it is, the graph shows performance from the outside to the inside of the hdd platter.
Then wouldn't it be helpful to everyone, if that was stated somewhere. I wouldn't call myself IT illiterate having been a software developer/analyst for 20 years, but I do wonder if developers have gone to the trouble of developing software (perhaps over several months) then would it be too much for them to go that extra mile and spend an hour writing a paragraph or two describing their efforts so those that aren't in the know might appreciate them even more ?

If performance really does improve towards the outer sections of the disk (i.e approaching 100%), shouldn't people be concerned if the performance decreases. In general, most of the results posted here appear to indicate that performance of the read-rate reduces with increasing disk capacity.

6. ## Re: System->Admin->Disk Utility->Benchmark results

Originally Posted by Onesimus
Then wouldn't it be helpful to everyone, if that was stated somewhere. I wouldn't call myself IT illiterate having been a software developer/analyst for 20 years, but I do wonder if developers have gone to the trouble of developing software (perhaps over several months) then would it be too much for them to go that extra mile and spend an hour writing a paragraph or two describing their efforts so those that aren't in the know might appreciate them even more ?
Generally those concerned about the performance of their drives and running benchmarks I think will be aware of this. I also actually find it pretty amazing they included such a nice benchmark here in what is just a disk utility, not a sisoft sandra benchmarking suite, its not like you will find anything like it on windows or os-x by default. Some people are never satisfied

But anyway, its open source, so feel free to contribute!

If performance really does improve towards the outer sections of the disk (i.e approaching 100%), shouldn't people be concerned if the performance decreases.
Im not sure I understand what you are saying. The graph represents
0% as the outside tracks (which is the beginning) of the hdd where performance (read/write throughput at least) is highest. This is completely normal and by design. Some people partition their drives accordingly, creating "fast" partitions at the beginning of the drive.

In general, most of the results posted here appear to indicate that performance of the read-rate reduces with increasing disk capacity.
?
No, not capacity. It decreases near the inside of the drive, but that has no direct relationship with a full or empty drive. It depends entirely where your partitions are and how your OS decides to fill them. You can tune this if you really want, and to some extent OSs are clever and try putting commonly used files at the beginning (outside), but its really not worth bothering with for most and avoiding fragmentation is a much bigger issue than this.
Last edited by P4man; April 29th, 2010 at 09:30 AM.

7. ## Re: System->Admin->Disk Utility->Benchmark results

I'm sorry for throwing in the word "capacity".

Originally Posted by Onesimus
Then wouldn't it be helpful to everyone, if that was stated somewhere.
Definitely.

8. Tea Glorious Tea!
Join Date
Apr 2007
Beans
307
Distro
Ubuntu 12.04 Precise Pangolin

@P4man

Thank you for taking the time to explain things patiently to me. I am impressed with the ever-growing functionality of the applications that are provided as standard with the default Ubuntu installation.

I suppose I just feel that if more explanation was provided it would be helpful to everyone.

Thanks once again

9. Just Give Me the Beans!
Join Date
Jul 2006
Location
The Beach. NC
Beans
66
Distro
Ubuntu 10.04 Lucid Lynx

Here are the results from my SSD Raid0

I am running 10.04 RC with an Intel S5000XVNSATAR mainboard, Dual L5310 Quad Core Xeons over clocked to 2.0Ghz (1333 bus), Thermalright True Black 120 coolers with Scythe 120mm fans, 8G (8x1G) FBDIMMS, OCZ 550W PSU, Yate Loon case and ram fans, Nvidia Quadro FX540 PCIe, Dual Intel X25v 40G SSDs in Raid (R1 /boot, R0 swap and /), Chieftec mid tower EAXT case, Rosewill PCI wireless card.

10. Just Give Me the Beans!
Join Date
Jul 2006
Location
The Beach. NC
Beans
66
Distro
Ubuntu 10.04 Lucid Lynx

Here are the SSDs by themselves.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•