Continuing with our SSD. Since December, we write loop on an SSD-based controller Indilinx checking it for wear using the information reported by SMART. Last week, just after the cell the more worn - without elaborating - spent 10 000 cycles (about 10 500 at that time), we restarted the computer test and we realized that errors occurred on the DSS.
This week, after a few more tests, we found that decreased its ability. The reason is quite simple: after writing each day - about 5 TB per day - we conducted an audit of the device with the Windows tool. And every time we got errors on the SSD. This morning, for example, the utility audit has found 31 168 bad clusters.
And so - when performing a formatting - the bad clusters are marked and are no longer used. Specifically, the DSS provided a capacity of 127 823 507 456 bytes at the beginning of our tests. Currently, the capacity is 127 689 289 728 bytes, a difference of 134,217,728 bytes (128 million exactly happy coincidence).
For the rest, the average wear of the DSS is 8953 cycles, the cell is the most worn 10 778 cycles and less waste to 1012 cycles. Since December 20, we wrote about 625 TB on the SSD with 5 TB per day on average. Remember that manufacturers indicate that a typical user written between 5 and 10 GB per day on an SSD.
Note, for those interested, the wear reported by SMART information is consistent: in theory we should write 128 GB to increase the average wear of one cycle and our measurements indicate between 100 and 110 GB per cycle in practice. Small point to consider if you use the software Indilinx Status: value "health" shown is actually a calculation based on the theoretical strength (10 000 cycles) and the average wear.
It is therefore normal that the meter "start from zero after the limit of 10 000 cycles reached. Finally, users Crucial M225 should not panic if this value is low for some unknown reason, the latest firmware (2030) of the SSD model is based on a design life of 5000 cycles instead of 10 000 regular cycles.
In view of the errors of our SSD, appeared before the fateful limit, I am inclined to think that life is a trifle exaggerated, even if it is not a problem in practice. 
This week, after a few more tests, we found that decreased its ability. The reason is quite simple: after writing each day - about 5 TB per day - we conducted an audit of the device with the Windows tool. And every time we got errors on the SSD. This morning, for example, the utility audit has found 31 168 bad clusters.
And so - when performing a formatting - the bad clusters are marked and are no longer used. Specifically, the DSS provided a capacity of 127 823 507 456 bytes at the beginning of our tests. Currently, the capacity is 127 689 289 728 bytes, a difference of 134,217,728 bytes (128 million exactly happy coincidence).
For the rest, the average wear of the DSS is 8953 cycles, the cell is the most worn 10 778 cycles and less waste to 1012 cycles. Since December 20, we wrote about 625 TB on the SSD with 5 TB per day on average. Remember that manufacturers indicate that a typical user written between 5 and 10 GB per day on an SSD.
Note, for those interested, the wear reported by SMART information is consistent: in theory we should write 128 GB to increase the average wear of one cycle and our measurements indicate between 100 and 110 GB per cycle in practice. Small point to consider if you use the software Indilinx Status: value "health" shown is actually a calculation based on the theoretical strength (10 000 cycles) and the average wear.
It is therefore normal that the meter "start from zero after the limit of 10 000 cycles reached. Finally, users Crucial M225 should not panic if this value is low for some unknown reason, the latest firmware (2030) of the SSD model is based on a design life of 5000 cycles instead of 10 000 regular cycles.
In view of the errors of our SSD, appeared before the fateful limit, I am inclined to think that life is a trifle exaggerated, even if it is not a problem in practice.
- Intel Said Set to Ship Five New SSDs, Hybrid Drive (28/04/2011)
- Intel SSD 720, 710 and 520 Series leak out, Larsen Creek and Paint Creek bring up rear guard (28/04/2011)
- Intel's Larsen Creek SSDs leak out, courtesy of ASRock's Z68 motherboard (28/04/2011)
- Is it really worth buying a cheap SSD (via The world of Jamie I) (18/04/2011)
- Viking Modular plugs flash chips into memory sockets - The Register (28/04/2011)
No comments:
Post a Comment