原帖由 tyranid 於 2009-6-26 11:49 PM 發表
So slow for GMonster
it only does 20MB/10MB(Read/Write@4K)

And look like 1000MB/s is the cause by cache on RAID controller/SSD, the steady statue is only 500MB/s

閃卡(SSD)速度慢是意料之內,但是PhotoFast的設計可以把原本是慢的SSD配合cache on RAID controller而提昇整體實質速度至1,000MB/s,值得一讚。
反觀Acard,把原本理應很快的DDR2-RAM降速,就是浪費!

TOP

原帖由 fastloopattack 於 2009-6-27 00:29 發表

閃卡(SSD)速度慢是意料之內,但是PhotoFast的設計可以把原本是慢的SSD配合cache on RAID controller而提昇整體實質速度至1,000MB/s,值得一讚。
反觀Acard,把原本理應很快的DDR2-RAM降速,就是浪費!

你都好夜下喎

TOP

原帖由 fastloopattack 於 2009-6-27 00:29 發表

閃卡(SSD)速度慢是意料之內,但是PhotoFast的設計可以把原本是慢的SSD配合cache on RAID controller而提昇整體實質速度至1,000MB/s,值得一讚。
反觀Acard,把原本理應很快的DDR2-RAM降速,就是浪費!

SSD is not slow,
But what i mean is Gmonster PCI-E SSD is just as fast as a single Barefoot@4K read/write.
and consider that you are unlikely to do use Wiper.exe for TRIM purpose on RAID,
Which mean it will getting slower and slower by time.
you might end up having 20MB/s write on each SSD some time later.....

imo, Intel SSD is still one of the best mainstream ssd on the market(forget about DDR x1/iodrive etc....)

TOP

原帖由 tyranid 於 2009-6-27 12:52 AM 發表

SSD is not slow,
But what i mean is Gmonster PCI-E SSD is just as fast as a single Barefoot@4K read/write.
and consider that you are unlikely to do use Wiper.exe for TRIM purpose on RAID,
Which mean  ...

Just curious, why you always concentrate on the 4k read/write result? How about the other results? Do they meaningless? For example, in CrystalDiskMark, the sequential read is 1,061MB/s, 512k is 441MB/s, 4k is 43.41MB/s, what does it mean? If you think that the result of 4k is unsatisfactory, then how about 512k which should be larger block of data. If 512k is faster than 4k, what does it mean?

TOP

原帖由 fastloopattack 於 2009-6-27 02:09 發表

Just curious, why you always concentrate on the 4k read/write result? How about the other results? Do they meaningless? For example, in CrystalDiskMark, the sequential read is 1,061MB/s, 512k is 441M ...

Because mult-threading 4k random read/write is a important figure of launching/installing programme . In days to days operation, it is like 80% in random and 20% in sequential.

Unless copying large file, you rarely can reach the stated sequential read/write.

TOP

原帖由 tyranid 於 2009-6-27 03:49 發表

Because mult-threading 4k random read/write is a important figure of launching/installing programme . In days to days operation, it is like 80% in random and 20% in sequential.

Unless copying large  ...

很期望看到CHING既裝置, 可否分享下!

[ 本帖最後由 chiushing01 於 2009-6-27 11:38 編輯 ]

TOP

原帖由 chiushing01 於 2009-6-27 07:35 發表

很期望看到CHING既裝置, 可否分享下!

I am just using a single x25-m,
unless file copying take place, the speed of x25-m is comparable to x25-e in most of the aspect(like seq read/ 4k random read/write)

TOP

原帖由 tyranid 於 2009-6-27 03:49 AM 發表

Because mult-threading 4k random read/write is a important figure of launching/installing programme . In days to days operation, it is like 80% in random and 20% in sequential.

Unless copying large  ...

但是4k的result是要小心地interpret。數值低,並不一定代表速度慢,極有可能是4k檔案太過細,用不盡頻寬,所以不需要很高的transfer rate就已經傳送完畢。理據是,我execute 4k檔案或programme的速度極高,不需要一秒已經完成,真是快如閃電。

TOP

原帖由 fastloopattack 於 2009-6-28 01:26 發表

但是4k的result是要小心地interpret。數值低,並不一定代表速度慢,極有可能是4k檔案太過細,用不盡頻寬,所以不需要很高的transfer rate就已經傳送完畢。理據是,我execute 4k檔案或programme的速度極高,不需要一秒已經完成,真是快如閃電。

in windows 7, the speed of 4k random read/write is the value to determine the overall system score, only storages with over 8MB/s at 4k random read/write are rated over 5.9+

it is what Microsoft telling you why 4k is always important in windows

TOP

原帖由 tyranid 於 2009-6-28 02:05 AM 發表

in windows 7, the speed of 4k random read/write is the value to determine the overall system score, only storages with over 8MB/s at 4k random read/write are rated over 5.9+

it is what Microsoft tel ...

Microsoft D 野你都好信ge?

TOP