標題: OCZ VERTEX 2 60G?? [打印本頁]

---

作者: hkalan    時間: 2011-3-28 12:46     標題: OCZ VERTEX 2 60G??

IS 0.25 OR 0.34 now?

---

作者: xc63774174    時間: 2011-3-28 13:06

提示: 作者被禁止或刪除 內容自動屏蔽

---

作者: Midas    時間: 2011-3-28 21:33

25nm 除咗 format 後少幾 M 之外，重有咩比 34nm 差？

---

作者: metal005    時間: 2011-3-28 22:42

提示: 作者被禁止或刪除 內容自動屏蔽

---

作者: dieago    時間: 2011-3-29 00:39

回復 4# Midas

startup time more... registry loading kills more time

---

作者: Midas    時間: 2011-3-29 00:42




    Do you think a firmware upgrade will solve the problems in the near future?

---

作者: dieago    時間: 2011-3-29 00:50

dont know... just asking their cs in this forum!

last time, failed to upgrade 1.3 to 1.6, then they agree +$ and exhange the 60g 2nd gen.

i want to start the pc faster, that's the main reason i prefer ssd... but it runs 6.5 bar now... ding

---

作者: Midas    時間: 2011-3-29 00:59



I also enjoy the fast bootup time with the SSD,  what do you mean by 6.5 bar? Sorry for my ignorance!

---

作者: dieago    時間: 2011-3-29 01:01

before the windows' logo appear, a bar is running

---

作者: Midas    時間: 2011-3-29 01:13




    I am using Win7 , no bars...

---

作者: sunnychan1020    時間: 2011-3-29 01:30




I just need run 1 bar only

---

作者: dieago    時間: 2011-3-29 01:35

it's very hard to define "how many bar", how fast a pc run!

my point is "compare to the past".

---

作者: Midas    時間: 2011-3-29 01:53

My Win7 runs 2 rings then full boot

---

作者: wilsonkf    時間: 2011-3-29 08:45

提示: 作者被禁止或刪除 內容自動屏蔽

---

作者: Midas    時間: 2011-3-29 12:38



From a business point of view, do you think it is something similar to the "Sandy Bridge affair"? or  Intel is trying to reduce cost boosting topline at the expense of the customer? Or are you saying 25nm flash chip has intrinsic hardware problem in it?

---

作者: Niel    時間: 2011-3-29 12:52



25nm NAND Flash has a greater density than its 3Xnm counterpart.

The first version of 25nm Vertex 2 60GB contains 8 64Gb density die (8x64Gb), which only has half of the channels of the original Vertex 2 60GB (34nm) using 32Gb die (16x32Gb). And because of that, the performance is severely affected.

According to OCZ forum, they are now shipping a new batch of Vertex 2 60GB SSD which uses 32Gb 25nm die and the problem caused by the reduction of available channels is now resolved.

---

作者: Midas    時間: 2011-3-29 13:07



Niel 師兄， good news ! Thanks for your info. How about the G3? I guess it's much to early for Intel to act on their onec again another issue

---

作者: Midas    時間: 2011-3-29 13:09

回復 17# Niel

From 16 X 32Gb  to  8 X 64Gb  is definately a cost cutting issue. No wander why...

---

作者: madox    時間: 2011-3-29 13:11


How to check it is 32Gb version?

---

作者: dieago    時間: 2011-3-29 13:14

回復 17# Niel

if so, will they do something and help the buyer who have 25nm?  it's not fair

---

作者: Niel    時間: 2011-3-29 13:16



You can check my post for Intel 320 info:
http://www.hkepc.com/forum/viewt ... &extra=page%3D1

But if you can afford a little premium (~$600) and are looking for the top of the line model, you should wait for Vertex 3. It simply dwarf all others.

---

作者: Niel    時間: 2011-3-29 13:19



OCZ is offering SSD replacement in US. But seems they are not going to expand to other non-US countries.

---

作者: Niel    時間: 2011-3-29 13:20



Check my post:
http://www.hkepc.com/forum/viewt ... 6087&highlight=

---

作者: madox    時間: 2011-3-29 13:29


Thanks..
不過要買左返黎先知,會唔會好易中招?

---

作者: Niel    時間: 2011-3-29 13:32



係60G先有依個問題，120G以上不受影響。 (90G我唔sure)

如果你等得可以等下Vertex 3 120G，到時再比較。

---

作者: Midas    時間: 2011-3-29 14:07




    Niel 師兄， 其實如果我唔打機！ 普普通通咁用， 最多睇下高清電視， 上下網咁！ 感唔感覺到兩者（25,32nm）嘅分別呢？ 我最中意就喺 SSD bootup, restart speed， 呢個應該喺 sequential ，應該唔會被 25 32 nm 問題音響，是嗎？ 呢個問題會唔會好難解答？

---

作者: big_nature    時間: 2011-3-29 14:18


師兄，
想問下如果一般的SSD使用，
只上下網、做下文書(以Excel為主)
每天使用8小時，
一隻SSD是否可用5年以上?
(講讀寫計，不計零件上的損壞)
想叫公司買隻畀我在公司用!

---

作者: Niel    時間: 2011-3-29 14:36



我記得你用Intel G2，你可以自己用Intel toolbox睇返host write，除返power on time，就可以簡單估計你每日用幾多GB host write。(通常最多host write係install windows/software的時候，之後個average 會下降)

以Intel 320 80GB為例 (3-5k 25nm NAND, 7% OP)，簡單估計下:

Intel話自己WA可以近1.1

Basic conversion:
80GB = 74.5058GiB
74.5058GiBx3k cycle (用下限計) = 218.2787TiB

OP (Over-provisioning):
218.2787TiB x (1+7%) = 233.558TiB

WA (Write Amplification):
233.558TiB/1.1 = 212.32TiB

Theoretical Host Write有212.32TiB (以3k write cycle計)

以讀寫計，不計零件上的損壞:
每日寫100GB，最少可以用6年
每日寫50GB，最少可以用12年
每日寫10GB，最少可以用60年
每日寫5GB，最少可以用120年
...

所以SSD真係幾難寫死...

P.S. 原來Intel無在25nm model上再加OP...同34nm一樣都係7%...好似小氣左d...

---

作者: Niel    時間: 2011-3-29 14:38



其實Vertex 2 60GB降速主要係channel問題，唔係3Xnm > 2Xnm問題。

Bootup, restart重視random performance多d。

---

作者: big_nature    時間: 2011-3-29 15:33

本帖最後由 big_nature 於 2011-3-29 15:41 編輯

回復 29# Niel

師兄非常詳細的數據，

隻SSD幾好用下!
用小小$得到開心係非常值得!

依家想說服公司買隻畀我工作用!

即係我隻120g G2 如果50G W/R計,
可以用大約12年!?

---

作者: Niel    時間: 2011-3-29 15:45



一日50GB係好heavy usage，office work通常都係5GB per day，你可以自己monitor下

---

作者: big_nature    時間: 2011-3-29 15:54

回復 32# Niel

唔係好識睇個Intel toolbox資料!
只係做過Check W/R  passed果項

師兄對SSD真係expert!

---

作者: Niel    時間: 2011-3-29 16:01



多謝贊賞

[attach]1158365[/attach]
將Host Writes除Power-On Hours court，就可以估計average

---

作者: big_nature    時間: 2011-3-29 16:34

本帖最後由 big_nature 於 2011-3-29 16:39 編輯

回復 34# Niel

明白晒啦師兄!
呢個test我都有做過，但以前不明所以!

如果我個120G 鎅左80G 同40G
(當然唔足數啦)
同唔鎅的話，壽命有冇差異?
技術研究而已!

---

作者: Niel    時間: 2011-3-29 16:43



除左個file system多左個Master File Table，多左少少overhead之外，我諗唔到有會對壽命有大影響。

---

作者: big_nature    時間: 2011-3-29 16:49

回復 36# Niel
如果個SSD用到5年，
我非常滿意了!
因為到時的儲存技術唔知發展到邊個地步!

---

作者: Niel    時間: 2011-3-29 16:54



無錯，SSD差不多1.5年快近一倍，5年後我諗大家都可能買新了

---

作者: Midas    時間: 2011-3-29 17:08




   多謝 Niel 師兄提點， 咁我都喺睇定啲先啦！ 橫掂都唔急！ 對我隻 Vertex 30G 都好滿意！

---

作者: computer1975    時間: 2011-3-29 19:25



keep more than 20% free space otherwise window will spend more time to search empty space to write a new data and data a less likely to write sequential to lower read speed

---

作者: Niel    時間: 2011-3-29 19:35



wrong....

As SSD adds a level of abstraction between the OS's file system and SSD's physical media, the OS has NO WAY to trace the exact location (address) of empty (or recycled) block and data location.

The block handling behavior is entirely done by the SSD controller (not the OS or the file system).

However, keeping a certain amount of free space does help to maintain the performance of the SSD.

---

作者: Midas    時間: 2011-3-29 21:04




    Niel 師兄， how much free space in % wise is good for healthy SSD performance? Mine is like below, I believe it's time to change!


[attach]1158474[/attach]

---

作者: big_nature    時間: 2011-3-29 21:08

本帖最後由 big_nature 於 2011-3-29 21:11 編輯

回復 40# computer1975

我個容量應該都有好多位剩!

---

作者: Midas    時間: 2011-3-29 21:12

今日想喺 Amazon 買一隻 G2 80G， 但喺 Amazon 唔寄香港！ 激死人！

---

作者: Midas    時間: 2011-3-29 21:15




    師兄你呢隻由按電製到入到 Win7 大概幾多秒？

---

作者: big_nature    時間: 2011-3-29 21:20

回復 41# Niel
照師兄咁講，
我鎅左C同D兩個Partition.
OS當然裝在C啦.
如果D只係放下照片同音樂!
而D好少使用(聽歌同睇相)的話!
咁controller 基本動作(W/R)都在C調動,
呢個理解有冇錯!?

---

作者: big_nature    時間: 2011-3-29 21:23

本帖最後由 big_nature 於 2011-3-29 21:49 編輯

回復 45# Midas
我部係note,
由出Thinkpad個logo到入windows 7 64 桌面,
25-30秒度!
power control set adaptive only

---

作者: Niel    時間: 2011-3-29 21:41



唔係，而係controller係無C/D等等idea，分唔分partition無分別。

---

作者: big_nature    時間: 2011-3-29 21:51

回復 48# Niel
即係controller 睇見邊度有空位(C or D) 就利用!
鎅左C/D唔係關鍵!

---

作者: big_nature    時間: 2011-3-29 23:08


咁我個表是否=15年?

---

作者: computer1975    時間: 2011-3-30 09:10

本帖最後由 computer1975 於 2011-3-30 09:43 編輯


I only had 4G SSD in my EeePC 701. I boot into Ubuntu 10.04 less than 40 second. 4G SSD have 700 ~ 800MB free space (18-21% free space). I will decide to try Debian 6.0.1 later to reclaim some free space to let SSD work more healthy.

---

作者: computer1975    時間: 2011-3-30 09:36

本帖最後由 computer1975 於 2011-3-30 09:41 編輯


I think you not known well design of filesystem and how filesystem interactive with SSD.
NTFS is designed for Magnetic HDD. Don't expect it is work very well for SSD.

Not enough free space (less than 20%) is more and more likely for OS to be not successful to write data to continuous free space (file fragment occur). In this situation, SSD will more likely to write more than one location.
It is very high change for SSD to do more erase block and then read performance will be suffered. At the conclusion, not enough free space contributes more wear tearing and deteriorates write and read performance.

---

作者: Niel    時間: 2011-3-30 12:08

本帖最後由 Niel 於 2011-3-30 12:17 編輯


Are you blind?



Everyone knows NTFS is not that suitable with SSD as other FS like exFAT.

If you read my reply CAREFULLY, I agree with you that keeping a certain amount of free space does help to maintain the performance of the SSD. And you have mention the reason behind it CORRECTLY.

However, you were wrong when you said


and


As aforementioned, SSD adds a level of abstraction between the OS's file system and SSD's physical media, the OS has NO WAY to trace the exact location (address) of empty (or recycled) block and data location (Unless you have access to engineering tools).

As a result, the file system will not be and does not need to be understood by the SSD controller. So it is not the Windows (OS) who spent more time to search for empty space, but the SSD controller itself.

---

作者: metal005    時間: 2011-3-30 12:45

提示: 作者被禁止或刪除 內容自動屏蔽

---

作者: Niel    時間: 2011-3-30 12:48



你隻SSD都好新wor haha，得10個鐘。

你用耐d再計啦，你一開始裝software + 行benchmark (如有) 先令到個host write咁高。

---

作者: big_nature    時間: 2011-3-30 14:38

回復 55# Niel

晚上只上下網!
睇下各師兄的評論，
做下公司成本計算!
真係用得好少時間!

個 host writes高嗎?
以為佢有15年壽命，
要換野都冇藉口!

---

作者: Niel    時間: 2011-3-30 14:40



你一開始裝software + 行benchmark (如有) 先令到個host write咁高。

之後正常用average write會drop

---

作者: big_nature    時間: 2011-3-30 21:02

本帖最後由 big_nature 於 2011-3-30 21:04 編輯

回復 57# Niel
裝過2次機!
只行過一次AS SSD benchmark!
就會咁高?
(15X G都係高?)

---

作者: Niel    時間: 2011-3-30 22:40



power on hours: 10
host writes: 155GB
Write per hour: 15.5GB...

不過，你一開始裝software + 行benchmark 先令到個host write咁高。

之後正常用average write會drop

---

作者: big_nature    時間: 2011-3-30 22:54

回復 59# Niel
原來條數係咁計!
以為=15年!

講真我每晚都用2-3個鐘度!
9號買到依家都20日應該有40-50個鐘!
點解會話我用得10個鐘?
我都可奇怪!
是否重裝機開始計!
或是!?
toolbox行過一次complete read scan!
唔知有冇關?

---

作者: dieago    時間: 2011-3-30 22:57

[attach]1158863[/attach][attach]1158864[/attach]
talk to their CS today, he ask me provide them the serial nbr for checking.

just check my ssd and post my benchmark for ching's review... can someone tell me that "bootup-> slower than vertex 1 30g... what's wrong now?"... "the running bar is now 7.5 before the windows' logo appear"... omg

hw+driver:
amd 770 chips, amd 3850 (10.2 driver), 620 u, 4g ram, hdx4, emu 0404, broadcom 5708,
sw:
just instal : vista 32-all patch update to date, (1) purecodec (2) itunes (3) foobar, (4) eset smart security 4

---

作者: loveyincc    時間: 2011-3-31 05:53



[attach]1159051[/attach]
我都係OCZ VERTEX 2 60G
同樓上位好似好大分別?

---

作者: tbpatrick    時間: 2011-3-31 13:30

好奇怪隻SSD用0左個幾月個POWER-ON HOURS COUNT都仲係停留0左0係"10"

---

作者: Midas    時間: 2011-3-31 15:56




    大哥， 我（Vertex 1 用家）喺用 Win7 64bit 啵！ 唔知會唔會有分別呢？  Vista 問題多多喺人都知道！ 不如你 DL 一隻 OEM Win7 試用一下睇下有冇分別重好啦！

---

作者: dieago    時間: 2011-3-31 20:04

回復 64# Midas

dont want to spend $ on w7, as run vista not more than 2 yrs...

i think the point is not the windows, is the ssd, when i use vertex 1, no such problem

---

作者: computer1975    時間: 2011-4-1 21:28



exFAT do not support OS boot drive
may be lack of journaling

---

作者: computer1975    時間: 2011-4-1 21:40



If you said SSD controller to be very intelligent, SSD is no need to co-operative with OS. Why a lot of Ching to do flash Alignment! I don't think today to SSD controller still can not handle them (include low free space of filesystem) perfectly so that why I recommend user to leave enough free space to keep SSD healthy.

I think I not tell to you but for another ching. I usually monitor disk fragmentation by analysis of Diskeeper without Inteli-write and Invisible task.  In enough free space, copy file from Drive D to Drive C. NTFS is very smart to avoid file fragmentation. But at low free space situtation, file fragmentation will be happened very easily.

---

作者: Niel    時間: 2011-4-1 21:59



我第一個reply就講左Keep Free Space係有助maintain performance



Free space下跌，速度同樣下跌同file system完全無關係，你要製造一隻有極高random write speed，而且在近全滿情況下不降速的SSD，唯一方法是加大over provisioning，而且controller block recycling要極快，同File System完全無關係。

你連基本concept都分不清楚，仲要誤導其他人SSD要defrag/prevent fragmentation，真係救命。之前我同你已經討論過:

http://www.hkepc.com/forum/viewt ... 0%E9%87%8D%E7%B5%84
   
NTFS is very smart to avoid file fragmentation. << 你無講錯，NTFS能防止HDD上的fragmentation，唔係SSD。SSD係喜歡製造fragmentation，不過如果fragment太細controller會自動作combination，整個過程File system都係不知情。

我重申一次，亦請各位網友留意，SSD是不需要重組的。

---

作者: Niel    時間: 2011-4-1 22:02

本帖最後由 Niel 於 2011-4-1 22:20 編輯


我之前係寫exFAT比NTFS更適合Flash Device，我沒有指exFAT可以boot機，最少現時未有OS support。

exFAT 唔 support boot drive係OS問題，唔係設計上問題。

FAT, FAT32都無journaling，咁大家用Windows 95/98/ME果時點開機?...

你去讀下基本concept先啦。

---

作者: computer1975    時間: 2011-4-1 22:48



Linux - ext3, jfs, XFS
BeOS - BeFS

以上全部都有journaling



無必要啦睇太多，ext2, jfs, ufs, ufs2, XFS, NTFS ... etc 好多filesystem whitepaper ＆ spec都睇過

---

作者: Niel    時間: 2011-4-1 22:53



睇多咁多都仲錯...

...

---

作者: computer1975    時間: 2011-4-2 07:54

本帖最後由 computer1975 於 2011-4-2 07:58 編輯


Your OS concept is very poor and do not known well how OS to work!

My standpoint free space fragment hurt the performance of SSD

I remember what myDefrag's autor said. It is the proof of file and space fragment to increase overhead of Windows during I/O operation of SSD.



MyDefrag 有 SSD defrag mode



source: http://www.mydefrag.com/FAQGener ... hSSDMemoryDisk.html

---

作者: Niel    時間: 2011-4-2 19:10

本帖最後由 Niel 於 2011-4-2 19:23 編輯


你所有concept用在Hard Drive身上都是對的，不過當你用Hard Drive的concept去了解SSD，所有野都錯曬，我一樣一樣咁解釋你聽:

你的句子:


我一開始就同意你的講法，不過同時我一開始也說了你的解釋不對。(結論對，解釋不對)可用空間下降引致降速是對的，不過降速不是由File System引致或可用defragment software解決，原因如下:

1. SSD physical layer被controller的FTL abstract左，File system中的logical address已經不是正確的physical address。(不同於HDD)
2. Defragmentation software由於經由File system access physical media，所以Defragmentation SW同樣不能正確運作(因為不能正確地得知physical address，引致不能正確地分析fragmentation)。
3. SSD為了增加速度，是會自行增加fragmentation，目的是使用multichannel的優勢。如果我地假設(留意係假設)有defragmentation software係work，defragmentation只會令SSD降速及增加write wearing，因為defragmentation會抵消了SSD multichannel的效果。幸好的是，現今市場上標榜可以defrag SSD的software都係唔work(原因1及2)，所以用左都唔會引致嚴重降速。

現時市場上標榜defrag SSD的software都是defrag file system (above FTL)，而不是defrag SSD上的實體data (below FTL)。現時只有TRIM command可以幫助SSD正確地進行block recycle。

我依排好忙，如果你都係唔明的話，我下星期寫一篇詳細分析比大家睇。

---

作者: computer1975    時間: 2011-4-2 22:15

本帖最後由 computer1975 於 2011-4-2 22:44 編輯


不認同 ... ...
1. File System下產生Free Space Fragmentation、File Fragmentation，降速相對上對於JMicro第一代會影響比較太
2. 另外別忘記SSD最麻煩是要erase block之後再能寫東西。Free Space Fragmentation、File Fragmentation長遠來說對SSD有影响，最近出trim技術麻煩少了不少。我所指overhead是Windows處理Free Space Fragmentation、File Fragmentation要額外附出CPU usage，不過當然可以俾錢解決，拿金錢買更快CPU去扺消呢方面I/O latency。
3. 只想純討論SSD沒有再必要爭落去，同時間都有睇下相關filesystem和SSD有乜關係對其他ching可以用，SSD和filesystem兩樣我都有興趣去睇睇。鑽研Filesystem都有10多年啦！



Diskeeper的HyperFast有針對Free Space Fragmentation作優化的軟件，不過只有benchmark，連一點技術上資料沒有公開過，那要等你或其它ching試下，trial版有一個月免費試用期。本身我只有4G SSD另外也不裝Windows所以無得試。



邊到可以找到資料？
知你不會用MyDefrag的SSD Mode去優化SSD，那就等其他ching去試吧。



認同，本身就是SSD原理

提外話問問ching，你現在用乜CPU、SSD？你係咪Microwork得“將”那位？

---

作者: Niel    時間: 2011-4-2 22:50



朋友，我知道你對file system有一定程度的了解，你之前所以說的對於傳統hard drive都是完全正確的。

而且我由第一個回覆就已經說了你的結論是對的 (i.e. 可用空間下降引致降速是對的, 不論HDD or SSD)

不過如果是SSD，成因不是由file system產生，而是由SSD自身的特性引致。而且file fragmentation的產生是對於SSD來說是一個優化手段 (drawback是需要有效的tracing及block recycling，所以才有TRIM command的出現)，不過對HDD來說，fragmentation的產生是不想發生的。

Fragmentation的多少及程度控制各廠商都不同，不但沒有標準而且不對外公開 (那些Algorithm好值錢)，而且SSD是一個完全abstracted的blackbox(有人比喻為computer inside computer)，所以physical layer是完全不受file system管理的。就算是TRIM command，不同的SSD都有不同的處理方法。

所以你說SSD的free space fragmentation及file fragmentation是由OS或FS管理是完全錯誤的。(對於HDD是完全正確)

Diskeeper等Software對第一代的SSD有效，是因為第一代SSD的設計是單純的1-to-1 mapping，那時候Defragmentation software依然可以分析SSD的內部情況，所以效能非常的低。

自Intel x25m第一代之後，主流SSD controller再沒有使用單純的mapping，效能大幅提升之餘，而且已經完全地abstract了。所以現時的SSD不可再被defragmentation software defrag。

時代不同，電腦架構已經完全不同了，舊有的理論已經唔再work。

---

作者: computer1975    時間: 2011-4-2 23:08



雖然意見不一玫，也沒有敵意，也先行謝過，沒必要去做

I still look for target SSD for 1015PEM notebook.

Kingspec SPF-SF12-S50 (S50) - this one is too pricey
http://www.price.com.hk/product.php?p=124982

Look for Intel G3 - have capacitor but have very short 3000 cycle life

I only use 4G SSD for Linux running on XFS.I tell something to done optimize for my SSD. Flash alignment was done
XFS is made to run in fake RAID mode (512byte x 8 = 4k). Let XFS try to operate to write 4K, 8K, 12K .... each time if possible
XFS is smarter than NTFS to prevent fragmentation because Linux filesystem to prevent fragmentation and delay allocation!

---

作者: Niel    時間: 2011-4-2 23:22

本帖最後由 Niel 於 2011-4-2 23:24 編輯


不打緊，我也想大家多了解一點，不過可能要遲一點才有時間寫。

其實3k-5k cycle MLC SSD的壽命是很長的，之前也解釋過:
http://www.hkepc.com/forum/redir ... 24&pid=24691798
所以不用太擔心。請留意計算方法是以日常使用計算 (Intel官方表示WA~1.1)，如果你random write多，WA會大幅提升。

看來朋友還對FS及SSD有一點誤解，請你等我下星期空閒些少再給你解釋。