FlashFire XP 32Bit (0.a0) - New Buffering Engine

Developing versions are uploaded here

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby mokka » Tue Feb 09, 2010 6:53 pm

Hi,

I've been using 0.9d on WinXP for some months now. I bought a low-budget-SSD some more months ago and keep using it since I know FlashFire.
Best regards for this tool! I should donate some money (but I don't find the donate-button) and I wish best success publishing your paper.

I tried out Version a0 again this evening. There are some HANGINGs from time to time, which affect even Windows Desktop. I installed with the option NOT "prefer reliability for performance" (I want performace first) and
DO bypass for big files.

I wonder how much is big and how to tune the registry, anyway there are some hangings beginning shortly after logon and repeatedly while playing around. They seem to be serious. I started watching FlashFire Status (insdispesably great) showing that "live clusters" are being counted down til zero, then "flush count" is increased by 1. Hit Ratios stall in the meantime.
After that the computer is working again for some period.

Did you encounter this behaviour as well?

greetings from europe where is it rather cold those days....
Flo
mokka
 
Posts: 2
Joined: Tue Feb 09, 2010 6:30 pm

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby mokka » Wed Feb 10, 2010 10:27 am

unfortunately, quite after the last post I uninstalled v0.a0 and after reboot, I retrieved Windows STOP 0x0000007B which made my disk no more bootable at all. Unfortunately as well I have to account some data loss due to missing backups )-:

Anyway, I restored a backup from old HDD, installed v0.9d again and right now my system is up and running FAST again.
mokka
 
Posts: 2
Joined: Tue Feb 09, 2010 6:30 pm

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby Orion » Tue Feb 23, 2010 2:14 am

ondra wrote:I tried FleshFire XP 32Bit(0.a0) and it really perfectly works with my XP’s, when are on master drive. Now I am using on master drive Linux a Windows on slave drive, but than the FlashFire do not work. Is it possible to make a release working with slave instead master? I think it would be very useful, as I think I am not only person booting windows from slave drive.


Look in:

HKEY_LOCAL_MACHINE/SYSTEM/CurrentControlSet/Services/ffire/Enum

there you can find the hard drives listed by number, find your slave drive number, probablly the number 1

then go down to Parameters and change DiskNum from 0 (the master) to your slave drive number

reboot and your slave will now be cached.
Orion
 

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby trancemaster2k » Sun Feb 28, 2010 9:57 am

I am using 0.a0 version with 64mb buffer right after testing 0.9f one for some days
without any problems.
During installation I choose "No" (speed/durability thing), "Yes" (big files).

My problem: low 32k values of my 8GB SSD (PATA-ZIF Module).


Here are my ATTO benchmark results:

Code: Select all
without       0.9f          0.a0          0.a0 64mb 

    w     r       w     r       w     r       w     r
02327 02178   09600 01659   10980 01227   10752 01480  0.5
04851 03873   16343 02625   17278 02236   21705 02733  1
08554 07710   22643 05494   26425 06038   26474 05263  2
13278 12738   27374 09570   33048 09822   40354 08991  4
18492 22032   29970 17638   31968 16530   40960 17554  8
24212 36499   36864 26609   37491 26796   47375 25772  16
33851 46152   15493 39479   14727 30986   21652 21652  32
26058 61536   39479 51501   36274 54727   37165 53281  64
46397 70834   36510 62864   46152 67108   45353 66807  128

256-8192 results are as close as the 128 results.


How can I improve better r/w values especially for 32k ?
And will it be possible to improve read speed ?

Bye,
Oliver
MSI Wind U115 Luxury / Atom Z530 / GMA500 / 1GB RAM / 8GB SSD
Mods: Windows XP Prof SP3 (Sereby Updatepack, Driverpacks, nLite), IEGD 10.3 drivers (FullHD 1080p), 128MB Ramdisk (IE8), 1.0E BIOS, Synaptics TouchPad drivers
trancemaster2k
 
Posts: 2
Joined: Sun Feb 28, 2010 9:19 am

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby Pepa » Sun Apr 25, 2010 8:15 am

zartoven wrote:New developing version is available now.
I implemented a new buffering engine to remove some limitations of previous version.

Here are key changes:
- Minimized computation overhead for big capacity SSDs
(I am expecting performance gain on regular SSDs )
- More than 32 Mbytes write buffer can be configured by registry setting.

FlashFire0a0.zip


This is registry setting to configure 64MB buffer size.
Extract zip file, right-click reg file, and choose merge from pop up menu, and reboot your system.
The file is simple text file, and will be able to find out how to configure the buffer size.

ffire-buffersize.zip
Pepa
 

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby asherlazarus » Tue Jun 29, 2010 5:57 pm

If I disable this version of flashfire will I be able to safely hibernate, and then reenable after?
asherlazarus
 
Posts: 3
Joined: Tue Jul 28, 2009 12:24 pm

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby octopus » Sun Jul 04, 2010 11:44 am

Hi,

Just some personal suggestions for your manuscript:

(1) Reliability problem: holding data in buffer may cause file system crash for sudden power failures.


Windows issues the "flush cache" ATA command to the underlying drive for filesystem integrity. Provided that your tool handles this command properly, it is feasible to convince the reviewers that your tool and SSDs using internal buffers support the same level of data integrity.

(2) The buffering technique is not novel: High-end SSDs are using write-coalescing in the drive already.
--> As the market mature, component cost will drop, and HW solution (on-board write buffer) will be more popular. Therefore FlashFire solution will not be useful at that time.


Try finding some optimizations that can be done in the OS-level but not in the device-level. Believe me, there can always be such opportunities.

(3) Log-structured File System is better than FlashFire Solution: Log-structured File System can change random writes (small writes) to sequential writes, and thus there will be no performance problem with log-structured file system.
(and so on)


True, but your tool is a layered design, which means that 1) any existing file systems can use your tool and 2) even LFSs can use your tool as well (try finding some activities that your tool can help LFSs with). Try convincing the reviewers from this viewpoint.

You may also need to include some technical comparison of your tool against related work such as Windows Steady State and "ReSSD: A Software Layer for Resuscitating SSDs from Poor Small Random Write Performance" published in ACM SAC2010.
octopus
 
Posts: 1
Joined: Sun Jul 04, 2010 11:09 am

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby timwjohn » Thu Jul 15, 2010 8:49 am

Hi! I have noticed a great overall improvement in speed using FF0.a0, however this is peppered with BSODs saying PAGE FAULT IN NONPAGED AREA.

I'm using XP SP3 with a Transcend 64GB SSD.

This seems to be during intensive/multiple file reads/writes.

Any tips? Or is it recommended I just stick to the stable version 0.99?
timwjohn
 
Posts: 1
Joined: Thu Jul 15, 2010 8:44 am

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby Zero3K » Thu Sep 02, 2010 11:50 am

Please make this version compatible with Windows 7. Thanks in advance.
Zero3K
 
Posts: 16
Joined: Thu Sep 02, 2010 11:43 am

Re: FlashFire XP 32Bit (0.a0) - New Buffering Engine

Postby Zero3K » Thu Sep 02, 2010 1:14 pm

Actually, it does work on Windows 7 (with no BSODs, so far). All I had to do was enable "Windows XP" compatiblity mode for the setup exe.
Zero3K
 
Posts: 16
Joined: Thu Sep 02, 2010 11:43 am

PreviousNext

Return to FlashFire Lab

Who is online

Users browsing this forum: No registered users and 1 guest