From vvijay03 at gmail.com Thu Oct 11 12:42:44 2012 From: vvijay03 at gmail.com (Vijay Chidambaram) Date: Thu, 11 Oct 2012 11:42:44 -0500 Subject: [Disksim-users] Compiling DiskSim 4.0 on 64-bit machine Message-ID: <9E442330-61B4-4DD1-B708-9D49FD92D97F@gmail.com> Hi, I tried compiling DiskSim on a Ubuntu 64-bit machine with 3.2.0-29-generic kernel. I ran into some problems and fixed them later with the help of some googling and some experimenting. I thought I would send an email collecting the changes I had to do. # Problem 1: Disksim-4.0 does not compile on 64-bit machine. Symptom: An error about "duplicate case" Solution: Apply the 64-bit patch found here: http://code.google.com/p/sst-simulator/downloads/detail?name=disksim_4.0_64bit.patch # Problem 2: Problems due to math linking errors Symptom: undefined reference to 'log', 'sqrt' and so on. Solution: Change the Makefiles so that in LDFLAGS definition, the "-lm" part is at the end of the line. For example: Change LDFLAGS = -L. -lm -ldisksim $(DISKMODEL_LDFLAGS) to LDFLAGS = -L. -ldisksim $(DISKMODEL_LDFLAGS) -lm Hope this helps anyone else running into the same problem. Thanks, Vijay -------------- next part -------------- An HTML attachment was scrubbed... URL: From saeid.barati157 at gmail.com Wed Oct 17 08:02:04 2012 From: saeid.barati157 at gmail.com (Saeid Barati) Date: Wed, 17 Oct 2012 15:32:04 +0330 Subject: [Disksim-users] Write request handling in Disksim Message-ID: Hi all I used Disksim with Microsoft SSD add-on. *Is there any particular function for handling write requests in Disksim? Or How disksim calculate the logical address of incoming request?* * * At first, I thought it would be the "ssd_logical_pageno" function in "ssd.c" file. But it seems that disksim is using another address translation between APN and LPN. Because changin the output of function will not result in different distribution of blocks lifetime. Does anyone know about this? Thanks in advance -------------- next part -------------- An HTML attachment was scrubbed... URL: From abdul.arfan at gmail.com Mon Oct 22 22:13:29 2012 From: abdul.arfan at gmail.com (Abdul Arfan) Date: Tue, 23 Oct 2012 11:13:29 +0900 Subject: [Disksim-users] access time estimation Message-ID: Dear disksim users, Is it possible to get an estimation of access time of a request in the disksim? I have looked at the code and it says that the acctime estimation is not currently supported, is there any other alternative to do this? the estimated access time does not have to be very accurate though. thank you. -- Abdul Arfan (+82 10 4449 6652) http://arfan86.blogspot.com From ykwok2 at miners.utep.edu Mon Oct 22 22:27:50 2012 From: ykwok2 at miners.utep.edu (Yipkei Kwok) Date: Mon, 22 Oct 2012 20:27:50 -0600 Subject: [Disksim-users] access time estimation In-Reply-To: References: Message-ID: Access time of each request *stripe* is passed to disk_acctimestats() in ./src/disksim_disk.c Hope this helps. On Mon, Oct 22, 2012 at 8:13 PM, Abdul Arfan wrote: > Dear disksim users, > > Is it possible to get an estimation of access time of a request in the disksim? > I have looked at the code and it says that the acctime estimation is > not currently supported, > is there any other alternative to do this? the estimated access time > does not have to be very > accurate though. > > thank you. > > -- > Abdul Arfan (+82 10 4449 6652) > http://arfan86.blogspot.com > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > -- ******************************************** Yipkei Kwok (@yipkeikwok) Ph.D. Candidate Research Assistant HiPerSys Lab Department of Computer Science The University of Texas at El Paso Phone: 915 747 6433 (O) E-mail: ykwok2 at miners dot utep dot edu ******************************************** From vahldiek at mpi-sws.org Tue Oct 23 00:30:26 2012 From: vahldiek at mpi-sws.org (Anjo Vahldiek) Date: Tue, 23 Oct 2012 06:30:26 +0200 Subject: [Disksim-users] access time estimation In-Reply-To: References: Message-ID: <50861D62.7040105@mpi-sws.org> Hi, Answering this question is not so simple, since I do not know where you would like to use the estimate. (outside or within disksim) I'm asuming that you would like to get the estimate within disksim, but outside the diskctlr. So that you could e.g. decide which disk or replica of content should serve the request, if multiple disks/blocks hold the same content (but are in different states/content has different location). I cannot give you an entire recipe, but I think what you're looking for can be found in the disk model interface dm_mech_if in diskmodel/dm.h. This interface holds functions to account for entire access time, rotational or seek latency, translate between various models of state representation... I'm not saying that it is very handy to use, but the functionality is there... Now the question is how do you get the implementation of the interface for a specific disk and the current state of the disk. Using the getdisk (globally available) fct call... It gives back a structure containing current state (which is required for most of the interface functions) as well as the implementation of the interface. getdisk itself needs only an integer to return the disk - and this makes it tricky to use. Depending on your implementation it can be difficult to know which disk should serve the request. The latest point at which one can decide the disk is in the controller (smart, simple or your implementation). Anjo On 10/23/2012 04:13 AM, Abdul Arfan wrote: > Dear disksim users, > > Is it possible to get an estimation of access time of a request in the disksim? > I have looked at the code and it says that the acctime estimation is > not currently supported, > is there any other alternative to do this? the estimated access time > does not have to be very > accurate though. > > thank you. > From abdul.arfan at gmail.com Wed Oct 24 22:51:58 2012 From: abdul.arfan at gmail.com (Abdul Arfan) Date: Thu, 25 Oct 2012 11:51:58 +0900 Subject: [Disksim-users] cache algorithm for cachemem/cachedev Message-ID: Dear disksim users, Is there any tutorial or documents on how to add a new cache algorithm for cachemem or cachedev? for example I want to extend the slru algorithm in the cache. Thank you. -- Abdul Arfan (+82 10 4449 6652) http://arfan86.blogspot.com From kmudzinhwuynh at gmail.com Thu Oct 25 06:13:55 2012 From: kmudzinhwuynh at gmail.com (KMU) Date: Thu, 25 Oct 2012 19:13:55 +0900 Subject: [Disksim-users] [DiskSimUser] Page Size & Time to Read/Write Message-ID: <508910E3.40800@gmail.com> Hello all Disksim-users, Currently I am working with Disksim. However, I am confused with many things so I would like to ask you many questions and hope you will be able spend a little time to help me. 1. In "ssd-iozone.parv" file: - Line 181, Page size = 8 - What is the unit sie of page size (Kilo byte or others) ? 2. If page size is 8 KB, so size of a block is 512KB. However, when I calculate read block transfer time and write block transfer time, there are no any results which math in parameter file ----------------------------------------- Read block transfer time = 0.0002384, Write block transfer time = 0.0002384, #Read block transfer time = 0.0016276, #Write block transfer time = 0.0016276, ------------------------------------------ My calculation is * If data transfer is 2GBps --> time = 0.000262144 (second) * If data transfer is 2Gbps --> time = 0.002097152 (second) * If data transfer is 30MBps --> time = 0.01747626 (second) Best Regards, -- /************************** ----- Richard Huynh ----- ----- KMU ----- **************************/ From abdul.arfan at gmail.com Sat Oct 27 19:16:53 2012 From: abdul.arfan at gmail.com (Abdul Arfan) Date: Sun, 28 Oct 2012 08:16:53 +0900 Subject: [Disksim-users] SLRU in disksim Message-ID: Dear all disksim users, Currently I'm trying to understand the working of cachemem in disksim, Is there any information on how SLRU cache replacement policy works in disk sim other than reading the source code? Thank you. -- Abdul Arfan (+82 10 4449 6652) http://arfan86.blogspot.com