What Happens When the Hard Drive Crashes?
You may never experience an actual hard drive crash but if you do you will undoubtedly know what has happened. Of all the computers I have owned I have never had a “real” hard drive crash. I have had my computer crash but not from a hard drive failure. I have recently worked on at least two different laptops than had hard drive failures and the only way for a repair was hard drive replacement.
Let us take a look at what actually happens when a hard drive crashes. In a recent article I referred to the distance between the platter and the read/write head and it is a matter of NM’s (Nanometers) above the platter. When you think about this thickness, a sheet of paper can be thicker than the distance between the platter and the read/write head. This is the reason for being careful of how we treat the computer and not be rough when moving it etc. another item of harm is dust and grit. The hard drive is somewhat sealed but things can happen and dirt can sometimes get into the hard drive itself. Recently I got a laptop to work on and when I plugged it in and turned it on you could hear the hard drive grinding against things it was not supposed to be touching. I knew without looking the hard drive was history; however, I still took it apart to investigate further.
Sometimes part of the data can be recovered but often it is not, whatever was being written at the time and all of the past memories on the drive may be lost forever. But what actually happened? The platter is certainly not indestructible and scratches can occur and dropping the computer may in fact cause extreme damage to the drive. Because the platter is spinning so fast when it comes in contact with something it should not, bad things are going to happen and it may be lost forever. This is a good reason for backup plans but like any good business plan, the backup plan must be before it is needed. An extra hard drive could be the solution and if it is setup as a RAID1 it will be a redundant copy of everything on the drive. If you have a RAID0 extra drive installed the data is alternated between drives so only part of your material would be lost or you could have a good external hard drive where data is backed up on a routine basis to secure at least a part of your data. I have my external set to backup every Saturday morning
Wednesday, April 28, 2010
Monday, April 19, 2010
Tracks, Heads, and Sectors
Broader Description of Tracks, Heads, and Sectors
In the previous writing I discussed a little about the Hard Drive Capacity but did not go into much depth. This could leave some confusion about the amount of data stored because of not knowing how each of these parts work. When looking at the “Head” this is the part of the Hard Drive that actually reads and writes data either to or from the platters on the hard drive. If you Google these parts you can find pictures and as the saying goes, a picture is worth a thousand words. In this case, that certainly applies. The picture will show the head as being more like an arm and this arm extends over the platters which are spinning. Some platters spin at a rate of 10,000 revolutions per minute (RPM) and this arm is a paper thin distance above the platters. This is measured in nanometres and if you looked to try to see between the head and the platters the naked eye probably could not detect a space in this small separation. Hard Drives normally have multiple platters and the platters are two sided with data on both sides. The heads or arms that read and write data are on both sides of the platter and work simultaneously transferring or reading data. Not all platter are two sided however as is the case with floppy disk drives.
As in earlier writings I have mentioned tracks but did not explain in depth. Tracks are sometimes referred to as cylinders but mean the same thing. These are the tracks going around the platter in ever decreasing circles if looking from the outside edge toward the inside where the small hole is in the center. You can see the results of writing to the platter if you will take the CD disk out and look at the shinny side. You will see a difference in the tracks and notice it writes from the inside toward the outside of the platter.
The sectors on the disk are addresses on the tracks and on the older disk there were 17 sectors and all were the same size and contained equal amounts of data or 512 bytes of data regardless of the radius of the track. Looking at the track you understand from the inside near the hole in the center toward the outside edge the radius the size changes but the sector which contains the data remains the same regardless of where it is on the track, and it contains 512 bytes of data. A bit is one single piece of data and a byte is equal to eight bits. If reading this in literature or from Compact Disk information if it is bytes it would be written as MB (Notice the capitol B) for megabytes or Mb (Notice lower case b) for megabits.
In the previous writing I discussed a little about the Hard Drive Capacity but did not go into much depth. This could leave some confusion about the amount of data stored because of not knowing how each of these parts work. When looking at the “Head” this is the part of the Hard Drive that actually reads and writes data either to or from the platters on the hard drive. If you Google these parts you can find pictures and as the saying goes, a picture is worth a thousand words. In this case, that certainly applies. The picture will show the head as being more like an arm and this arm extends over the platters which are spinning. Some platters spin at a rate of 10,000 revolutions per minute (RPM) and this arm is a paper thin distance above the platters. This is measured in nanometres and if you looked to try to see between the head and the platters the naked eye probably could not detect a space in this small separation. Hard Drives normally have multiple platters and the platters are two sided with data on both sides. The heads or arms that read and write data are on both sides of the platter and work simultaneously transferring or reading data. Not all platter are two sided however as is the case with floppy disk drives.
As in earlier writings I have mentioned tracks but did not explain in depth. Tracks are sometimes referred to as cylinders but mean the same thing. These are the tracks going around the platter in ever decreasing circles if looking from the outside edge toward the inside where the small hole is in the center. You can see the results of writing to the platter if you will take the CD disk out and look at the shinny side. You will see a difference in the tracks and notice it writes from the inside toward the outside of the platter.
The sectors on the disk are addresses on the tracks and on the older disk there were 17 sectors and all were the same size and contained equal amounts of data or 512 bytes of data regardless of the radius of the track. Looking at the track you understand from the inside near the hole in the center toward the outside edge the radius the size changes but the sector which contains the data remains the same regardless of where it is on the track, and it contains 512 bytes of data. A bit is one single piece of data and a byte is equal to eight bits. If reading this in literature or from Compact Disk information if it is bytes it would be written as MB (Notice the capitol B) for megabytes or Mb (Notice lower case b) for megabits.
Friday, April 16, 2010
Floppies and Hard Drives
STORAGE – HARD DRIVES AND FLOPPIES
Floppy Drives
In the study of a hard drive the first place to look is at floppy drive technology. Even though floppy drives are old and obsolete they are still around in a few computers and there is a relationship between a floppy and a hard drive. Looking back at older computers you may remember having an A drive and a B drive. A drive was in some cases the boot drive and required a disk to boot; however, that is going way back. A drive was a 3.5" drive and held 1.44 MB of storage. Both A & B hooked to the motherboard via a data cable with a 34 pin connector. On these older computers both A & B were connected by the same data cable; however, newer computers will only have one cable and just the 3.5 floppy drive. Drive B was a 5.25" floppy and since they were both hooked to the motherboard by the same data cable, the distinction of these two floppies was made by the BIOS. The 5.25" drive has gone the way of the Dinosaurs and is no longer available on most computers. The cable used was identifiable by a twist in the cable for drive A and number 1 pin had a color along the edge of the cable.
In preparing the floppy for use you need to know a little about its construction. Inside that disk is a flat, blank, magnetically coated plastic cylinder with nothing on it. There are no groves, no sectors, and no writing. In order to use this cylinder we must prepare it or make it ready to accept data by a process called formatting. The first thing done is to format the disk by inserting it into the proper drive, going to the DOS prompt (A ://>) and typing the command in the line telling the computer to prepare this disk for use. When you hit enter it will begin its work of creating circles or tracks and sections called sectors on the disk. When it is finished a high density disk will have 80 circles or tracks on each side of the disk and 18 sectors on each side and the circles will be numbered 0 through 79 and the sectors numbered 0 through 18. All sectors will store the same amount of data. The data is written to the disk in binary code of 1's and 0's with each bit being magnetized. I normally do formatting from the dos prompt but you can also format from a shortcut in Windows Explorer. All material is written using magnetism for the binary code 1's and 0's and a different magnetic code for the spaces. When it is read the computer uses a magnetic read/write head mechanism in the floppy drive. There is some difference in the method used by a floppy drive as opposed to a hard drive. The floppy drive head lightly touch the surface of the disk while the disk is rotating at 360 revolutions per minute. The floppy drive has two heads, one on top of the cylinder and one beneath and they move back and forth in unison. A hard drive works the same way but the drive heads never touch the disk.
Let's dig a little deeper in this hole of knowledge. Sectors can be grouped as one or more sectors together making up a cluster. A cluster is the smallest unit of space on a disk used for storage and just like your office at home you need a filing system to keep everything organized. On your computer that filing system is a "FAT" or File Allocation Table and on some newer units it will be a NTFS or New Technology File System used on Windows NT, 2000, and XP. NTFS provided better security and more storage than FATs.
Now that we have a File Allocation System we need some explanations about the system. The first sector in the file cabinet is the boot record and it tells us how many sectors the disk has and how many sectors are allocated per cluster and how many bits are in each FAT entry. It will also advise of the version of DOS or Windows used in the formatting process of the disk.
If your floppy drive has a problem the idea now is to replace it because the price has made repairing of floppy drives obsolete. I am not going to go into detail here on replacing the floppy. It is very simple and directions for your computer can be found on the internet in Google. One word of caution, if you do decide to replace a defective floppy don't forget about electro static discharge (ESD) and make sure you are properly grounded before entering the case. In the beginning of this discussion I stated the floppy drive and hard drive were similar. In fact they are very similar. Where a floppy has one Mylar disk a hard drive will have two and possibly more.
HARD DRIVE
Let me emphasize hard drives are fragile and should be handled gently. If you listen to your hard drive as it spins up and hear crunching or unusual sounds, prepare to install a new hard drive. The platters or disk spin in unison and all that I have seen were housed in a metal case. Inside this case are firmware chips and devices to read and write information to the platters or disks. Just like floppies the key to a hard drive is magnetism. Material is written to the platters by a magnetic spot and this is one reason nothing with magnetism should ever get close to your hard drive or for that matter the computer. Magnetism can totally destroy the data on a hard drive in microseconds. Another similarity to a floppy is if you have multiple platters data is written to both sides. Similar to a floppy data is written to the platter beginning from the outside of the circle and continuing to the inside and even though it would seem the sectors are smaller they all contain the same amount of data. This does not hold true with newer units because a new technology was introduced called zone bit recording. Even though it differs from the older drives each sector still holds 512 bits of information. Just like the floppy, hard drives are disposable and when sectors begin to fade or go bad, pull the drive out and replace with a new one. Just be sure to try and recover as much of your data as possible. Hard drives eventually fail and will begin rendering "Bad Sector or Sector Not Found" errors and this is usually caused by fading of the low-level formatting done by the factory whereby it set track and sector markings. As the computer ages these lines gradually fade to the point of eventual failure errors. They have a life expectancy of, "the life of the drive."
Older computers built during the first 1990 years were rather primitive and their controllers couldn't handle complicated arrangements that changed between tracks. As a result, every track had the same number of sectors. This wasted a great deal of space but since all sectors held the same number of bytes, it was easy to calculate the size of the hard drive. If you knew how many tracks, heads, and sectors were on the hard drive you could simply multiply the number of heads, by tracks, by sectors, by 512 bytes and this gave you the size of the hard drive. An example is (855 X 7 X 17 X 512.) then you must divide by 1024 and divide again by 1024 to convert to MB and this gives the size of the drive. I remember my first 25 MB computer and at that time in 1993 it was state of the art. Today it would not have enough memory to install even the smallest available PC operating system.
Thursday, April 8, 2010
MORE ABOUT RAM
WHAT ABOUT MEMORY?
Ram Technologies
Computer Memory is divided into two categories: RAM and ROM. RAM stands for random access memory and ROM stands for read only memory. ROM retains its memory when the PCI where it is located is turned off; however, RAM memory will be lost. The Central Processor Unit CPU contains a small amount of STATIC RAM embedded within the processor housing. On the motherboard there is a firmware chip that holds the ROM BIOS.
Most RAM is stored on the motherboard in modules called DIMMs, RIMMs, or SIMMs. There are two types of RAM, dynamic and static. When ordering RAM it will be listed as DRAM or SRAM. Of the two, SRAM is the faster because it does not have to be refreshed nearly as often and does not lose data as quickly. Located inside the processor of most computers is a memory cache used to speed up memory access. This memory will be SRAM because SRAM is faster than DRAM. Some computers even have a small amount of SRAM embedded on the motherboard in addition to the SRAM in the cache location. SRAM and DRAM are made of transistors and a transistor is a switch. To keep a transistor switch open it must have a small amount of voltage at all times. DRAM transistor switches are charged by capacitors and they must be recharged; however, SRAM does not use capacitors so SRAMs are faster, they are also more expensive. Because of cost you will have a small amount of SRAM on a computer and a lot of DRAM.
As stated above DRAM is always stored in DIMM, RIMM, or SIMM modules which are plugged directly into the motherboard. The difference in these modules is the width of the data path each module accommodates and the manner data is transferred from the system bus to the module. The DDR2 chip is into a DIMM module. It has 240 pins and is the fastest memory at this time. This chip can be inserted as a single channel or dual. The difference is that the dual channel sticks are matched pairs of memory. To work the modules must be technically identical: same size, same timings, etc. Single data rate (SDR) DDR2 is able to operate with faster bus speeds; your memory doesn't hold back the performance of your processor. Generally speaking, motherboards are built to support only one type of memory. You cannot mix and match SDRAM, DDR, or DDR2 memory on the same motherboard in any system. They will not function and will not even fit in the same sockets. DDR stands for Double Data Rate; DDR2 is the newer and faster technology and is replacing DDR.
The following is a discussion of the advantages of DDR2 Memory over DDR.
The difference in DDR2 is the architecture that supports improvements in chipsets and in other system components. DDR2 processes memory more efficiently and reduces system latency time and can operate at a data rate of 667 MHz and sometimes faster. It is also known for consuming lower power rates which prolongs notebook battery life. DDR uses 2.5 volts of electricity and SDRAM consumes 3.3 volts but DDR2 only uses 1.8 volts. These lower voltages result in lower operating temperatures which is a big plus factor for servers and notebooks.
The third picture is a 168 pin SDR DIMM or Single Duty Rate and has 168 pins and has two notches instead of one. The two notches regulate the voltage and rather it is buffered or unbuffered, registered or unregistered.
RIMM stands for Rambus inline memory modules and RIMM modules have 184 pins used for Rambus DRAM. I have only seen one RIMM module and they are recognizable by the heat shield around the chips due to the excessive heat generated. Rambus Inc developed RDRAM or Direct Rambus DRAM and RDRAM uses RIMM. Every slot must be filled when using RIMM memory with either a RIMM memory card or a C-RIMM (Continuity RIMM) also called a placeholder module in order to maintain continuity throughout all slots. The C-RIMM does not contain a memory chip but the slot must be filled with a card.
The last memory chip I will mention is a 72 pin and a 30 pin SIMM. The 72 pin must be installed with two modules to a bank of memory and the 32 pin SIMM module and must be installed four modules to a memory bank. SIMM means single inline memory module. SIMMs are rated by speed and measured in nanoseconds, the data path is 32 bit. Speed is measured in nanoseconds and in rating the lower nanosecond is faster than the larger number because it is measured from the time it takes for the processor to request the data, the memory controller to find the data on the module chip and load the data onto the bus and return to the processor to be read and then the memory controller to refresh the memory chip on the SIMM. So we are measuring elapsed time measured in nanoseconds instead of how fast the data is traveling.
Asynchronously or Synchronously
Older DRAM operated asynchronously but newer memory is synchronous. Synchronous keeps time with the system clock and asynchronous does not and this makes the newer DRAM a better choice. As technology improves the goal remains to increase speed but maintain accuracy.
DIMM Technology
The data width of DIMM modules is a 64 bit path and this by itself would make DIMM faster that SIMM at its 32 bit. This means that instead of sending 32 bits of information at a time the DIMM is sending 64. DIMM’s can have chips on one side or both sides and can have double or single sided contacts. If they are double sided each side operates independently of the other. This double sided technology makes them faster than SIMM’s. DIMM’s can hold from 8 MB up to 2 GB of Random Access Memory (RAM) and have 168, 184 or 240 pins on the connector side of the board. SDRAM or synchronous DRAM was the first DIMM to be synchronized with the system clock. Improvements to DIMMs are the DDR, DDR2, BUFFERED, REGISTERED, and dual channel DIMMs. Extended Data Out or (EDO) extended or improved the performance of Fast Page Mode by eliminated wait states by keeping the output buffer active until the next cycle began. Burst EDO or (Burst EDO) was a faster type of EDO that gained speed by using an address counter for next addresses and a pipeline stage that overlapped operations. These types of memory probably do not have significant meaning to you but they do show a history of memory improvements and some still in use. Typically the memory modules I run across are the DDR2 DIMMs.
BUFFERED AND REGISTERED DIMMS
The purpose for buffering and registering is to amplify the signal just before the data is written to the module. Not all DIMMs use buffering or registering and some are either buffered or registered; however if a DIMM uses buffers it is called a buffered DIMM; if it uses registers it is called a registered DIMM. This seems kind of redundant but if it does not support registers or buffers, it is always referred to as an unbuffered DIMM.
“Extended Data Out (EDO)” and “Fast Page Memory (FPM)” modules used buffers and buffering did not degrade performance. Registers do reduce performance in SDRAM some; however, motherboard can support more memory modules when SDRAM uses registers. Servers using advanced buffering technologies are fully buffered DIMMs (FB-DIMMs) and this makes it possible to support more DIMMs.
If you read the identifying description the position of the notches on memory modules identifies the technologies it supports. The major purpose for the notches, while it does help identify the module, it also determines where the module will go. Because of the notches you will not put the wrong module in the wrong slot, it will not fit. When looking at the module, the notch on the left identifies if it is buffered, registered, unbuffered or unregistered memory and the notch on the right determines the voltage used by the module.
SOME MOTHERBOARDS USE DUAL CHANNELING
When DDR or DDR2 uses dual channeling on the motherboard memory and performance is improved. An SDRAM stands for Synchronous Dynamic Random Access Memory. DDR is short for "DDR SDRAM" and stands for Double Data Rate. It is dual channeling technology that improves memory and enables the controller to communicate with two DIMMs at the same time and this ability effectively doubles the speed of access. The circuit board on the motherboard has two channels and they are simply called Channel A and Channel B. Each of the channels implements dual channeling. To support dual channeling you install two matched DIMMs in Channel A and two matched DIMMs in Channel B and each can be accessed at the same time. For clarification the two DIMMs in Channel A are accessed at the same time and the two DIMMs in Channel B are accessed at the same time. By matching the DIMMs this doubles the access time. To explain how this is possible the DIMMs are both at the same time placing 64 bits of data at the same time and because it is two different DIMMs this will effectively double the access rate from 64 bit to 128 bit. These DIMMs must be equally matched in speed, size, and features for dual channeling to work. If you have the computer case open you will see the two slots are different colors. The Channel A slots are yellow and the two Channel B slots are black. If you have matched DIMMs in the yellow slots and decide to put DIMMs in Channel B black slots they must match but they do not have to match the DIMMs in Channel A, the two Channels run independent of each other. If the DIMMs are not matched the motherboard will revert to single channeling.
ERROR CHECKING AND PARITY
In parity we must first remember machine language there are no letters and every input is in binary code, either a 1 or 0. Older computers the memory banks or a row was nine chips on each memory card. The data path of older machines was eight bits with a bit being a single bit of information, either a 1 or 0. The computer recognizes this bit as a switch for electrical current and the switch is either on or off and in the case of an 8 bit card each of the switches would either be on or off. A 1 turns the switch on and a 0 turns it off. The switch being on is represented by a 1 and if off it will be a 0. When explaining “parity,” each bank held one byte and as you know a byte equals 8 bits. In parity, every byte has an even number of one’s or every byte has an odd number of ones. In describing parity every bank has an extra chip to store an extra bit so every byte has nine bits rather than eight. The ninth bit is an error checking bit and if the computer uses odd parity it makes the ninth bit or the parity bit either a one or zero in order to comply with parity of the computer. In other words it makes the ninth bit either a one or a zero representing whatever is needed for conformity. Simplifying even more, when the error checking bit counts the number of 1’s and finds 4 and 4 zeroes it determines it is even parity and the computer needs it to be odd parity it will make the ninth bit either another one or a zero thus correcting the error. This system of error checking is no longer used. Error correcting has been improved and today it uses error-correcting code (ECC) and can detect and correct an error in a single bit. ECC is supported by some SDRAM, DDR, DDR2 and RIMM memory modules. Not all memory modules support error correcting code and where it does it will have an odd number of chips on the module. A DIMM normally is a 64-bit module, but if it supports ECC it will be a 71 OR 72 bit module. Important to remember is to always replace memory with what the motherboard supports.
CAS LATENCY AND RAS LATENCY
Two additional memory features to think about are column access strobe (CAS) and row access strobe (RAS) these are two means of measuring speed relative to clock cycles or the number of clock cycles it takes to write or read a row or column of data from a memory module. CAS stands for column access strobe and RAS means row access strobe.
DO YOU WANT THAT IN TIN OR GOLD?
Both the connectors on the memory modules and the connectors inside the memory slots are made of either tin or gold. On older computers the connectors were made of gold but as a cost reduction factor modern machines are now using tin. Mismatching the connectors, or placing gold chips in tin slots can result in a chemical reaction causing corrosion which may lead to memory errors and may prevent your computer from booting.
DIFFERENT MEASURMENTS OF MEMORY SPEED
Memory speed is measured in different ways and sometimes this makes it difficult for the non technician to determine what they are buying when purchasing a new computer or ordering replacement parts. There are also several factors that will determine how fast memory will run. Several ways speed is measured are nanoseconds, MHz, PC rating, RAS Latency and CAS Latency. Earlier we discussed parity, ECC, dual channeling, and buffering and each of these will affect memory speed. SDRAM, DDR, DDR2, and RIMMs are measured in MHz but sometimes DIMMs and RIMMs are measured according to PC rating. It is not unusual to find DDR listed as DDR266 or DDR333 meaning it runs at either 266 MHz or 333MHz. A way to understand this is a module has a 64 bit data path which means it yields a path of 8 bytes and 8 x 8 = 64 or 64 bits. This then would effectively yield a speed of 2128 MD pr second or DDR266 x 8 = 2128MB/second.
When buying a new computer determine how much memory is installed because the more memory, the faster the system and you should get as much memory as the system and your wallet can support. Look at the technology used and know that a DDR2 is faster than a DDR and DDR is faster than SDR SDRAM. Regardless of the technology when replacing or adding RAM always use what the mother board and its technology supports.
If you install modules of different speeds in the same computer system, the system will run at the slowest speed or it will become unstable. You should use the fastest memory the motherboard supports. You usually look for the higher number but in the case of CAS OR RAS lower numbers are better. You probably will not see either of these ratings because they are not normally listed. And last to use dual channeling make sure you use matching pairs of memory in each channeling slots.
Some of this material was found in search engines on the internet and some form Guide to Managing and Maintaining your PC. To the best of my knowledge there are no direct quotes however the material was learned while studying from this book and referencing on the web while preparing for the A+ exam.
UPGRADING MEMORY
To upgrade memory simply means to add more RAM. Some people confuse memory with storage, memory is RAM, storage is like your HARD DRIVE or DVD, etc. If you are getting an error message of “Insufficient memory” it means you need to add RAM. If your computer is new and never had the case off it may have memory slots with no cards installed. If this is the case, how much memory do you need? Previously I emphasized buy whatever the computer and your wallet will support and certainly that is true here.
HOW MUCH MEMORY TO BUY?
Jean Andrews in her book “Guide to Maintaining and Managing your PC” when you are adding memory to ask yourself these questions:
*How much memory do I need?
*How much RAM is currently installed in my system?
*How many and what kind of memory modules are currently installed on my motherboard?
*How much and what kind of memory can I fit on my motherboard?
*How do I select and purchase the right memory for my upgrade?
Then she proceeds to answer these important questions. First she states the same thing I have written before, “All you can get.” In my writing I state you probably should get all you can afford but of course this is relative to the other questions Jean asks. It would be foolish to buy more RAM than can go in your PC but if you take my answer at face value you may have a shelf with unused memory. Before we buy RAM we must know what is required by the software we have installed and the hardware it is going into. The different operating systems have different requirements for memory. In my previous articles I have written to determine memory right click on “My Computer,” then click on properties and you could find what was installed. A better method because it gives much more information is to click on the “Start” button then type in the “Run” box “Msinfo32” and hit Enter. A new window will be displayed with the information needed. The amount of physical and virtual memory can affect how your computer operates and if it is sluggish.
A physical examination of the inside of the computer will appraise you of the number of memory slots available and the type, size, manufacturer, and a real close exam will reveal if the memory was relabeled or not. It is not unusual for some unscrupulous builders to reuse memory as a cheap method of constructing a computer. Do a little scrape test with your fingernail to tell if it has been relabeled.
Ram Technologies
Computer Memory is divided into two categories: RAM and ROM. RAM stands for random access memory and ROM stands for read only memory. ROM retains its memory when the PCI where it is located is turned off; however, RAM memory will be lost. The Central Processor Unit CPU contains a small amount of STATIC RAM embedded within the processor housing. On the motherboard there is a firmware chip that holds the ROM BIOS.
Most RAM is stored on the motherboard in modules called DIMMs, RIMMs, or SIMMs. There are two types of RAM, dynamic and static. When ordering RAM it will be listed as DRAM or SRAM. Of the two, SRAM is the faster because it does not have to be refreshed nearly as often and does not lose data as quickly. Located inside the processor of most computers is a memory cache used to speed up memory access. This memory will be SRAM because SRAM is faster than DRAM. Some computers even have a small amount of SRAM embedded on the motherboard in addition to the SRAM in the cache location. SRAM and DRAM are made of transistors and a transistor is a switch. To keep a transistor switch open it must have a small amount of voltage at all times. DRAM transistor switches are charged by capacitors and they must be recharged; however, SRAM does not use capacitors so SRAMs are faster, they are also more expensive. Because of cost you will have a small amount of SRAM on a computer and a lot of DRAM.
As stated above DRAM is always stored in DIMM, RIMM, or SIMM modules which are plugged directly into the motherboard. The difference in these modules is the width of the data path each module accommodates and the manner data is transferred from the system bus to the module. The DDR2 chip is into a DIMM module. It has 240 pins and is the fastest memory at this time. This chip can be inserted as a single channel or dual. The difference is that the dual channel sticks are matched pairs of memory. To work the modules must be technically identical: same size, same timings, etc. Single data rate (SDR) DDR2 is able to operate with faster bus speeds; your memory doesn't hold back the performance of your processor. Generally speaking, motherboards are built to support only one type of memory. You cannot mix and match SDRAM, DDR, or DDR2 memory on the same motherboard in any system. They will not function and will not even fit in the same sockets. DDR stands for Double Data Rate; DDR2 is the newer and faster technology and is replacing DDR.
The following is a discussion of the advantages of DDR2 Memory over DDR.
The difference in DDR2 is the architecture that supports improvements in chipsets and in other system components. DDR2 processes memory more efficiently and reduces system latency time and can operate at a data rate of 667 MHz and sometimes faster. It is also known for consuming lower power rates which prolongs notebook battery life. DDR uses 2.5 volts of electricity and SDRAM consumes 3.3 volts but DDR2 only uses 1.8 volts. These lower voltages result in lower operating temperatures which is a big plus factor for servers and notebooks.
The third picture is a 168 pin SDR DIMM or Single Duty Rate and has 168 pins and has two notches instead of one. The two notches regulate the voltage and rather it is buffered or unbuffered, registered or unregistered.
RIMM stands for Rambus inline memory modules and RIMM modules have 184 pins used for Rambus DRAM. I have only seen one RIMM module and they are recognizable by the heat shield around the chips due to the excessive heat generated. Rambus Inc developed RDRAM or Direct Rambus DRAM and RDRAM uses RIMM. Every slot must be filled when using RIMM memory with either a RIMM memory card or a C-RIMM (Continuity RIMM) also called a placeholder module in order to maintain continuity throughout all slots. The C-RIMM does not contain a memory chip but the slot must be filled with a card.
The last memory chip I will mention is a 72 pin and a 30 pin SIMM. The 72 pin must be installed with two modules to a bank of memory and the 32 pin SIMM module and must be installed four modules to a memory bank. SIMM means single inline memory module. SIMMs are rated by speed and measured in nanoseconds, the data path is 32 bit. Speed is measured in nanoseconds and in rating the lower nanosecond is faster than the larger number because it is measured from the time it takes for the processor to request the data, the memory controller to find the data on the module chip and load the data onto the bus and return to the processor to be read and then the memory controller to refresh the memory chip on the SIMM. So we are measuring elapsed time measured in nanoseconds instead of how fast the data is traveling.
Asynchronously or Synchronously
Older DRAM operated asynchronously but newer memory is synchronous. Synchronous keeps time with the system clock and asynchronous does not and this makes the newer DRAM a better choice. As technology improves the goal remains to increase speed but maintain accuracy.
DIMM Technology
The data width of DIMM modules is a 64 bit path and this by itself would make DIMM faster that SIMM at its 32 bit. This means that instead of sending 32 bits of information at a time the DIMM is sending 64. DIMM’s can have chips on one side or both sides and can have double or single sided contacts. If they are double sided each side operates independently of the other. This double sided technology makes them faster than SIMM’s. DIMM’s can hold from 8 MB up to 2 GB of Random Access Memory (RAM) and have 168, 184 or 240 pins on the connector side of the board. SDRAM or synchronous DRAM was the first DIMM to be synchronized with the system clock. Improvements to DIMMs are the DDR, DDR2, BUFFERED, REGISTERED, and dual channel DIMMs. Extended Data Out or (EDO) extended or improved the performance of Fast Page Mode by eliminated wait states by keeping the output buffer active until the next cycle began. Burst EDO or (Burst EDO) was a faster type of EDO that gained speed by using an address counter for next addresses and a pipeline stage that overlapped operations. These types of memory probably do not have significant meaning to you but they do show a history of memory improvements and some still in use. Typically the memory modules I run across are the DDR2 DIMMs.
BUFFERED AND REGISTERED DIMMS
The purpose for buffering and registering is to amplify the signal just before the data is written to the module. Not all DIMMs use buffering or registering and some are either buffered or registered; however if a DIMM uses buffers it is called a buffered DIMM; if it uses registers it is called a registered DIMM. This seems kind of redundant but if it does not support registers or buffers, it is always referred to as an unbuffered DIMM.
“Extended Data Out (EDO)” and “Fast Page Memory (FPM)” modules used buffers and buffering did not degrade performance. Registers do reduce performance in SDRAM some; however, motherboard can support more memory modules when SDRAM uses registers. Servers using advanced buffering technologies are fully buffered DIMMs (FB-DIMMs) and this makes it possible to support more DIMMs.
If you read the identifying description the position of the notches on memory modules identifies the technologies it supports. The major purpose for the notches, while it does help identify the module, it also determines where the module will go. Because of the notches you will not put the wrong module in the wrong slot, it will not fit. When looking at the module, the notch on the left identifies if it is buffered, registered, unbuffered or unregistered memory and the notch on the right determines the voltage used by the module.
SOME MOTHERBOARDS USE DUAL CHANNELING
When DDR or DDR2 uses dual channeling on the motherboard memory and performance is improved. An SDRAM stands for Synchronous Dynamic Random Access Memory. DDR is short for "DDR SDRAM" and stands for Double Data Rate. It is dual channeling technology that improves memory and enables the controller to communicate with two DIMMs at the same time and this ability effectively doubles the speed of access. The circuit board on the motherboard has two channels and they are simply called Channel A and Channel B. Each of the channels implements dual channeling. To support dual channeling you install two matched DIMMs in Channel A and two matched DIMMs in Channel B and each can be accessed at the same time. For clarification the two DIMMs in Channel A are accessed at the same time and the two DIMMs in Channel B are accessed at the same time. By matching the DIMMs this doubles the access time. To explain how this is possible the DIMMs are both at the same time placing 64 bits of data at the same time and because it is two different DIMMs this will effectively double the access rate from 64 bit to 128 bit. These DIMMs must be equally matched in speed, size, and features for dual channeling to work. If you have the computer case open you will see the two slots are different colors. The Channel A slots are yellow and the two Channel B slots are black. If you have matched DIMMs in the yellow slots and decide to put DIMMs in Channel B black slots they must match but they do not have to match the DIMMs in Channel A, the two Channels run independent of each other. If the DIMMs are not matched the motherboard will revert to single channeling.
ERROR CHECKING AND PARITY
In parity we must first remember machine language there are no letters and every input is in binary code, either a 1 or 0. Older computers the memory banks or a row was nine chips on each memory card. The data path of older machines was eight bits with a bit being a single bit of information, either a 1 or 0. The computer recognizes this bit as a switch for electrical current and the switch is either on or off and in the case of an 8 bit card each of the switches would either be on or off. A 1 turns the switch on and a 0 turns it off. The switch being on is represented by a 1 and if off it will be a 0. When explaining “parity,” each bank held one byte and as you know a byte equals 8 bits. In parity, every byte has an even number of one’s or every byte has an odd number of ones. In describing parity every bank has an extra chip to store an extra bit so every byte has nine bits rather than eight. The ninth bit is an error checking bit and if the computer uses odd parity it makes the ninth bit or the parity bit either a one or zero in order to comply with parity of the computer. In other words it makes the ninth bit either a one or a zero representing whatever is needed for conformity. Simplifying even more, when the error checking bit counts the number of 1’s and finds 4 and 4 zeroes it determines it is even parity and the computer needs it to be odd parity it will make the ninth bit either another one or a zero thus correcting the error. This system of error checking is no longer used. Error correcting has been improved and today it uses error-correcting code (ECC) and can detect and correct an error in a single bit. ECC is supported by some SDRAM, DDR, DDR2 and RIMM memory modules. Not all memory modules support error correcting code and where it does it will have an odd number of chips on the module. A DIMM normally is a 64-bit module, but if it supports ECC it will be a 71 OR 72 bit module. Important to remember is to always replace memory with what the motherboard supports.
CAS LATENCY AND RAS LATENCY
Two additional memory features to think about are column access strobe (CAS) and row access strobe (RAS) these are two means of measuring speed relative to clock cycles or the number of clock cycles it takes to write or read a row or column of data from a memory module. CAS stands for column access strobe and RAS means row access strobe.
DO YOU WANT THAT IN TIN OR GOLD?
Both the connectors on the memory modules and the connectors inside the memory slots are made of either tin or gold. On older computers the connectors were made of gold but as a cost reduction factor modern machines are now using tin. Mismatching the connectors, or placing gold chips in tin slots can result in a chemical reaction causing corrosion which may lead to memory errors and may prevent your computer from booting.
DIFFERENT MEASURMENTS OF MEMORY SPEED
Memory speed is measured in different ways and sometimes this makes it difficult for the non technician to determine what they are buying when purchasing a new computer or ordering replacement parts. There are also several factors that will determine how fast memory will run. Several ways speed is measured are nanoseconds, MHz, PC rating, RAS Latency and CAS Latency. Earlier we discussed parity, ECC, dual channeling, and buffering and each of these will affect memory speed. SDRAM, DDR, DDR2, and RIMMs are measured in MHz but sometimes DIMMs and RIMMs are measured according to PC rating. It is not unusual to find DDR listed as DDR266 or DDR333 meaning it runs at either 266 MHz or 333MHz. A way to understand this is a module has a 64 bit data path which means it yields a path of 8 bytes and 8 x 8 = 64 or 64 bits. This then would effectively yield a speed of 2128 MD pr second or DDR266 x 8 = 2128MB/second.
When buying a new computer determine how much memory is installed because the more memory, the faster the system and you should get as much memory as the system and your wallet can support. Look at the technology used and know that a DDR2 is faster than a DDR and DDR is faster than SDR SDRAM. Regardless of the technology when replacing or adding RAM always use what the mother board and its technology supports.
If you install modules of different speeds in the same computer system, the system will run at the slowest speed or it will become unstable. You should use the fastest memory the motherboard supports. You usually look for the higher number but in the case of CAS OR RAS lower numbers are better. You probably will not see either of these ratings because they are not normally listed. And last to use dual channeling make sure you use matching pairs of memory in each channeling slots.
Some of this material was found in search engines on the internet and some form Guide to Managing and Maintaining your PC. To the best of my knowledge there are no direct quotes however the material was learned while studying from this book and referencing on the web while preparing for the A+ exam.
UPGRADING MEMORY
To upgrade memory simply means to add more RAM. Some people confuse memory with storage, memory is RAM, storage is like your HARD DRIVE or DVD, etc. If you are getting an error message of “Insufficient memory” it means you need to add RAM. If your computer is new and never had the case off it may have memory slots with no cards installed. If this is the case, how much memory do you need? Previously I emphasized buy whatever the computer and your wallet will support and certainly that is true here.
HOW MUCH MEMORY TO BUY?
Jean Andrews in her book “Guide to Maintaining and Managing your PC” when you are adding memory to ask yourself these questions:
*How much memory do I need?
*How much RAM is currently installed in my system?
*How many and what kind of memory modules are currently installed on my motherboard?
*How much and what kind of memory can I fit on my motherboard?
*How do I select and purchase the right memory for my upgrade?
Then she proceeds to answer these important questions. First she states the same thing I have written before, “All you can get.” In my writing I state you probably should get all you can afford but of course this is relative to the other questions Jean asks. It would be foolish to buy more RAM than can go in your PC but if you take my answer at face value you may have a shelf with unused memory. Before we buy RAM we must know what is required by the software we have installed and the hardware it is going into. The different operating systems have different requirements for memory. In my previous articles I have written to determine memory right click on “My Computer,” then click on properties and you could find what was installed. A better method because it gives much more information is to click on the “Start” button then type in the “Run” box “Msinfo32” and hit Enter. A new window will be displayed with the information needed. The amount of physical and virtual memory can affect how your computer operates and if it is sluggish.
A physical examination of the inside of the computer will appraise you of the number of memory slots available and the type, size, manufacturer, and a real close exam will reveal if the memory was relabeled or not. It is not unusual for some unscrupulous builders to reuse memory as a cheap method of constructing a computer. Do a little scrape test with your fingernail to tell if it has been relabeled.
Subscribe to:
Posts (Atom)