Among the components of a PC, there are several essential elements that should
not choose at random: the processor, motherboard, graphics card and memory.
These four elements have a direct and significant influence on the functionality
and performance that you issue your computer. Now lets see memory timings and
their usefulness in practice.
Memory is a crucial part in the
performance and stability of a PC.There are other criteria such as speed,
latency, capacity, etc.. It is these elements that to deal with numerical
calculation.
RAM
The storage of information is the same as other media used in computers, ie
binary form of 0 and 1. The bit cell is the unit of this type of memory, he took
the value of a 0 or a 1. It includes these bits table cells in multi-million for
memories of different sizes.
The table structure allows you to send a
bit cell with a coordinated intersection of a row and a column. It is therefore
easy to read or write data when you want it to memory, knowing only the address
of the bit cell concerned. The reading or writing data is done from anywhere of
memory with the same access time. That is why it is called Random Access Memory
or RAM. All submissions current work on this principle.
Dynamic memory or DRAM
DRAM or Dynamic RAM is so named because it is obliged to refresh to work
properly.
The bit cell consists of a couple physically transistor and
capacitor. The transistor connects the intersections of the table (in a sense,
it’s a switch) and is then connected to a capacitor that stores the state 1 or 0
corresponding to a load or no load.
The capacitor was chosen for its
simplicity of manufacture, its very small footprint and hence its cost. It is
easy to use on a much smaller area to produce large memory while modules with
small and inexpensive.
Drawbacks:
- The load and discharge are not instantaneous which determines the latency of
memory, - A capacitor loses its charge over time, they say it leaks. It may not keep
the information in a short time, that is why the memory is refreshed
periodically.
A memory card is composed of several chips. Each chip is divided into "banks".
Each bank is an array . Each table cell contains 1 byte of data (8 bits). For
example: 256 MB of memory consists of 8 chips of 32MB each chip contains 4 banks
of 8 MB (8192 = 024 lines x 1 column x 1 byte).
For DDR-2 are made from
chips of 128 MB (by increasing the fine print), which are then used to construct
512 MB DIMMs (equipped with 4 chips), 2GB (equipped with 16 chips) …
The memory currently in the most widespread is the DDR SDRAM (Double Data Rate)
which, like the RAMBUS uses two fronts so as to treat twice as much data as the
frequency equal to SDRAM (DDR PC1600 to PC2100 and 100 MHz to 133 MHz). The
format is still used but the DIMM slots from 84 to 92 or 184 pin connectors. The
frequencies available for this memory range from 100 to 300 MHz (which is rare)
but this increase in frequency was achieved at the expense of latency of less
aggressive as the rise, particularly in Above 200 MHz …
DDR2
introduced with the Intel i925X chipset and i915.Physically DDR2 has a number of
connectors highest: 240 against 184 for the DDR standard. High frequency is the
main feature of the DDR2. Thus, the lowest frequency of DDR2 400 MHz (200 MHz
FSB). Chipsets Intel i915 and i925 also support DDR2 266 MHz FSB is 533 MHz DDR.
The modules the highest speeds PC5400 DIMMs at 675 MHz DDR, an actual frequency
of 337.5 MHz.
DDR2 has a function different from the classic DDR. Where
DDR first name returns 2 bits at each cycle, the 4 refers to DDR2. It operates
internally rather than DDR (dual data rate) but in QDR (quad data rate). So with
an internal frequency of 100 MHz, we are left with the DDR2-400 where DDR400
used a classical internal frequency of 200 Mhz. In short, the objective of DDR2
is clearly mounting frequency. If the DDR2 modules arrive one day at a clock
frequency of 200 MHz internally, then we will have DDR2-800 modules with a
bandwidth of 6.4 GB / s. Here is also a summary of the bandwidth provided by DDR
and DDR2:
DDR | Frequency internal | bandwidth |
DDR166 | 133 mhz | 2.1 gb/s |
DDR333 | 166mhz | 2.7 gb/s |
DDR400 | 200 mhz | 3.2 gb/s |
DDR433 | 217 mhz | 3.5 gb/s |
DDR466 | 233 mhz | 3.7 gb/s |
DDR500 | 250 mhz | 4 gb/s |
DDR533 | 266 mhz | 4.5 gb/s |
DDR550 | 175 mhz | 4.6 gb/s |
DDR2-400 | 200 mhz | 3.2 gb/s |
DDR2- 533 | 266 mhz | 4.2 gb/s |
DDR2-667 | 333 mhz | 5.3 gb/s |
Memory access
When the system wants to access an area of memory to read or write data, it
sends a command ACTIVE containing the line number and an order of selection, the
RAS (Row Access Strobe). The selected line on the full extent of the matrix
memory is called memory page. The page is opened with the command selection.
spoke here the first time the latency tRCD (RAS to CAS Delay) which is the time
at the RAM to accomplish this selection line before executing the next
instruction.
The command of reading or writing is executed (at time t +
tRCD). This command contains the number of the column and the order of
selection, the CAS (Column Address Strobe). Then the second time the latency TCL
(CAS Latency) is the time given to the RAM to perform the read operation or
writing before the next instruction.
The PRECHARGE command is executed
to close the memory previously used. Spoke here a third time to the latency tRAS
(Active to precharge Delay) which is the minimum period between the first
instruction ACTIVE and PRECHARGE command. PRECHARGE command s’ therefore
executes at t + tRAS and closes a page memory.
The next instruction is a
new memory access, but the memory requires a time to prepare to perform a new
direction. Spoke here the fourth time the TRP latency (RAS precharge time) that
is time left in memory after a PRECHARGE to run a new command ACTIVE.
Note:
- An instruction runs tRCD + TCL + (tRAS – tRCD + TCL) + tRAS + TRP or TRP,
- TRCD and TCL have no connection with the tRAS, tRCD , except that TCL <
tRAS. It uses command = tRAS tRCD + TCL + 2 - You can configure to your liking the different values of latency, however,
all the RAM will not follow. If you do not leave enough time in memory to
prepare or execute an instruction problems like data corruption, system
instability, slow or crash occur.
The SPD
The Bios of machines can configure the timings of the modules through a memory
chip eprom present at the same bars,the SPD (Serial Presence Detect). The BIOS
reads the information on this chip (capacity, frequency, latency, as well as
manufacturer, serial number …) to boot the computer to adjust the operating
parameters of the RAM.
An extension of the SPD ,the EPP comes on the
market for high-end memory. This additional memory can store more information on
the possibilities of memory modules with respect to the frequencies and
latencies possible. This allows to use the statements in automatic overclocking
accessible by everyone. This requires memory modules with EPP chipset allowing
the EPP and activate the option in the BIOS of the machine.
Another point
of DDR2: latency times very high. Currently the best memory timings are
3-3-3-8 for DDR2 533 and 4-4-4-12 for DDR2 675 … It is hoped that this will
improve in the coming months.
What is the timings?
The frequency is not the only determining factor for good performance. The
timings are indeed related to the performance of your PC and a proper adjustment
will allow you to optimize, especially in games. Today, most manufacturers of
RAM do not hesitate to put forward the famous timings shock as an argument for
marketing themselves to a particular brand and especially strongly augmenter the
selling price. With the disappearance of the famous Winbond BH5, BH6 and others,
had lost the famous aggressive CAS2 timings.All memory timing even DDR3 RAM timing can be done this way.
The timings are commonly encountered CAS, RAS to CAS Delay Time RAS precharge
and RAS Active Time:
- CAS (Column Access Strobe) (tCL) Timing is the minimum time required to
access a column of a bench.It controls the number of time cycles between sending
and the time to act on it.The lower value of these in cycles, the higher the
performance.The CAS is often put forward by the manufacturers, and this is
partly due to the fact that many still believe that this single parameter is
sufficient to ensure the best performance. Yet other parameters also influence
and direct the performance. That is why manufacturers also calculate these
settings
e.g.: 2-2-3-10 The “2” is the CAS timing. - tRAS Timing is RAS Active Time which is the time required to access a line
to a bench.until tRAS has completed row cannot be deactivated. The lower valu is
the faster the performance, but if it is too low then possibility of data
corruption.
tRAS = tCL + tRCD + tRP e.g.: 2-2-3-10 The bold “10”
is the tRAS timing. - tRCD Timing is the RAS to CAS Delay (Row Access Strobe to CAS) is the
minimum time between an access line to that of a column.
e.g.:
2-2-3-10 The bold “2” is the tRCD timing. - tRP timing is RAS precharge Time which is the minimum time between two
signals RAS
e.g.: 2-2-3-10 The bold “3” is the tRP
timing.
Configure ram timing setting
Whatever the timing, the more value you give is lower the better performance of
your PC. However there is a limit not to exceed the memory modules are sold to
operate at certain timings certified. Before any change in the timings should
check the stability of your memory timings to the original (those programmed in
the SPD) using the excellent utility Memtest86 +. You need to create a boot disk
containing the utility or if you do not drive or floppy disk to create a
bootable CD .You have to download the ISO at the address http://www.memtest.org or download the following utility
UltimateBootCD.All in there is voltage and frequency timing important for successive result.
Memtest86 + will launch when several sequences of tests in the form of
passes. The ideal is to spend two consecutive passes without errors to affirm /
confirm the stability of the memory settings. If these tests were successful,
you can now change the settings through the bios of your motherboard, usually in
the menu under Advanced Chipset Features, but for security reasons it is better
not to modify a value at a time and check each time the stability of your memory
always with Memtest86 +.
In case of instability you can always return to
the previous value or increase the voltage of your slides in steps of ± 0.1V.
However, do not exceed the 2.85V for DDR failing to take the risk of damaging.
if you use more than one memory the timings will be limited to the less powerful
card provided they are different. This situation is also not recommended on any
platform supporting dual channel memory. It is also common today to see
manufacturers such as OCZ, Corsair or KINGSTON Dual Channel kits certifying
specifications identical bars to ensure a flawless operation, but also not to be
blocked by one of them for the settings of timings.
Memory Latency
The RAM’s CAS latency is nothing but what we saw above it is 4 numbers, such as
2-2-3-10, which correspond with CAS / tRCD / tRP / tRAS. The CAS Latency can be
cahnge from bios setting .so what we all see above is memory latency.
Intel platform
It allow you to enjoy more aggressive ram timings overclock the Pentium 4
3.2Ghz to 3.4 Ghz FSB from 228 to 243 Mhz. In both cases,an overclocked CPU and
memory synchronized with the latter.It is observed that the performance of the
Pentium 4 running at 3.4 GHz are slightly lower than the Pentium 4 3.2Ghz
advantage of aggressive timings and note well the differences between
conventional and aggressive timings at the same configuration.
AMD Athlon 64 platform
With Athlon 64 but different graphics card than the one used with the Pentium4
.here are the results obtained with an Athlon 64 3500 + 0.09μ and the DDR400
Corsair 2-2-2-5, the most aggressive timings show their interest by winning FPS.
the FPS are not earned enormous and difficult to justify the extra cost to
dispose of barrettes 2-2-2-5 .. It should also be noted that the platform used
here is less sensitive to latency time could be as a platform combined with an
nForce2 Athlon XP.
As we have seen, the timings affect the overall
performance of the PC. This influence is not very obvious in some games and will
be in others. However, the choice of bars is not left to chance.you will have to
see what the best price / timings / frequencies that you can offer. Therefore,
only you to judge whether it is worth or not, the budget can not afford some of
stick.However, these expensive bars have good guaranteed for life.
The
choice of memory should be based on your processor and choice of the memory
capacity even if it is certified only for traditional timings .it would be
detrimental to the risk of instability or incompatibilities. To ensure
compatibility and quality of memory with your current or future motherboard,
please visit the manufacturer web site before purchase.