PCI express 3.0 expansion slot. What is PCI Express. PCI-E bus formats

PCI - Express (PCIePCI -E)– serial, universal bus first unveiled July 22, 2002 of the year.

Is general, unifying a bus for all nodes of the system board, in which all devices connected to it coexist. Came to replace an outdated tire PCI and its variations AGP, due to increased requirements for bus throughput and the inability to improve the speed performance of the latter at reasonable cost.

The tire acts as switch, simply sending a signal from one point to another without changing it. This allows, without obvious loss of speed, with minimal changes and errors transmit and receive a signal.

Data on the bus goes simplex(full duplex), that is, simultaneously in both directions at the same speed, and signal along the lines flows continuously, even when the device is turned off (as D.C., or a bit signal of zeros).

Synchronization constructed using a redundant method. That is, instead of 8 bit information is transmitted 10 bits, two of which are official (20% ) and serve in a certain sequence beacons For synchronization clock generators or identifying errors. Therefore, the declared speed for one line in 2.5 Gbps, is actually equal to approximately 2.0 Gbps real.

Nutrition each device on the bus, selected separately and regulated using technology ASPM (Active State Power Management). It allows when the device is idle (without sending a signal) lower its clock generator and put the bus into mode reduced energy consumption. If no signal is received within a few microseconds, the device considered inactive and switches to mode expectations(time depends on device type).

Speed ​​characteristics in two directions PCI - Express 1.0 :*

1 x PCI-E~ 500 Mbps

4x PCI-E~ 2 Gbps

8 x PCI-E~ 4 Gbps

16x PCI-E~ 8 Gbps

32x PCI-E~ 16 Gbps

*Data transfer speed in one direction is 2 times lower than these indicators

January 15, 2007, PCI-SIG released an updated specification called PCI-Express 2.0

The main improvement was in 2 times increased speed data transmission ( 5.0 GHz, against 2.5GHz V old version). Also improved point-to-point communication protocol(dot-to-dot), modified software component and added system software monitoring according to the tire speed. At the same time, it was preserved compatibility with protocol versions PCI-E 1.x

In the new version of the standard ( PCI -Express 3.0 ), the main innovation will be modified coding system And synchronization. Instead of 10 bit systems ( 8 bit information, 2 bits official), will apply 130 bit (128 bit information, 2 bits official). This will reduce losses in speed from 20% to ~1.5%. Will also be redesigned synchronization algorithm transmitter and receiver, improved PLL(phase-locked loop).Transmission speed expected to increase 2 times(compared to PCI-E 2.0), wherein compatibility will remain with previous versions PCI-Express.

When changing just one video card, be sure to take into account that new models may simply not fit your motherboard, since there are not just several different types of expansion slots, but also several different versions of them (for both AGP and PCI Express). If you are not confident in your knowledge on this topic, please read the section carefully.

As we noted above, the video card is inserted into a special expansion slot on the computer’s motherboard, and through this slot the video chip exchanges information with central processor systems. On motherboards Most often, there are expansion slots of one or two different types, differing in bandwidth, power settings and other characteristics, and not all of them are suitable for installing video cards. It is important to know the connectors available in the system and buy only the video card that matches them. Different expansion connectors are physically and logically incompatible, and a video card designed for one type will not fit into another and will not work.

Fortunately, over the past time, not only the ISA and VESA Local Bus expansion slots (which are of interest only to future archaeologists) and the corresponding video cards have sunk into oblivion, but also the video cards for PCI slots have practically disappeared, and all AGP models are hopelessly outdated. And everyone is modern GPUs They use only one type of interface - PCI Express. Previously, the AGP standard was widely used; these interfaces differ significantly from each other, including throughput, the capabilities provided for powering the video card, as well as other less important characteristics.

Only a very small part of modern motherboards do not have PCI Express slots, and if your system is so old that it uses an AGP video card, then you won’t be able to upgrade it - you need to change the entire system. Let's take a closer look at these interfaces; these are the slots you need to look for on your motherboards. See photos and compare.

AGP (Accelerated Graphics Port or Advanced Graphics Port) is a high-speed interface based on the PCI specification, but created specifically for connecting video cards and motherboards. The AGP bus, although better suited for video adapters compared to PCI (not Express!), provides a direct connection between the central processor and the video chip, as well as some other features that increase performance in some cases, for example, GART - the ability to read textures directly from RAM , without copying them to video memory; higher clock speeds, simplified data transfer protocols, etc., but this type of slot is hopelessly outdated and new products with it have not been released for a long time.

But still, for the sake of order, let’s mention this type. The AGP specifications appeared in 1997, when Intel released the first version of the specification, including two speeds: 1x and 2x. In the second version (2.0) AGP 4x appeared, and in 3.0 - 8x. Let's consider all the options in more detail:
AGP 1x is a 32-bit link operating at 66 MHz, with a throughput of 266 MB/s, which is twice the PCI bandwidth (133 MB/s, 33 MHz and 32 bits).
AGP 2x is a 32-bit channel operating with double the bandwidth of 533 MB/s at the same frequency of 66 MHz due to data transmission on two fronts, similar to DDR memory (only for the direction “to the video card”).
AGP 4x is the same 32-bit channel operating at 66 MHz, but as a result of further tweaks, a quadruple “effective” frequency of 266 MHz was achieved, with a maximum throughput of more than 1 GB/s.
AGP 8x - additional changes in this modification made it possible to obtain throughput up to 2.1 GB/s.

Video cards with an AGP interface and the corresponding slots on motherboards are compatible within certain limits. Video cards rated for 1.5V do not work in 3.3V slots, and vice versa. However, there are also universal connectors that support both types of boards. Video cards designed for a morally and physically outdated AGP slot have not been considered for a long time, so to learn about old AGP systems, it would be better to read the article:

PCI Express (PCIe or PCI-E, not to be confused with PCI-X), formerly known as Arapahoe or 3GIO, differs from PCI and AGP in that it is a serial rather than parallel interface, allowing for fewer pins and higher bandwidth. PCIe is just one example of the move from parallel to serial buses; other examples of this movement are HyperTransport, Serial ATA, USB, and FireWire. An important advantage of PCI Express is that it allows multiple single lanes to be stacked into one channel to increase throughput. Multi-channel serial design increases flexibility, slow devices can be allocated fewer lines with a small number of contacts, and fast devices can be allocated more.

The PCIe 1.0 interface transfers data at 250 MB/s per lane, which is almost double the capacity of conventional PCI slots. The maximum number of lanes supported by PCI Express 1.0 slots is 32, which gives a throughput of up to 8 GB/s. A PCIe slot with eight working lanes is approximately comparable in this parameter to the fastest AGP version - 8x. Which is even more impressive when you consider the ability to transmit simultaneously in both directions at high speeds. The most common PCI Express x1 slots provide single lane bandwidth (250 MB/s) in each direction, while PCI Express x16, which is used for video cards and combines 16 lanes, provides up to 4 GB/s bandwidth in each direction.

Although the connection between two PCIe devices is sometimes made up of several lanes, all devices support a single lane at a minimum, but can optionally handle more of them. Physically, PCIe expansion cards fit and work normally in any slots with an equal or greater number of lanes, so a PCI Express x1 card will work smoothly in x4 and x16 slots. Also, a physically larger slot can work with a logically smaller number of lines (for example, it looks like a regular x16 connector, but only 8 lines are routed). In any of the above options, PCIe itself will select the highest possible mode and will work normally.

Most often, x16 connectors are used for video adapters, but there are also boards with x1 connectors. And most motherboards with two PCI Express x16 slots operate in x8 mode to create SLI and CrossFire systems. Physically, other slot options, such as x4, are not used for video cards. Let me remind you that all this applies only to the physical level; there are also motherboards with physical PCI-E x16 connectors, but in reality with 8, 4 or even 1 channels. And any video cards designed for 16 channels will work in such slots, but with lower performance. By the way, the photo above shows the x16, x4 and x1 slots, and for comparison, PCI is also left (below).

Although the difference in games is not that big. Here, for example, is a review of two motherboards on our website, which examines the difference in the speed of 3D games on two motherboards, a pair of test video cards in which operate in 8-channel and 1-channel modes, respectively:

The comparison we are interested in is at the end of the article, pay attention to the last two tables. As you can see, the difference at medium settings is very small, but in heavy modes it begins to increase, and a large difference is noted in the case of a less powerful video card. Please take note.

PCI Express differs not only in throughput, but also in new power consumption capabilities. This need arose because the AGP 8x slot (version 3.0) can transfer only no more than 40 watts in total, which was already lacking in the video cards of that time designed for AGP, which were installed with one or two standard four-pin power connectors. The PCI Express slot can carry up to 75W, with an additional 75W available through the standard six-pin power connector (see last section of this part). Recently, video cards have appeared with two such connectors, which in total gives up to 225 W.

Subsequently, the PCI-SIG group, which develops relevant standards, presented the main specifications of PCI Express 2.0. The second version of PCIe doubled the standard bandwidth, from 2.5 Gbps to 5 Gbps, so that the x16 connector can transfer data at speeds of up to 8 GB/s in each direction. At the same time, PCIe 2.0 is compatible with PCIe 1.1; old expansion cards usually work fine in new motherboards.

The PCIe 2.0 specification supports transfer speeds of both 2.5 Gbps and 5 Gbps, this is done to ensure backward compatibility with existing solutions PCIe 1.0 and 1.1. PCI Express 2.0 backwards compatibility allows legacy 2.5 Gb/s solutions to be used in 5.0 Gb/s slots, which will then simply operate at a lower speed. And devices designed to version 2.0 specifications can support speeds of 2.5 Gbps and/or 5 Gbps.

Although the main innovation in PCI Express 2.0 is the speed doubled to 5 Gbps, this is not the only change; there are other modifications to increase flexibility, new mechanisms for program control connection speed, etc. We are most interested in changes related to the power supply of devices, since the power requirements of video cards are steadily increasing. PCI-SIG has developed a new specification to address the increasing power consumption of graphics cards, which expands current power supply capabilities to 225/300 W per graphics card. To support this specification, a new 2x4-pin power connector is used, designed to provide power to high-end graphics cards.

Video cards and motherboards with support for PCI Express 2.0 appeared on wide sale already in 2007, and now you can’t find others on the market. Both major video chip manufacturers, AMD and NVIDIA, have released new lines of GPUs and video cards based on them, supporting the increased bandwidth of the second version of PCI Express and taking advantage of new electrical power capabilities for expansion cards. All of them are backward compatible with motherboards that have PCI Express 1.x slots on board, although in some rare cases there is incompatibility, so you need to be careful.

Actually, the emergence of the third version of PCIe was an obvious event. In November 2010, the specifications for the third version of PCI Express were finally approved. Although this interface has a transfer rate of 8 Gt/s instead of 5 Gt/s in version 2.0, it throughput again increased exactly twice compared to the PCI Express 2.0 standard. To do this, we used a different coding scheme for data sent over the bus, but it was compatible with previous versions PCI Express remains the same. The first products of the PCI Express 3.0 version were presented in the summer of 2011, and real devices have only just begun to appear on the market.

A whole war broke out among motherboard manufacturers for the right to be the first to introduce a product with support for PCI Express 3.0 (mainly based on Intel chipset Z68), and several companies presented corresponding press releases at once. Although at the time of updating the guide, there are simply no video cards with such support, so it’s simply not interesting. By the time PCIe 3.0 support is needed, completely different boards will appear. Most likely, this will happen no earlier than 2012.

By the way, we can assume that PCI Express 4.0 will be introduced over the next few years, and the new version will also once again double the bandwidth in demand by that time. But this will not happen soon, and we are not interested yet.

External PCI Express

In 2007, the PCI-SIG, a formal standards group PCI solutions Express, announced the adoption of the PCI Express External Cabling 1.0 specification, which describes the data transfer standard over the PCI Express 1.1 external interface. This version allows data transfer at a speed of 2.5 Gbps, and the next one should increase the throughput to 5 Gbps. The standard includes four external connectors: PCI Express x1, x4, x8 and x16. The older connectors are equipped with a special tongue that makes connection easier.

The external version of the PCI Express interface can be used not only for connecting external video cards, but also for external drives and other expansion cards. The maximum recommended cable length is 10 meters, but it can be increased by connecting the cables through a repeater.

Theoretically, this could make life easier for laptop lovers, when they use a low-power built-in video core when running on batteries, and a powerful external video card when connected to a desktop monitor. Upgrading such video cards is significantly easier; there is no need to open the PC case. Manufacturers can make completely new cooling systems that are not limited by the features of expansion cards, and there should be fewer problems with power supply - most likely, external power supplies will be used, designed specifically for a specific video card; they can be built into one external case with a video card, using one cooling system. It may make it easier to assemble systems on multiple video cards (SLI/CrossFire), and given the constant growth in popularity of mobile solutions, such external PCI Express should have gained some popularity.

They should have, but they didn’t win. As of fall 2011 external options There are practically no video cards on the market. Their range is limited by outdated models of video chips and a narrow selection of compatible laptops. Unfortunately, the business of external video cards did not go any further and slowly died out. We don’t even hear winning advertising statements from laptop manufacturers anymore... Perhaps the power of modern mobile video cards has simply become enough even for demanding 3D applications, including many games.

There remains hope for the development of external solutions in a promising interface for connectivity peripheral devices Thunderbolt, formerly known as Light Peak. It was developed by Intel Corporation based on DisplayPort technology, and the first solutions have already been released by Apple. Thunderbolt combines the capabilities of DisplayPort and PCI Express and allows you to connect external devices. However, so far they simply do not exist, although cables already exist:

In this article we do not touch on outdated interfaces; the vast majority of modern video cards are designed for the PCI Express 2.0 interface, so when choosing a video card, we suggest considering only it; all data on AGP is provided for reference only. The new boards use the PCI Express 2.0 interface, combining the speed of 16 PCI Express lanes, which gives a throughput of up to 8 GB/s in each direction, which is several times more than the same characteristic of the best AGP. In addition, PCI Express operates at such speeds in each direction, unlike AGP.

On the other hand, products with support for PCI-E 3.0 have not really come out yet, so it doesn’t make much sense to consider them either. If we are talking about upgrading an old one or purchasing new board or simultaneously changing the system and video cards, then you just need to purchase boards with the PCI Express 2.0 interface, which will be quite sufficient and most widespread for several years, especially since products of different versions of PCI Express are compatible with each other.

In the spring of 1991, Intel completed development of the first prototype version of the PCI bus. The engineers were tasked with developing an inexpensive and high-performance solution that would realize the capabilities of the 486, Pentium and Pentium Pro processors. In addition, it was necessary to take into account the mistakes made by VESA when designing the VLB bus (the electrical load did not allow connecting more than 3 expansion cards), and also to implement automatic setup devices.

In 1992, the first version of the PCI bus appeared, Intel announced that the bus standard would be open, and created the PCI Special Interest Group. Thanks to this, any interested developer has the opportunity to create devices for the PCI bus without having to purchase a license. The first version of the bus had a clock frequency of 33 MHz, could be 32- or 64-bit, and devices could operate with signals of 5 V or 3.3 V. Theoretically, the bus throughput was 133 MB / s, but in reality the throughput was about 80 MB/s

Main characteristics:

• bus frequency - 33.33 or 66.66 MHz, synchronous transmission;
• bus width - 32 or 64 bits, multiplexed bus (address and data are transmitted over the same lines);
• peak throughput for the 32-bit version operating at 33.33 MHz is 133 MB/s;
• memory address space - 32 bits (4 bytes);
• address space of I/O ports - 32 bits (4 bytes);
• configuration address space (for one function) - 256 bytes;
• voltage - 3.3 or 5 V.

Photos of connectors:

MiniPCI - 124 pin
MiniPCI Express MiniSata/mSATA - 52 pin
Apple MBA SSD, 2012
Apple SSD, 2012
Apple PCIe SSD
MXM, Graphics Card, 230 / 232 pin
MXM2 NGIFF 75 pins KEY A PCIe x2 KEY B PCIe x4 Sata SMBus
MXM3, Graphics Card, 314 pin
PCI 5V
PCI Universal
PCI-X 5v
AGP Universal
AGP 3.3v
AGP 3.3 v + ADS Power
PCIe x1
PCIe x16
Custom PCIe
ISA 8bit
ISA 16bit
eISA
VESA
NuBus
PDS
PDS
Apple II/GS Expasion slot
PC/XT/AT expasion bus 8 bit
ISA (industry standard architecture) - 16 bit
eISA
MBA - Micro Bus architecture 16 bit
MBA - Micro Bus architecture with 16 bit video
MBA - Micro Bus architecture 32 bit
MBA - Micro Bus architecture with 32 bit video
ISA 16 + VLB (VESA)
Processor Direct Slot PDS
601 Processor Direct Slot PDS
LC Processor Direct Slot PERCH
NuBus
PCI (Peripheral Computer Interconnect) - 5v
PCI 3.3v
CNR (Communications / network riser)
AMR (Audio/Modem Riser)
ACR (Advanced communication riser)
PCI-X (Peripheral PCI) 3.3v
PCI-X 5v
PCI 5v + RAID option - ARO
AGP 3.3v
AGP 1.5v
AGP Universal
AGP Pro 1.5v
AGP Pro 1.5v+ADC power
PCIe (peripheral component interconnect express) x1
PCIe x4
PCIe x8
PCIe x16

PCI 2.0

The first version of the basic standard to become widespread used both cards and slots with a signal voltage of only 5 volts. Peak throughput - 133 MB/s.

PCI 2.1 - 3.0

They differed from version 2.0 by the possibility of simultaneous operation of several bus masters (English bus-master, the so-called competitive mode), as well as the appearance of universal expansion cards capable of operating both in slots using a voltage of 5 volts, and in slots using 3 .3 volts (with a frequency of 33 and 66 MHz, respectively). Peak throughput for 33 MHz is 133 MB/s, and for 66 MHz it is 266 MB/s.

• Version 2.1 - work with cards designed for a voltage of 3.3 volts, and the presence of appropriate power lines were optional.
• Version 2.2 - expansion cards made in accordance with these standards have a universal power connector key and are able to work in many later types of PCI bus slots, as well as, in some cases, in version 2.1 slots.
• Version 2.3 - Incompatible with PCI cards designed to use 5 volts, despite the continued use of 32-bit slots with a 5 volt key. Expansion cards have a universal connector, but are not able to work in 5-volt slots of earlier versions (up to 2.1 inclusive).
• Version 3.0 - completes the transition to 3.3 volt PCI cards, 5 volt PCI cards are no longer supported.

PCI 64

An extension of the basic PCI standard, introduced in version 2.1, that doubles the number of data lanes, and therefore the throughput. The PCI 64 slot is an extended version of the regular PCI slot. Formally, the compatibility of 32-bit cards with 64-bit slots (provided there is a common supported signal voltage) is full, but the compatibility of a 64-bit card with 32-bit slots is limited (in any case there will be a loss of performance). Operates at a clock frequency of 33 MHz. Peak throughput - 266 MB/s.

• Version 1 - uses a 64-bit PCI slot and a voltage of 5 volts.
• Version 2 - uses a 64-bit PCI slot and a voltage of 3.3 volts.

PCI 66

PCI 66 is a 66 MHz evolution of PCI 64; uses 3.3 volts in the slot; the cards have a universal or 3.3 V form factor. Peak throughput is 533 MB/s.

PCI 64/66

The combination of PCI 64 and PCI 66 allows for four times the data transfer speed compared to basic standard PCI; uses 64-bit 3.3V slots, compatible only with universal ones, and 3.3V 32-bit expansion cards. PCI64/66 standard cards have either a universal (but with limited compatibility with 32-bit slots) or a 3.3-volt form factor (the latter option is fundamentally incompatible with 32-bit 33-MHz slots of popular standards). Peak throughput - 533 MB/s.

PCI-X

PCI-X 1.0 is an expansion of the PCI64 bus with the addition of two new operating frequencies, 100 and 133 MHz, as well as a separate transaction mechanism to improve performance when multiple devices operate simultaneously. Generally backward compatible with all 3.3V and generic PCI cards. PCI-X cards are usually implemented in a 64-bit 3.3B format and have limited backward compatibility with PCI64/66 slots, and some PCI-X cards are in a universal format and are capable of working (although this has almost no practical value) in the usual PCI 2.2/2.3. In difficult cases, in order to be completely confident in the functionality of the combination of the motherboard and expansion card, you need to look at the compatibility lists of the manufacturers of both devices.

PCI-X 2.0

PCI-X 2.0 - further expansion of the capabilities of PCI-X 1.0; frequencies of 266 and 533 MHz have been added, as well as parity error correction during data transmission (ECC). Allows splitting into 4 independent 16-bit buses, which is used exclusively in built-in and industrial systems ; The signal voltage has been reduced to 1.5 V, but the connectors are backward compatible with all cards using a signal voltage of 3.3 V. Currently, for the non-professional segment of the high-performance computer market (powerful workstations and servers entry level), in which the PCI-X bus is used, very few motherboards supporting the bus are produced. An example of a motherboard for this segment is ASUS P5K WS. In the professional segment it is used in RAID controllers and SSD drives for PCI-E.

Mini PCI

Form factor PCI 2.2, intended for use mainly in laptops.

PCI Express

PCI Express, or PCIe, or PCI-E (also known as 3GIO for 3rd Generation I/O; not to be confused with PCI-X and PXI) - computer bus(although at the physical level it is not a bus, being a point-to-point connection), using software model PCI buses and a high-performance physical protocol based on serial data transmission. The development of the PCI Express standard was started by Intel after abandoning the InfiniBand bus. Officially, the first basic PCI Express specification appeared in July 2002. The development of the PCI Express standard is carried out by the PCI Special Interest Group.

Unlike the PCI standard, which used a common bus for data transfer with multiple devices connected in parallel, PCI Express, in general, is a packet network with star topology. PCI Express devices communicate with each other through a medium formed by switches, with each device directly connected by a point-to-point connection to the switch. In addition, the PCI Express bus supports:

• hot swap cards;
• guaranteed bandwidth (QoS);
• energy management;
• monitoring the integrity of transmitted data.

The PCI Express bus is intended to be used only as a local bus. Because software model PCI Express is largely inherited from PCI, existing systems and controllers can be modified to use the PCI Express bus by replacing only physical level, without modification software. The high peak performance of the PCI Express bus allows it to be used instead of AGP buses, and even more so PCI and PCI-X. De facto, PCI Express replaced these buses in personal computers.

• MiniCard (Mini PCIe) - replacement for the Mini PCI form factor. The Mini Card connector has the following buses: x1 PCIe, 2.0 and SMBus.
  • M.2 is the second version of Mini PCIe, up to x4 PCIe and SATA.
• ExpressCard - similar to PCMCIA form factor. The ExpressCard connector supports x1 PCIe and USB 2.0 buses; ExpressCard cards support hot plugging.
• AdvancedTCA, MicroTCA - form factor for modular telecommunications equipment.
  
• Mobile PCI Express Module (MXM) is an industrial form factor created for laptops by NVIDIA. It is used to connect graphics accelerators.
• PCI Express cable specifications allow the length of one connection to reach tens of meters, which makes it possible to create a computer whose peripheral devices are located at a considerable distance.
  
• StackPC - specification for building stackable computer systems. This specification describes the expansion connectors StackPC, FPE and their relative positions.

Despite the fact that the standard allows x32 lines per port, such solutions are physically quite bulky and are not available.

Year release	Version PCI Express	Coding	Speed transfers	Bandwidth on x lines
				×1	×2	×4	×8	×16
2002	1.0	8b/10b	2.5 GT/s	2	4	8	16	32
2007	2.0	8b/10b	5 GT/s	4	8	16	32	64
2010	3.0	128b/130b	8 GT/s	~7,877	~15,754	~31,508	~63,015	~126,031
2017	4.0	128b/130b	16 GT/s	~15,754	~31,508	~63,015	~126,031	~252,062
2019	5.0	128b/130b	32 GT/s	~32	~64	~128	~256	~512

PCI Express 2.0

The PCI-SIG released the PCI Express 2.0 specification on January 15, 2007. Key innovations in PCI Express 2.0:

• Increased throughput: bandwidth of one line 500 MB/s, or 5 GT/s ( Gigatransactions/s).
• Improvements have been made to the transfer protocol between devices and the software model.
• Dynamic speed control (to control the communication speed).
• Bandwidth Alert (to notify software of changes in bus speed and width).
• Access Control Services - Optional point-to-point transaction management capabilities.
• Execution timeout control.
• Function level reset is an optional mechanism for resetting PCI functions within a PCI device.
• Redefining the power limit (to redefine the slot power limit when connecting devices that consume more power).

PCI Express 2.0 is fully compatible with PCI Express 1.1 (old ones will work in motherboards with new connectors, but only at a speed of 2.5 GT/s, since old chipsets cannot support double data transfer rates; new video adapters will work without problems in old PCI Express 1.x connectors).

PCI Express 2.1

In terms of physical characteristics (speed, connector) it corresponds to 2.0; in the software part, functions have been added that are planned to be fully implemented in version 3.0. Since most motherboards are sold with version 2.0, having only a video card with 2.1 does not allow you to use 2.1 mode.

PCI Express 3.0

In November 2010, the specifications for PCI Express 3.0 were approved. The interface has a data transfer rate of 8 GT/s ( Gigatransactions/s). But despite this, its actual throughput was still doubled compared to the PCI Express 2.0 standard. This was achieved thanks to a more aggressive 128b/130b encoding scheme, where 128 bits of data sent over the bus are encoded in 130 bits. At the same time, full compatibility with previous versions of PCI Express is maintained. PCI Express 1.x and 2.x cards will work in slot 3.0 and, conversely, a PCI Express 3.0 card will work in slots 1.x and 2.x.

PCI Express 4.0

The PCI Special Interest Group (PCI SIG) stated that PCI Express 4.0 could be standardized before the end of 2016, but in mid-2016, when a number of chips were already being prepared for production, media reported that standardization was expected in early 2017. will have a throughput of 16 GT/s, that is, it will be twice as fast as PCIe 3.0.

Leave your comment!

IntroductionIn the past, the mass consumer was primarily interested in only two types of SSDs: either high-speed premium models like the Samsung 850 PRO, or value-for-money offerings like the Crucial BX100 or SanDisk Ultra II. That is, the segmentation of the SSD market was extremely weak, and competition between manufacturers, although it developed in the areas of performance and price, the gap between top- and bottom-level solutions remained quite small. This state of affairs was partly due to the fact that SSD technology itself significantly improves the user experience of working with a computer, and therefore issues of specific implementation fade into the background for many. For the same reason, consumer SSDs were fit into the old infrastructure, which was originally focused on mechanical hard disks. This greatly facilitated their implementation, but placed SSDs within a fairly narrow framework, which largely restrained both the growth of throughput and the reduction in latency of the disk subsystem.

But until a certain time, this state of affairs suited everyone. SSD technology was new, and users migrating to SSDs were happy with their purchases even though they were essentially getting products that didn't actually perform at their best, with performance being held back by artificial barriers. However, by today, SSDs can perhaps be considered truly mainstream. Any self-respecting owner of a personal computer, if he does not have at least one SSD in his system, is very serious about purchasing one in the very near future. And in these conditions, manufacturers are simply forced to think about how to finally develop full-fledged competition: to destroy all barriers and move on to producing wider product lines that are fundamentally different in the characteristics offered. Fortunately, all the necessary ground has been prepared for this, and, first of all, most SSD developers have the desire and opportunity to start producing products that work not through the legacy SATA interface, but through the much more productive PCI Express bus.

Since SATA bandwidth is limited to 6 Gb/s, the maximum speed of flagship SATA SSDs does not exceed about 500 MB/s. However, modern flash memory-based drives are capable of much more: after all, if you think about it, they have more in common with system memory than with mechanical hard drives. As for the PCI Express bus, it is now actively used as a transport layer when connecting graphic cards and other additional controllers that require high-speed data exchange, for example, Thunderbolt. A single Gen 2 PCI Express lane provides 500 MB/s of bandwidth, while a PCI Express 3.0 lane can reach speeds of up to 985 MB/s. Thus, an interface card installed in a PCIe x4 slot (with four lanes) can exchange data at speeds of up to 2 GB/s in the case of PCI Express 2.0 and up to almost 4 GB/s when using PCI Express third generation. These are excellent indicators that are quite suitable for modern solid-state drives.

From the above, it naturally follows that in addition to SATA SSDs, high-speed drives using the PCI Express bus should gradually become widespread on the market. And this is really happening. In stores you can find several models of consumer SSDs from leading manufacturers, made in the form of expansion cards or M.2 cards that use different versions of the PCI Express bus. We decided to put them together and compare them in terms of performance and other parameters.

Test participants

Intel SSD 750 400 GB

In the solid-state drive market, Intel adheres to a rather unconventional strategy and does not pay too much attention to the development of SSDs for the consumer segment, concentrating on products for servers. However, this does not make her proposals uninteresting, especially when it comes to a solid-state drive for the PCI Express bus. In this case, Intel decided to adapt its most advanced server platform for use in a high-performance client SSD. This is exactly how the Intel SSD 750 400 GB was born, which received not only impressive performance characteristics and a number of server-level technologies responsible for reliability, but also support for the newfangled NVMe interface, about which a few words should be said separately.

If we talk about specific improvements to NVMe, then the reduction in overhead costs deserves mention first. For example, sending the most common 4K blocks in the new protocol requires issuing only one command instead of two. And the entire set of control instructions has been simplified so much that their processing at the driver level reduces the processor load and the resulting delays by at least half. The second important innovation is support for deep pipelining and multitasking, which consists in the ability to create multiple request queues in parallel instead of the previously existing single queue for 32 commands. The NVMe interface protocol is capable of servicing up to 65536 queues, and each of them can contain up to 65536 commands. In fact, any restrictions are eliminated altogether, and this is very important for server environments where the disk subsystem may be subject to a huge number of simultaneous I/O operations.

But despite working through the NVMe interface, the Intel SSD 750 is still not a server drive, but a consumer drive. Yes, almost the same hardware platform as in this drive is used in server-class SSDs Intel DC P3500, P3600 and P3700, but the Intel SSD 750 uses cheaper ordinary MLC NAND, and in addition the firmware is modified. The manufacturer believes that thanks to such changes, the resulting product will appeal to enthusiasts, since it combines high power, fundamentally new interface NVMe and not too scary cost.

The Intel SSD 750 is a half-height PCIe x4 card that can use four 3.0 lanes and achieve sequential transfer rates of up to 2.4 GB/s and random operation speeds of up to 440 thousand IOPS. True, the most capacious 1.2 TB modification has the highest performance, but the 400 GB version we received for testing is a little slower.

The drive board is completely covered with armor. On the front side it is an aluminum radiator, and on the back side there is a decorative metal plate that does not actually come into contact with the microcircuits. It should be noted that the use of a radiator here is a necessity. The main controller of an Intel SSD generates a lot of heat, and under high load, even a drive equipped with such cooling can heat up to temperatures of about 50-55 degrees. But thanks to the pre-installed cooling, there is no hint of throttling - performance remains constant even during continuous and intensive use.

The Intel SSD 750 is based on a server controller Intel level CH29AE41AB0, which operates at a frequency of 400 MHz and has eighteen (!) channels for connecting flash memory. When you consider that most consumer SSD controllers have either eight or four channels, it becomes clear that the Intel SSD 750 can actually pump significantly more data across the bus than conventional SSD models.

As for the flash memory used, the Intel SSD 750 does not make any innovations in this area. It is based on regular Intel-made MLC NAND, produced using a 20-nm process technology and having cores with a volume of both 64 and 128 Gbit interspersed. It should be noted that most other SSD manufacturers abandoned such memory quite a long time ago, switching to chips made to thinner standards. And Intel itself has begun converting not only its consumer, but also server drives to 16nm memory. However, despite all this, the Intel SSD 750 is equipped with older memory, which supposedly has a higher resource.

The server origin of the Intel SSD 750 can also be traced in the fact that the total amount of flash memory in this SSD is 480 GiB, of which only about 78 percent is available to the user. The rest is allocated to the replacement fund, garbage collection and data protection technologies. The Intel SSD 750 implements a RAID 5-like scheme, traditional for flagship drives, at the MLC NAND chip level, which allows you to successfully restore data even if one of the chips completely fails. In addition, the Intel SSD provides full protection data from power failures. The Intel SSD 750 has two electrolytic capacitors, and their capacity is sufficient for normal shutdown of the drive in offline mode.

Kingston HyperX Predator 480 GB

Kingston HyperX Predator is a much more traditional solution compared to the Intel SSD 750. Firstly, it works via the AHCI protocol, not NVMe, and secondly, this SSD requires the more common PCI Express 2.0 bus to connect to the system. All this makes the Kingston version somewhat slower - peak speeds for sequential operations do not exceed 1400 MB/s, and random ones - 160 thousand IOPS. But HyperX Predator does not impose any special requirements on the system - it is compatible with any, including older platforms.

At the same time, the drive has a not entirely simple two-component design. The SSD itself is a board in the M.2 form factor, which is complemented by a PCI Express adapter that allows you to connect M.2 drives through regular full-size PCIe slots. The adapter is designed as a half-height PCIe x4 card that uses all four PCI Express lanes. Thanks to this design, Kingston sells its HyperX Predator in two versions: as a PCIe SSD for desktops and as an M.2 drive for mobile systems (in this case, the adapter is not included in the delivery).

Kingston HyperX Predator is based on the Marvell Altaplus controller (88SS9293), which, on the one hand, supports four PCI Express 2.0 lanes, and on the other, has eight channels for connecting flash memory. On this moment This is Marvell's fastest commercially available SSD controller with PCI Express support. However, Marvell will soon have faster successors with support for NVMe and PCI Express 3.0, which the Altaplus chip does not have.

Because she herself Kingston company does not produce either controllers or memory, assembling its SSDs from the element base purchased from other manufacturers, there is nothing strange in the fact that the HyperX Predator PCIe SSD is based not only on a third-party controller, but also on 128-gigabit 19-nm MLC NAND chips Toshiba company. Such memory has a low purchase price and is now installed in many products from Kingston (and other companies), and primarily in consumer models.

However, the use of such memory has given rise to a paradox: despite the fact that, according to its formal positioning, the Kingston HyperX Predator PCIe SSD is a premium product, it only comes with a three-year warranty, and the stated mean time between failures is significantly less than that of flagship SATA SSDs other manufacturers.

Kingston HyperX Predator also does not provide any special data protection technologies. But the drive has a relatively large area hidden from the user's eyes, the size of which is 13 percent of the total capacity of the drive. The backup flash memory included in it is used for garbage collection and wear leveling, but is primarily spent on replacing failed memory cells.

It remains only to add that the HyperX Predator design does not provide any special means to remove heat from the controller. Unlike most other high-performance solutions, this drive does not have a heatsink. However, this SSD is not at all prone to overheating - its maximum heat dissipation is only slightly higher than 8 W.

OCZ Revodrive 350 480 GB

The OCZ Revodrive 350 can rightfully be called one of the oldest consumer SSDs with a PCI Express interface. Back in the days when none of the other manufacturers even thought about releasing client PCIe SSDs, in model range OCZ company had RevoDrive 3 (X2) - the prototype of the modern Revodrive 350. However, the roots of the OCZ PCIe drive, which go back to the past, make it a somewhat strange proposal compared to current competitors. While most manufacturers of high-performance PC drives use modern controllers with native support for the PCI Express bus, the Revodrive 350 implements a very intricate and clearly suboptimal architecture. It is based on two or four (depending on the volume) SandForce SF-2200 controllers, which are assembled into a zero-level RAID array.

If we talk about the OCZ Revodrive 350 480 GB model that took part in this testing, then it is actually based on four SATA SSDs with a capacity of 120 GB, each of which is based on its own SF-2282 chip (analogue of the widely used SF-2281) . These elements are then combined into a single four-part RAID 0 array. However, for this purpose, not a very familiar RAID controller is used, but a proprietary virtualization processor (VCA 2.0) OCZ ICT-0262. However, it is very likely that this name hides a redesigned Marvell 88SE9548 chip, which is a four-port SAS/SATA 6 Gb/s RAID controller with a PCI Express 2.0 x8 interface. But even if so, OCZ engineers wrote their own firmware and driver for this controller.

The uniqueness of the RevoDrive 350 software component lies in the fact that it implements not quite the classic RAID 0, but something similar to it with interactive load balancing. Instead of breaking the data stream into fixed-size blocks and sequentially transmitting them to different SF-2282 controllers, VCA 2.0 technology involves analysis and flexible redistribution of I/O operations depending on the current occupancy of flash memory controllers. Therefore, the RevoDrive 350 looks like a monolithic SSD to the user. It is impossible to enter its BIOS, and it is impossible to discover that a RAID array is hidden in the depths of this SSD without a detailed acquaintance with the hardware. Moreover, unlike conventional RAID arrays, RevoDrive 350 supports all typical SSD functions: SMART monitoring, TRIM and Secure Erase operation.

RevoDrive 350 is available in the form of boards with PCI Express 2.0 x8 interface. Despite the fact that all eight interface lines are actually used, the stated performance figures are noticeably lower than their total theoretical throughput. The maximum speed of sequential operations is limited to 1800 MB/s, and the performance of random operations does not exceed 140 thousand IOPS.

It is worth noting that the OCZ RevoDrive 350 is made as a full-height PCI Express x8 board, that is, this drive is physically larger than all the other SSDs participating in testing, and therefore it cannot be installed in low-profile systems. The front surface of the RevoDrive 350 board is covered with a decorative metal casing, which also acts as a radiator for the base RAID controller chip. The SF-2282 controllers are located on the reverse side of the board and do not have any cooling.

To form the flash memory array, OCZ used chips from its parent company, Toshiba. Chips produced using a 19-nm process technology and having a capacity of 64 Gbit are used. The total amount of flash memory in the RevoDrive 350 480 GB is 512 GB, but 13% is reserved for internal needs - wear leveling and garbage collection.

It is worth noting that the architecture of the RevoDrive 350 is not unique. There are several more models of similar SSDs on the market, operating on the principle of a “RAID array of SATA SSDs based on SandForce controllers.” However, all such solutions, like the OCZ PCIe drive under consideration, have an unpleasant drawback - their performance on write operations degrades over time. This is due to the peculiarities of the internal algorithms of SandForce controllers, the TRIM operation of which does not return the write speed to the original level.

The indisputable fact that the RevoDrive 350 is one step lower than the PCI Express drives of the new generation is emphasized by the fact that this drive has only a three-year warranty, and its guaranteed recording resource is only 54 TB - several times less than that of its competitors. Moreover, despite the fact that RevoDrive 350 is based on the same design as the server Z-Drive 4500, it does not have any protection against power surges. However, all this does not prevent OCZ, with its characteristic audacity, from positioning the RevoDrive 350 as a premium solution at the Intel SSD 750 level.

Plextor M6e Black Edition 256 GB

It should be immediately noted that the Plextor M6e Black Edition drive is a direct successor to the well-known M6e model. The similarity of the new product to its predecessor can be seen in almost everything, if we talk about the technical rather than the aesthetic component. The new SSD also has a two-component design, including the drive itself in the M.2 2280 format and an adapter that allows you to install it in any regular PCIe x4 (or faster) slot. It is also based on an eight-channel Marvell 88SS9183 controller, which communicates with the outside world via two PCI Express 2.0 lines. Just like the previous modification, the M6e Black Edition uses Toshiba MLC flash memory.

This means that while the M6e Black Edition looks like a half-height PCI Express x4 card when assembled, this SSD actually only uses two PCI Express 2.0 lanes. Hence the not very impressive speeds, which are only slightly higher than the performance of traditional SATA SSDs. The nominal performance for sequential operations is limited to 770 MB/s, and for arbitrary operations – 105 thousand IOPS. It is worth noting that Plextor M6e Black Edition operates using the legacy AHCI protocol, and this ensures its wide compatibility with various systems.

Despite the fact that the Plextor M6e Black Edition, like the Kingston HyperX Predator, is a combination of a PCI Express adapter and a “core” in M.2 card format, it is impossible to determine this from the front side. The entire drive is hidden under a figured black aluminum casing, in the center of which there is a red radiator embedded, which should remove heat from the controller and memory chips. The designers’ calculation is clear: a similar color scheme is widely used in various gaming hardware, so the Plextor M6e Black Edition will look harmonious next to many gaming motherboards and video cards from most leading manufacturers.

The flash memory array in the Plextor M6e Black Edition is equipped with Toshiba's second-generation 19-nm MLC NAND chips with a capacity of 64 Gbit. The reserve used for the replacement fund and the operation of internal algorithms for leveling wear and garbage collection is allocated 7 percent of the total volume. Everything else is available to the user.

Due to the use of a rather weak Marvell 88SS9183 controller with an external PCI Express 2.0 x2 bus, the Plextor M6e Black Edition drive should be considered a rather slow PCIe SSD. However, this does not prevent the manufacturer from classifying this product in the upper price category. On the one hand, it is still faster than a SATA SSD, and on the other, it has good reliability characteristics: it has a long MTBF and is covered by a five-year warranty. However, no special technologies that can protect the M6e Black Edition from voltage surges or increase its service life are implemented in it.

Samsung SM951 256 GB

The Samsung SM951 is the most elusive drive in today's testing. The fact is that initially this is a product for computer assemblers, so it is presented in retail sales rather poorly. However, if you wish, it is still possible to buy it, so we did not refuse to consider the SM951. Moreover, judging by the characteristics, this is a very fast-acting model. It is designed to work on the PCI Express 3.0 x4 bus, uses the AHCI protocol and promises impressive speeds: up to 2150 MB/s for sequential operations and up to 90 thousand IOPS for random operations. But most importantly, with all this, the Samsung SM951 is cheaper than many other PCIe SSDs, so its search for sale may have a very specific economic justification.

Another feature of the Samsung SM951 is that it comes in M.2 format. Initially this solution is aimed at mobile systems, so no adapters for full-size PCIe slots are included with the drive. However, this can hardly be considered a serious drawback - most flagship motherboards also have M.2 interface slots on board. In addition, the necessary adapter boards are widely available for sale. The Samsung SM951 itself is a board of the M.2 2280 form factor, the connector of which has an M type key, indicating the need for an SSD with four PCI Express lines.

The Samsung SM951 is based on an exceptionally powerful Samsung UBX controller, developed by the manufacturer specifically for SSDs with a PCI Express interface. It is based on three cores with ARM architecture and, in theory, is capable of working with both AHCI and NVMe commands. In the SSD in question, only the AHCI mode is enabled in the controller. But the NVMe version of this controller could soon be seen in a new consumer SSD that Samsung is due to launch this fall.

Due to the OEM focus, neither the warranty period nor the predicted endurance are provided for the drive in question. Builders of systems into which the SM951 will be installed, or sellers must declare these parameters. However, it should be noted that 3D V-NAND, which is now actively promoted by Samsung in consumer SSDs as a faster and more reliable type of flash memory, is not used in the SM951. Instead, it uses conventional planar Toggle Mode 2.0 MLC NAND, presumably produced using 16nm technology (some sources suggest a 19nm process technology). This means that the SM951 should not be expected to have the same high endurance as the flagship SATA 850 PRO drive. In this parameter, the SM951 is closer to conventional mid-level models; moreover, only 7 percent of the flash memory array is allocated for redundancy in this SSD. The Samsung SM951 does not have any special server-level technologies to protect data from power failures. In other words, the emphasis in this model is solely on speed, and everything else is cut off to reduce cost.

One more point is worth noting. Under high load, the Samsung SM951 exhibits quite serious heating, which ultimately can even lead to throttling. Therefore, in high-performance systems, it is advisable to organize at least airflow for the SM951, or better yet, cover it with a radiator.

Comparative characteristics of tested SSDs

Compatibility issues

Like any new technology, solid-state drives with a PCI Express interface cannot yet boast of 100% trouble-free operation with any platform, especially older ones. Therefore, you have to choose a suitable SSD not only based on consumer characteristics, but also with an eye to compatibility. And here it is important to keep two points in mind.

First of all, different SSDs can use different numbers of PCI Express lanes and different generations this tire is 2.0 or 3.0. Therefore, before purchasing a PCIe drive, you need to make sure that the system where you plan to install it has a free slot with the required bandwidth. Of course, faster PCIe SSDs are backwards compatible with slow slots, but in this case, purchasing a high-speed SSD does not make too much sense - it simply will not be able to unleash its full potential.

The Plextor M6e Black Edition has the widest compatibility in this sense - it requires only two PCI Express 2.0 lanes, and such a free slot will probably be found on almost any motherboard. The Kingston HyperX Predator already requires four PCI Express 2.0 lanes: many boards also have such PCIe slots, but some cheap platforms may not have extra slots with four or more PCI Express lanes. This is especially true for motherboards built on lower-level chipsets, the total number of lines of which can be reduced to six. Therefore, before purchasing a Kingston HyperX Predator, be sure to check that the system has a free slot with four or more PCI Express lanes.

OCZ Revodrive 350 poses a more difficult problem - it already requires eight PCI Express lanes. Such slots are usually implemented not by the chipset, but by the processor. Therefore, the optimal place for using such a drive is LGA 2011/2011-3 platforms, where the PCI Express processor controller has an excess number of lanes, allowing it to service more than one video card. In systems with LGA 1155/1150/1151 processors, the OCZ Revodrive 350 will be appropriate only if the graphics built into the CPU are used. Otherwise, in favor of the solid-state drive, you will have to take away half of the lines from the GPU, switching it to PCI Express x8 mode.

Intel SSD 750 and Samsung SM951 are somewhat similar to the OCZ Revodrive 350: they are also preferable to use in PCI Express slots powered by the processor. However, the reason here is not the number of lanes - they only require four PCI Express lanes, but the generation of this interface: both of these drives are capable of using the increased bandwidth of PCI Express 3.0. However, there is an exception: the latest Intel chipsets of the 100th series, designed for processors of the Skylake family, have received support for PCI Express 3.0, so in the latest LGA 1151 boards they can be installed without a twinge of conscience in chipset PCIe slots, to which at least four lines.

There is a second part to the compatibility problem. In addition to all the restrictions associated with the throughput of various variations of PCI Express slots, there are also restrictions associated with the protocols used. The most problem-free in this sense are SSDs that operate via AHCI. Due to the fact that they emulate the behavior of a regular SATA controller, they can work with any, even old, platforms: they are seen in the BIOS of any motherboards, they can be boot disks, and for their operation in the operating system no additional drivers are required. In other words, Kingston HyperX Predator and Plextor M6e Black Edition are two of the most hassle-free PCIe SSDs.

What about the other pair of AHCI drives? The situation with them is a little more complicated. The OCZ Revodrive 350 runs in the operating system through its own driver, but even despite this, there are no problems with making this drive bootable. The situation is worse with Samsung SM951. Although this SSD communicates with the system via the legacy AHCI protocol, it does not have its own BIOS, and therefore must be initialized Motherboard BIOS fees. Unfortunately, not all motherboards, especially old ones, support this SSD. Therefore, we can only speak with complete confidence about its compatibility with boards based on the latest Intel chipsets of the 90th and 100th series. In other cases, he may simply not be seen motherboard. Of course, this will not prevent you from using the Samsung SM951 in an operating system where it is easily initialized by the AHCI driver, but in this case you will have to forget about the possibility of booting from a high-speed SSD.

But the biggest inconvenience can be caused by the Intel SSD 750, which operates via the new NVMe interface. The drivers required to support SSDs that use this protocol are only available on the latest operating systems. Thus, in Linux, NVMe support appeared in kernel version 3.1; The “innate” NVMe driver is available in Microsoft systems, starting with Windows 8.1 and Windows Server 2012 R2; and in OS X, compatibility with NVMe drives was added in version 10.10.3. In addition, NVMe SSD is not supported by all motherboards. In order for such drives to be used as boot drives, the motherboard BIOS must also have the appropriate driver. However, manufacturers have built the necessary functionality only into the most latest versions firmware released for the latest motherboard models. Therefore, download support operating system with NVMe drives are only available on the most modern boards for enthusiasts, based on kits Intel logic Z97, Z170 and X99. In older and cheaper platforms, users will be able to use NVMe SSDs only as second drives in a limited set of OSes.

Despite the fact that we tried to describe all possible combinations of platforms and PCI Express drives, the main conclusion from the above is this: the compatibility of PCIe SSDs with motherboards is not as obvious a question as in the case of SATA SSDs. Therefore, before purchasing any high-speed solid-state drive that operates via PCI Express, be sure to check its compatibility with a specific motherboard on the manufacturer’s website.

Test configuration, tools and testing methodology

Testing is carried out in the operating room Microsoft system Windows 8.1 Professional x64 with Update, which correctly recognizes and services modern solid-state drives. This means that during the testing process, as in normal everyday use of the SSD, the TRIM command is supported and actively used. Performance measurements are performed with drives in a “used” state, which is achieved by pre-filling them with data. Before each test, the drives are cleaned and maintained using the TRIM command. There is a 15-minute pause between individual tests, allotted for the correct development of garbage collection technology. All tests use randomized, incompressible data unless otherwise noted.

Applications and tests used:

Iometer 1.1.0

Measuring the speed of sequential reading and writing data in blocks of 256 KB (the most typical block size for sequential operations in desktop tasks). The speeds are estimated within a minute, after which the average is calculated.
Measuring the speed of random reading and writing in 4 KB blocks (this block size is used in the vast majority of real-life operations). The test is carried out twice - without a request queue and with a request queue with a depth of 4 commands (typical for desktop applications that actively work with a branched file system). Data blocks are aligned relative to the flash memory pages of the drives. The speed assessment is performed for three minutes, after which the average is calculated.
Establishing the dependence of random read and write speeds when operating a drive with 4 KB blocks on the depth of the request queue (ranging from one to 32 commands). Data blocks are aligned relative to the flash memory pages of the drives. The speed assessment is performed for three minutes, after which the average is calculated.
Establishing the dependence of random read and write speeds when the drive operates with blocks of different sizes. Blocks ranging in size from 512 bytes to 256 KB are used. The request queue depth during the test is 4 commands. Data blocks are aligned relative to the flash memory pages of the drives. The speed assessment is performed for three minutes, after which the average is calculated.
Measuring performance under mixed multi-threaded workloads and determining its dependence on the ratio between read and write operations. The test is carried out twice: for sequential reads and writes in 128 KB blocks, executed in two independent threads, and for random operations with 4 KB blocks, executed in four threads. In both cases, the ratio between read and write operations varies in 20 percent increments. The speed assessment is performed for three minutes, after which the average is calculated.
Study of SSD performance degradation when processing a continuous stream of random write operations. Blocks of 4 KB in size and a queue depth of 32 commands are used. Data blocks are aligned relative to the flash memory pages of the drives. The test duration is two hours, instantaneous speed measurements are carried out every second. At the end of the test, the ability of the drive to restore its performance to its original values ​​is additionally checked due to the operation of garbage collection technology and after running the TRIM command.

CrystalDiskMark 5.0.2
A synthetic test that provides typical performance indicators for solid-state drives, measured on a 1-gigabyte disk area “on top” file system. Of the entire set of parameters that can be assessed using this utility, we pay attention to the speed of sequential read and write, as well as the performance of random read and write of 4 KB blocks without a request queue and with a queue depth of 32 commands.
PCMark 8 2.0
A test based on emulating a real disk load, which is typical for various popular applications. On the drive being tested, a single partition is created in the file NTFS system for the entire available capacity, and PCMark 8 runs the Secondary Storage test. The test results take into account both the final performance and the execution speed of individual test traces generated by various applications.
File copy tests
This test measures the speed of copying file directories different types, as well as the speed of archiving and unarchiving files inside the drive. For copying, use the standard Windows tool– Robocopy utility, when archiving and unzipping – 7-zip archiver version 9.22 beta. The tests involve three sets of files: ISO – a set that includes several disk images with program distributions; Program – a set that is a pre-installed software package; Work – a set of work files, including office documents, photographs and illustrations, pdf files and multimedia content. Each set has a total file size of 8 GB.

A computer with a motherboard is used as a test platform ASUS board Z97-Pro Core processor i5-4690K with integrated graphics Intel core HD Graphics 4600 and 16 GB DDR3-2133 SDRAM. Drives with a SATA interface connect to the SATA 6 Gb/s controller built into the motherboard chipset and operate in AHCI mode. Drives with a PCI Express interface are installed in the first full-speed PCI Express 3.0 x16 slot. The drivers used are Intel Rapid Storage Technology (RST) 13.5.2.1000 and Intel Windows NVMe driver 1.2.0.1002.

The volume and speed of data transfer in benchmarks are indicated in binary units (1 KB = 1024 bytes).

In addition to the five main heroes of this test - client SSDs with a PCI Express interface, we also added the fastest SATA SSD - Samsung 850 PRO.

As a result, the list of tested models took the following form:

Intel SSD 750 400 GB (SSDPEDMW400G4, firmware 8EV10135);
Kingston HyperX Predator PCIe 480 GB (SHPM2280P2H/480G, firmware OC34L5TA);
OCZ RevoDrive 350 480 GB (RVD350-FHPX28-480G, firmware 2.50);
Plextor M6e Black Edition 256 GB (PX-256M6e-BK, firmware 1.05);
Samsung 850 Pro 256 GB (MZ-7KE256, firmware EXM01B6Q);
Samsung SM951 256 GB (MZHPV256HDGL-00000, firmware BXW2500Q).

Performance

Sequential reads and writes

The new generation of solid-state drives, transferred to the PCI Express bus, should primarily be distinguished by high sequential read and write speeds. And this is exactly what we see on the graph. All PCIe SSDs turn out to be more productive than the best SATA SSD – Samsung 850 PRO. However, even something as simple as sequential reads and writes shows huge differences between SSDs from different manufacturers. Moreover, the version of the PCI Express bus used is not decisive. The best performance here can be achieved by the PCI Express 3.0 x4 drive of the Samsung SM951, and in second place is the Kingston HyperX Predator, working via PCI Express 2.0 x4. The progressive NVMe drive Intel SSD 750 was only in third place.

Random reads

If we talk about random reading, then, as can be seen from the diagrams, PCIe SSDs are not particularly different in speed from traditional SATA SSDs. Moreover, this applies not only to AHCI drives, but also to the product that works with the NVMe channel. Actually better than Samsung 850 PRO performance Only three participants in this test can demonstrate random read operations on small request queues: Samsung SM951, Intel SSD 750 and Kingston HyperX Predator.

Although deep query queue operations for personal computers are not typical, we will still look at how the performance of the SSD in question depends on the depth of the request queue when reading 4-kilobyte blocks.

The graph clearly shows how solutions running via PCI Express 3.0 x4 can outperform all other SSDs. The curves corresponding to the Samsung SM951 and Intel SSD 750 are significantly higher than the graphs of other drives. Based on the above diagram, one more conclusion can be drawn: the OCZ RevoDrive 350 is a shamefully slow solid-state drive. In random read operations, it is about half as good as a SATA SSD, which is due to its RAID architecture and the use of outdated second-generation SandForce controllers.

In addition to this, we suggest looking at how the random read speed depends on the size of the data block:

Here the picture is a little different. As the block size increases, operations begin to resemble sequential ones, so not only the architecture and power of the SSD controller begins to play a role, but also the bandwidth of the bus they use. On large blocks better performance provide Samsung SM951, Intel SSD 750 and Kingston HyperX Predator.

Random writes

Somewhere, the benefits of the low-latency NVMe interface and the high-parallel Intel SSD 750 controller had to show up. In addition, the large DRAM buffer available in this SSD allows for very efficient data caching. As a result, the Intel SSD 750 delivers unmatched random write speeds even when the request queue is minimal.

You can see more clearly what happens to random write performance as the request queue depth increases at next schedule, showing the dependence of the speed of random writing in 4-kilobyte blocks on the depth of the request queue:

The performance of the Intel SSD 750 scales until the queue depth reaches 8 commands. This is typical behavior for consumer SSDs. However, Intel's new product is different in that its random write speeds are significantly higher than any other solid-state drives, including the fastest PCIe models like the Samsung SM951 or Kingston HyperX Predator. In other words, for occasional write workloads, the Intel SSD 750 offers fundamentally better performance than any other SSD. In other words, switching to the NVMe interface allows you to improve the random write speed. And this is certainly an important characteristic, but primarily for server drives. Actually, the Intel SSD 750 is precisely a close relative of such models as the Intel DC P3500, P3600 and P3700.

The following graph shows random write performance as a function of data block size.

As block sizes increase, the Intel SSD 750 loses its unconditional advantage. Samsung SM951 and Kingston HyperX Predator are starting to produce approximately the same performance.


As SSDs become cheaper, they are no longer used as purely system drives and are becoming regular work drives. In such situations, the SSD receives not only a refined load in the form of writing or reading, but also mixed requests, when read and write operations are initiated by different applications and must be processed simultaneously. However, full-duplex operation remains a significant problem for modern SSD controllers. When mixing reads and writes in the same queue, the speed of most consumer-grade SSDs noticeably drops. This became the reason for conducting a separate study, in which we check how SSDs work when it is necessary to process sequential operations arriving interspersed. The next couple of charts show the most typical case for desktops, where the ratio of read to write operations is 4 to 1.

With a sequential mixed load with predominant read operations, which is typical for conventional personal computers, the Samsung SM951 and Kingston HyperX Predator provide the best performance. A random mixed load turns out to be a more difficult test for SSDs and leaves the Samsung SM951 in the lead, but the Intel SSD 750 moves into second place. At the same time, the Plextor M6e Black Edition, Kingston HyperX Predator and OCZ RevoDrive 350 generally turn out to be noticeably worse than a regular SATA SSD.

The next pair of graphs gives a more detailed picture of performance under mixed loads, showing the dependence of SSD speed on the ratio of read and write operations on it.

Everything said above is well confirmed by the above graphs. With a mixed load with sequential operations, the best performance is shown by the Samsung SM951, which feels like a fish in water when working with any serial data. For arbitrary mixed operations the situation is slightly different. Both Samsung drives, the SM951 running via PCI Express 3.0 x4, and the regular SATA 850 PRO, give very good results in this test, outperforming almost all other SSDs. In some cases, only the Intel SSD 750 can resist them, which, thanks to the NVMe command system, is perfectly optimized for working with random writes. And when the share of records in the mixed transaction flow increases to 80 percent or higher, it leaps ahead.

Results in CrystalDiskMark

CrystalDiskMark is a popular and simple benchmark application that runs on top of the file system and produces results that are easily repeatable by ordinary users. The performance indicators obtained in it should complement the detailed graphs we built based on tests in IOMeter.

The four diagrams shown are of theoretical value only, showing peak performance that is not achievable in typical client workloads. There is never a request queue depth of 32 commands in personal computers, but in special tests it allows you to get maximum performance indicators. And in this case, the leading performance by a large margin is given by the Intel SSD 750, which has an architecture inherited from server drives, where a large request queue depth is quite normal.

But these four diagrams are of practical interest - they display performance under load, which is typical for personal computers. And here the best performance is given by the Samsung SM951, which lags behind the Intel SSD 750 only with random 4 KB writes.

PCMark 8 2.0, real use cases

The Futuremark PCMark 8 2.0 test package is interesting because it is not of a synthetic nature, but, on the contrary, is based on how real applications work. During its passage, real scenarios-traces of using the disk in common desktop tasks are reproduced, and the speed of their execution is measured. The current version of this test simulates workloads that are taken from real-life gaming applications of Battlefield 3 and World of Warcraft and software packages from Abobe and Microsoft: After Effects, Illustrator, InDesign, Photoshop, Excel, PowerPoint and Word. The final result is calculated in the form of the average speed that the drives show when passing test routes.

The PCMark 8 2.0 test, which evaluates the performance of storage systems in real applications, clearly tells us that there are only two PCIe drives, the speed of which is fundamentally higher than that of conventional models with a SATA interface. These are Samsung SM951 and Intel SSD 750, which win in many other tests. Other PCIe SSDs, for example, Plextor M6e Black Edition and Kingston HyperX Predator, lag behind the leaders by more than one and a half times. Well, the OCZ ReveDrive 350 demonstrates frankly poor performance. It is more than twice as slow as the best PCIe SSDs and is even slower than the Samsung 850 PRO, which operates via a SATA interface.

The integral result of PCMark 8 must be supplemented with performance indicators produced by flash drives when passing individual test traces that simulate various real-life load options. The fact is that under different loads, flash drives often behave slightly differently.

Whatever the application we are talking about, in any case, the highest performance is provided by one of the SSDs with a PCI Express 3.0 x4 interface: either Samsung SM951 or Intel SSD 750. Interestingly, other PCIe SSDs in some cases generally only provide speeds at the level of SATA SSDs . In fact, the advantage of the same Kingston HyperX Predator and Plextor M6e Black Edition over the Samsung 850 PRO can only be seen in Adobe Photoshop, Battlefield 3 and Microsoft Word.

Copying files

Keeping in mind that solid-state drives are being introduced into personal computers more and more widely, we decided to add to our methodology a measurement of performance during common file operations - when copying and working with archivers - which are performed “inside” the drive. This is a typical disk activity that occurs when the SSD acts not as a system drive, but as a regular disk.

In the copying tests, the leaders are still the same Samsung SM951 and Intel SSD 750. However, if we are talking about large sequential files, then the Kingston HyperX Predator can compete with them. I must say that with simple copying, almost all PCIe SSDs turn out to be faster than the Samsung 850 PRO. There is only one exception - Plextor M6e Black Edition. And the OCZ RevoDrive 350, which in other tests consistently found itself in the position of a hopeless outsider, unexpectedly outperforms not only the SATA SSD, but also the slowest PCIe SSD.

The second group of tests was carried out when archiving and unarchiving a directory with working files. The fundamental difference in this case is that half of the operations are performed with separate files, and the second half with one large archive file.

The situation is similar when working with archives. The only difference is that here the Samsung SM951 manages to confidently break away from all its competitors.

How TRIM and Background Garbage Collection Work

When testing various SSDs, we always check how they handle the TRIM command and whether they are able to collect garbage and restore their performance without support from the operating system, that is, in a situation where the TRIM command is not issued. Such testing was carried out this time as well. The design of this test is standard: after creating a long continuous load on writing data, which leads to write speed degradation, we disable TRIM support and wait 15 minutes, during which the SSD can try to recover on its own using its own garbage collection algorithm, but without outside help operating system, and measure the speed. Then the TRIM command is forced onto the drive - and after a short pause, the speed is measured again.

The results of this testing are shown in the following table, which shows for each model tested whether it responds to TRIM by clearing unused flash memory and whether it can procure clean flash memory pages for future operations if a TRIM command is not issued to it. For drives that were able to perform garbage collection without the TRIM command, we also indicated the amount of flash memory that was independently freed by the SSD controller for future operations. If the drive is used in an environment without TRIM support, this is exactly the amount of data that can be saved to the drive with a high initial speed after inactivity.

Despite the fact that high-quality support for the TRIM command has become an industry standard, some manufacturers consider it acceptable to sell drives that do not fully implement this command. Such a negative example is demonstrated by the OCZ Revodrive 350. Formally, it understands TRIM, and even tries to do something when receiving this command, but there is no talk of a complete return of the write speed to its original values. And there is nothing strange about this: the Revodrive 350 is based on SandForce controllers, which are distinguished by their irreversible performance degradation. Accordingly, it is also present in Revodrive 350.

All other PCIe SSDs work with TRIM just like their SATA counterparts. That is, ideal: in operating systems that issue this command to drives, performance remains at a consistently high level.

However, we want more - a high-quality drive should be able to perform garbage collection without issuing the TRIM command. And here the Plextor M6e Black Edition stands out - a drive that can independently free up significantly more flash memory for upcoming operations than its competitors. Although, of course, to one degree or another, autonomous garbage collection works for all SSDs we tested, with the exception of the Samsung SM951. In other words, during normal use in modern environments The performance of the Samsung SM951 will not degrade, however, in cases where TRIM is not supported, using this SSD is not recommended.

conclusions

We should probably start summarizing the results by stating the fact that consumer SSDs with the PCI Express interface are no longer exotic or some experimental products, but an entire market segment in which the fastest performing solid-state drives for enthusiasts play. Naturally, this also means that there have been no problems with PCIe SSDs for a long time: they support all the functions that SATA SSDs have, but at the same time they are more productive and sometimes have some new interesting technologies.

At the same time, the client PCIe SSD market is not so crowded, and so far only companies with high engineering potential have been able to enter the cohort of manufacturers of such solid-state drives. This is due to the fact that independent developers of mass-produced SSD controllers do not yet have design solutions that allow them to begin producing PCIe drives with minimal engineering effort. Therefore, each of the PCIe SSDs currently presented on store shelves is original and unique in its own way.

In this testing, we were able to bring together the five most popular and most common PCIe SSDs, aimed at operation as part of personal computers. And based on the results of getting to know them, it becomes clear that buyers who want to switch to using solid-state drives with a progressive interface will not face any serious pains of choice yet. In most cases, the choice will be clear, the tested models differ so much in their consumer qualities.

Overall, the most attractive PCIe SSD model turned out to be Samsung SM951. This is a brilliant solution from one of the market leaders, operating over the PCI Express 3.0 x4 bus, which not only turns out to be able to provide the highest performance in typical common workloads, but is also significantly cheaper than all other PCIe drives.

However, the Samsung SM951 is still not perfect. Firstly, it does not contain any special technologies aimed at increasing reliability, but in premium-level products one would still like to have them. Secondly, this SSD is quite difficult to find for sale in Russia - it is not supplied to our country through official channels. Fortunately, we can suggest paying attention to a good alternative - Intel SSD 750. This SSD also runs via PCI Express 3.0 x4, and is only slightly behind the Samsung SM951. But it is a direct relative of server models, and therefore has high reliability and works using the NVMe protocol, which allows it to demonstrate unsurpassed speed in random write operations.

In principle, compared to the Samsung SM951 and Intel SSD 750, other SSDs with a PCIe interface look rather weak. However, there are still situations when they will have to prefer some other PCIe SSD model. The fact is that advanced Samsung and Intel drives are compatible only with modern motherboards built on Intel chipsets of the ninetieth or hundredth series. In older systems, they can only work as a “second disk”, and loading the operating system from them will be impossible. Therefore, neither the Samsung SM951 nor the Intel SSD 750 are suitable for upgrading platforms of previous generations, and the choice will have to be on the drive Kingston HyperX Predator, which, on the one hand, can provide good performance, and on the other, is guaranteed not to have any compatibility problems with older platforms.

I have been asked this question more than once, so now I will try to answer it as clearly and briefly as possible. To do this, I will provide pictures of the PCI Express and PCI expansion slots on the motherboard for a clearer understanding and, of course, I will indicate the main differences in the characteristics, i.e. .e. very soon you will find out what these interfaces are and what they look like.

So, first, let's briefly answer the question, what exactly is PCI Express and PCI?

What is PCI Express and PCI?

PCI is a computer parallel input/output bus for connecting peripheral devices to the computer motherboard. PCI is used to connect: video cards, sound cards, network cards, TV tuners and other devices. The PCI interface is outdated, so you probably won’t be able to find, for example, a modern video card that connects via PCI.

PCI Express(PCIe or PCI-E) is a computer serial bus I/O for connecting peripheral devices to the computer motherboard. Those. in this case, bidirectional serial connection, which can have several lines (x1, x2, x4, x8, x12, x16 and x32) the more such lines, the higher the throughput of the PCI-E bus. The PCI Express interface is used to connect devices such as video cards, sound cards, network cards, SSD drives and others.

There are several versions of the PCI-E interface: 1.0, 2.0 and 3.0 (version 4.0 will be released soon). This interface is usually designated, for example, like this PCI-E 3.0 x16, which means PCI Express 3.0 version with 16 lanes.

If we talk about whether, for example, a video card that has a PCI-E 3.0 interface will work on a motherboard that only supports PCI-E 2.0 or 1.0, the developers say that everything will work, just of course keep in mind that the bandwidth will be limited by the capabilities of the motherboard. Therefore, in this case, overpay for a video card with more new version PCI Express I think is not worth it ( if only for the future, i.e. Are you planning to purchase a new motherboard with PCI-E 3.0?). Also, and vice versa, let’s say you have motherboard supports version PCI Express 3.0, and the video card version, say, 1.0, then this configuration should also work, but only with PCI-E 1.0 capabilities, i.e. There is no limitation here, since the video card in this case will work at the limit of its capabilities.

Differences between PCI Express and PCI

The main difference in characteristics is, of course, throughput; for PCI Express it is much higher, for example, PCI at 66 MHz has a throughput of 266 MB/sec, and PCI-E 3.0 (x16) 32 Gb/s.

Externally, the interfaces are also different, so connecting, for example, a PCI Express video card to a PCI expansion slot will not work. PCI Express interfaces with different numbers of lanes are also different, I will now show all this in pictures.

PCI Express and PCI expansion slots on motherboards

PCI and AGP slots

PCI-E x1, PCI-E x16 and PCI slots