Talk:PCI Express
Old/Unititled Conversations
PCI-Express is not, however, fast enough to be used as a memory bus.
First, I think you mean FSB (or some chipset to chipset bus, similiar to the Intel Hub interface, or the PCI bus which glues the North and South bridges together), not "memory bus." The memory bus is handled by the Noth bridge or Memory Hub controller. Secondly, a "high-bandwidth configuration with 8 lanes allowing 8 bits to be sent direction simultaneosly would allow up to 2,000MBps bandwith (each way)..." Also "future increases in signaling speed could increase that to 8,000MBps each way over the same 40 pins." "It...can also be used to replace the existing Intel hub architecture, HyperTransport, and similiar high-soeed interfaces between motherboard chipset components." In light of this information, I don't believe the case that PCI-Express is not "fast enough" for FSB/chipset to chipset communications(?) is supportable.
- Quotes taken from Scott Mueller's Upgrading and Repairing PCs, 15th Anniversary Edition, Chapter 4, Types of I/O Buses, page 333
--MSTCrow 06:31, Jun 22, 2004 (UTC)
Apple's running a 1.25Gbps/line x 64 line bus in their latest G5's, which due to overhead and ping-ponging, delivers about 20GBps continual. That is to say the current on-market machines already outperform the theoretical performance of not-yet-existing versions of PCI-X by almost 3x. HyperTransport is even more effected by this speed issue, as it is only offered as a backplane bus, whereas PCI-X has a future as a PCI replacement (maybe, we'll see). Generally though my statement is clearly supportable even today.
Maury 12:10, 22 Jun 2004 (UTC)
When you say PCI-X, you mean PCI-Express, and not PCI-X, right? PCI-X is another PCI-based bus, apart from PCI-Express.
--MSTCrow 16:25, Jun 23, 2004 (UTC)
Yeah, sorry, my bad short form! Maury 12:04, 24 Jun 2004 (UTC)
pic request
It would be nice if the article had photos of some pci-x devices, i.e. are the connectors similar to PCI?
There are plenty of pictures in the documents at pcisig.com. You could perhaps pick something out and ask their permission to include them if you really want one, but I'm going on to write an article on dielectric loss budgets. Dennis 21:31, 14 Jan 2005 (UTC)
- I think this user meant photos of PCIe devices. The connectors are visually simillar and phyiscally incompatible. --Cmiyc 01:36, 3 August 2006 (UTC)
Non Tech Question
Much goodie techie stuff here, but what about putting a PCI-e card in a PCI slot, does that work ? regards a Non techie geek
- As stated later on, PCIe and PCI/PCI-X are not mechanically compatible. Their physical layers (electrically) are completely different.--Cmiyc 01:36, 3 August 2006 (UTC)
Why is the article's name PCI-Express?
Why is the article's name PCI-Express? PCI SIG uses only PCI Express in its website, this article uses both spellings. --Jannex 09:10, 28 Feb 2005 (UTC)
- Googling around, it seems that PCI Express is the most accepted spelling, so the title and the use in the article should be changed. Thue | talk 13:20, 28 Feb 2005 (UTC)
"vast majority of new computers PCI Express capable"
I remove the sentense
- The vast majority of new computers from both Intel and AMD are PCI Express capable.
from the article. A visit to the website of my local computer store tells me that this is not yet the case. Thue | talk 8 July 2005 12:59 (UTC)
- Um, your local computer store has nowhere near the volume of an OEM such as Dell, HP, etc. (Local shops also tend to lag on newer products. Kind of like how walmart STILL sells nVidia 5200s.) Hence, using it as the basis for PCI Express marketshare is totally baseless. Back on topic, I do believe all the OEMs have pretty much made the switch. So, unless you or anyone else can give a link or show otherwise, that statement should be re-inserted.the1physicist 8 July 2005 17:32 (UTC)
- Well, this is rather seriously offtopic for a page about PCI-express, but I don't consider Wal-Mart to be a choice place for parts of any sort, especially for computers; it's more in the league of an office store: emergency parts only. If you're referring to Nvidia FX5200 cards, they're perfectly capable for 99.5% of machines anyway, so nobody cares. Fsiler 03:36, 5 November 2005 (UTC)
- My local computer shop has quite a high throughput, so it is not _that_ out-dated. Anyway, I checked out Dell and HP's websites. A survey of 6 Dell PCs showed that all of them had PCI-ex graphic card support. A survey of 6 HP PCs (their HP pavillion line) showed that half of them had PCI graphics card support. NONE of the computers had any other PCI-ex slots apart from the graphics one! Based on this quick survey I will agree that PCI-ex seems to be taking over the graphics slot, but not even starting to replace standard PCI slots yet; I would not call a motherboard unqualified "PCI express capable" with only one PCI-ex slot which is taken by the graphics card. The 75% of motherboard which had a graphics card PCI-ex slot also does not justify the formulation "vast majority", even ignoring that they had no generel-purpose PCI-ex slots. Thue | talk 9 July 2005 06:34 (UTC)
- Ok, here's where your argument breaks down. A motherboard does _not_ have to be fully PCI Express to be considered PCI Express capable. That statement is true even if it's mostly just for graphics right now. And actually, most PCI-E motherboards *do* have one or two x1 slots, even if they're not being used. Anyhow, I suppose I retract the word "vast", but the majority of new computers do have some form of PCI Express.the1physicist 9 July 2005 07:37 (UTC)
- In my book it is misleading to say that a motherboard is "PCI express capable" when only the graphics slots are PCI express; reading that I would expect them to have at least one free generel-purpose PCIe slot. None of the 12 systems I checked out had any 1x PCIe slots! We can however say in the article that "the majority of systems seems to have a PCIe slot for graphics". Thue | talk 9 July 2005 07:44 (UTC)
- Alrighty then. You do know that you can plug a x1 card in to a 16X slot, right? Not to mention that most motherboards with 16X slots usually have 1-2 1x slots. You seem to have gotten unlucky or something.the1physicist 9 July 2005 10:07 (UTC)
Hot pluggability?
On pinouts.ru, the PCI Express pinout has some hot-plug detect pins. Is PCI Express hot-pluggable? And if it is, why? Mrdelayer~ 10:20, 15 July 2005 (UTC)
- I believe they were planning on making a cable specification. Something like USB, but better.the1physicist 15:56, 15 July 2005 (UTC)
- For servers and other high-up time systems. --Cmiyc 01:36, 3 August 2006 (UTC)
Card/slot combinations
What card/slot size combinations are possible? For example, can an x4 card be plugged into an x16 slot? Can an x8 card be plugged into an x4 slot?
I believe that with PCI, a 32-bit card could be plugged into a 64-bit slot, and a 64-bit card could (at least in some cases) be plugged into a 32-bit slot (it would just operate at the lower bandwidth). What are the comparable possibilities for PCI Express?
I hoped to find an answer to this question here, so presumably others may also.
--Beric 14:38, 26 July 2005 (UTC)
You can't overhang the cards like you can with PCI, but you can install lower speed cards into faster slots (and the rest of the pins are [of course] unused). So, you could put a x1 or x4 into your x16 slot if you wanted.
-Dan November 14, 2005
i was looking for this info, too. it appears you can, based on a 2004 test by Tom's (see link i added at bottom of link section (altho it may be a special/nonavailible BIOS). also, an ASRock Core 2 mobo either has just a x4 slot or a x16 slot that only runs at x4. i was curious to see if it was even worth considering it and that's the info my first 30 seconds of websearching turned up. looks to be "acceptable" at least up to a 6800 Ultra. not sure about higher... Plonk420 21:04, 17 August 2006 (UTC)
General Compatibility
I'd heard that PCI-E is "backwards compatible" with PCI. When I got my new computer with PCI-E, I wanted to keep one of my old PC cards, but it would not physically fit: the PCI-E slots have a space between the pins near the panel side that the PCI card I wanted to keep didn't have a matching gap for, so it could not go in the slot. However, I checked some of my other PCI cards, and some of them did have that gap and did fit. Now, I want to buy a new card (USB2 & Firewire), but the only PCIe cards available so far cost a lot, and I'm fine with a PCI card (speedwise), but since my old one didn't fit, how can I tell if a new one will fit? I'm not looking for someone to directly answer my questions on Wikipedia, I can find a forum for that. It's just that there's no section addressing compatibility with PCI cards, I expected to get some info on this here, if anyone could add it.
-Tom, June 6, 2006
- PCI-E is NOT backwards compatible with PCI PERIOD, the software layers may be the same but the physical interface is totally different. BTW the intro to this article reads rather like a marketing peice. Plugwash 23:11, 15 June 2006 (UTC)
Performance Difference
Is there really noticible (higher than 5%) performance difference between PCI-e and AGP graphics card? I have used both versions of 6600GT and I cannot notice any difference even though the bridging chip should slightly slow it down. --Antilived 11:01, 6 August 2005 (UTC)
- For the most part, there is little to no performance penalty for AGP. I do believe that a PCI-E 8X will be about 2% slower than 16X, and a 4X about 5% slower than a 16X.the1physicist 22:11, 6 August 2005 (UTC)
- This is application-dependant. PCIex does have much less latency and overhead for small transactions. On an application doing lots of small texture updates (google for "uberflow" to have an example) the PCIex machine ended up being 80% faster than the AGP one on similar hardware! 83.176.117.138 08:03, 31 July 2006 (UTC)
[[User:|User:]] 14:38, 28 October 2005 (UTC)
Looking at the Specfications for the TYAN Tomcat i7230A, it claims to have 2 PCIe x16 slots. One with x8 Signal and one with x4 signal. What does this mean?
There's three different speeds - card, slot, and signal. As long as the card is <= the slot speed, it will work. However, all of the slots could have less signal lines so they're really slower than expected. However, I don't know exactly where the difference is (perhaps in the bridge chip??).
-Dan November 14, 2005
PCI-X is merged with the PCI article. Suggest that this be merged with the PCI article, too? --BleachInjected 02:56, 4 November 2005 (UTC)
- I'd* have to say no, because unlike PCI-X (which was an extension of PCI), this is different physically. It's serial and has dedicated channels. It's designed to be compatable with PCI programming (only requiring a new transport), but it's not electrically similar.
-Dan November 14, 2005
article language
This is by no means a criticism of the PCIe specification -- merely a reminder that PCIe is not exempt to the rules of a layered packet protocol, no more so than other comparable high-speed serial technologies (such as Serial ATA and Fibre Channel). It seems to me this language doesn't belong in an encyclopedia article. How about (This is not an issue peculiar to the PCI express interface.) If nobody complains, I'll change it. Feel free to revert (with discussion).
Help with 'conversion'
"As it is based on the existing PCI system, cards and systems can be converted to PCI Express by changing the physical layer only – existing systems could be adapted to PCI Express without any change in software." Does this mean a PCI slot can support a PCI-express card? Or do you have to have a PCI-express slot that could possibly support an older PCI card? I'd like to get a new video card, but the one I want is PCI-Express, while I only have PCI-X slots. EDIT: I think they're PCI-X slots, but I'm not completely sure on if they are or how to find out. PirateMonkey 02:18, 28 January 2006 (UTC)
- The slots are not compatible in any way. I presume the new PCIe slots were designed to make it impossible to insert old PCI cards. That section is simply saying that there needs to be no change in the software interface. Makers of PCI cards/chips can make a PCIe version by changing the physical interface and the card will work in a PCIe system. The system still sees a host bridge, one or more buses behind it, and devices with vendor:device ID's. No massive change to the way that works, except for graphics cards moving from AGP. Imroy 06:05, 28 January 2006 (UTC)
HowStuffWorks article
I've read on howstuffworks in article about PCI next:
“ | PCI-Express slots will also accept older PCI cards, which will help them become popular more quickly than they would if everyone's PCI components were suddenly useless. | ” |
How can it be true? Or may be that article too old? PCIe slots are completely different from PCI - or I'm mistaking? If my motherboard have both PCI & PCIe stots - is it mean that all PCI slots - are PCIe slots but with PCI-formfactor ? Or motherboard have 2 different buses PCI & PCIe ? And PCI slots will be works as usual with bus multiplexing ? — Preceding unsigned comment added by 81.200.2.184 (talk • contribs)
- That information is just wrong. The slots are different, and the electrical interface is *very* different. Many (most?) motherboards with PCIe slots also have legacy PCI slots as well. It's also possible the author of that information was confusing PCIe with PCI-X.
- Imroy 09:51, 30 March 2006 (UTC)
PCI Express
Please wikipedia
In Hardware level. Talk me in bit or bits or b. In Software level. Talk me in BYTE OR BYTES or B.
No confusion for final user, PLEASE !
or always b = bit(s) and B = BYTE(S)
PEG ≡ PCI Express
Is PEG (PCI-Express For Graphics) 100% identical with normal PCIe or not? --Hhielscher 01:05, 16 April 2006 (UTC)
- PEG == PCI Express Graphics. It is 100% identical, but is specifically called PEG because the slots sit directly off of the northbridge/MCH, so they have less latency for memory and CPU cycles. The regular PCI Express slots sit off the southbridge, and there's extra latency in communicating upstream to the northbridge. Rmcii 20:09, 13 May 2006 (UTC)
PCI Express 2.0 Specification
Does anyone have a soild source regarding the upcoming PCI Express 2.0 specification? Most notably the expected final release date and the added hardware support for virtualization in this new standard?
I'm missing information on the PCIE 1.0a spec. What is different for 1.0a?
What is newer, 1.0 or 1.0a? What does it mean for a graphics card to be 1.0a compliant?
adaptors
i guess an adaptor could theoretically be made to fit a card into a smaller slot then intended but there would be physical issues with the machines case in doing so. Would that be a correct analysis? Plugwash 23:15, 15 June 2006 (UTC)
- Yes that would be correct. The added height would cause physical problems. There are companies that make "shims" to convert to lower lane widths. They are useful for testing.
Outlook
The article says "As of 2006, PCI Express appears to be well on its way to becoming the new backplane standard in personal computers." However, since each PCI Express lane stands alone (as contrasted to the shared 32 or 64bit bidirection pins of regular PCI), it doesn't appear to match the definition of backplane given. I suggest that this be revised to "As of 2006, PCI Express appears to be well on its way to becoming the new internal expansion standard in personal computers." JDBoyd 13:49, 09 Aug 2006 (EST)
The article says that as of 2005 it looks like PCI-E may become the standard for PC's. Well it's half past 2006 and I think PCI-E has indeed become the standard for PC's. Anyone agree? Dionyseus 21:49, 14 July 2006 (UTC)
- Its certainly dominated all other newcomers (PCI-X agp etc) but afaict its still got some way to go to kill off PCI as the dominiant general purpose expansion slot system. Plugwash 02:25, 26 August 2006 (UTC)
BW/clock
Where is the bandwidth and clock info? I believe this should be pointed out somewhere. 83.176.117.138 08:05, 31 July 2006 (UTC)
- This question doesn't really make sense to me. What clock info? The bit rate is 2.5Gbit/s. The reference clock is 100MHz. There are sources on the actual bandwidth, after you account for some of the overhead and 8b/10b encoding.
250 MiB or 250 MB
It is definitely 250 MB/s.
It's 1.25GHz * 2 bits/clock / (10 bits/byte) = 250 million bytes per second = 250 MB/s
To quote the IBM Redbook (see Link in the article):
"The 8b/10b encoding essentially requires 10 bits per character, or about 20% channel overhead. This encoding explains differences in the published spec speeds of 250 MBps (with the embedded clock overhead) and 200 MBps (data only, without the overhead)."
And especially to DmitryKo: If 2.5 Gbit/s was the real data bandwdith, you'd get 298 MiB/s and not 250 MiB/s.
I hope that's proof enough for you. If not, please show me how to get 250 MiB/s with a clock of 1.25GHz, DDR and a 8b/10b encoding. In my opinion, this is a calculation you should have done before changing the MB to MiB.
JogyB 21:22, 27 August 2006 (UTC)
- I just didn't realize bus bandwidth is expressed in decimal units... sorry. --Dmitry (talk •contibs ) 06:22, 28 August 2006 (UTC)