IBM ThinkPad Dock II (Model 2877) + T43. PCI performance

[baryluk]

Hi,

I'm considering buying IBM ThinkPad Dock II (Model 2877), to my ThinkPad T43. And have few questions to the users of both.

First.

I would like to put some SATA controler in the PCI slot. I know it should fit (I will use flexible PCI raiser if it will not) , however I'm wondering what possible speed I can get. Is this full PCI (33 or 66MHz probably) slot? Or is there some kind of bridge between it and the thinkpad (in Dock, or in both dock and T43)? I guess, there is some bridge anyway, and it is shared by this PCI slot, and IDE controller (CMD 648 for the Ultrabay 2000 slot) and PCMCIA cards (second Ricoh controller probably, right?). Can it get full 133MB/s speed in both direction (or more)?

Second.

How about hot plugging and unplugging this dock? I am using Linux, so unmounting file systems etc, unloading PCI card and Ultrabay drivers, and disabling PCI should do the trick. Is there anything to remember when pluging or unpluging it on running T43? Will power to the other components (includin PCI card) go off, when I unplug T43 from it?

AFAIK there should not be big issue, as actually PCI is behind PCIe to PCI bridge (most of T43 interal chips are on PCIe, with exception of the Wifi/bluetooths and modem card only probably).

I was also wondering about actually architecture of PCI bus in T43 and Dock II. My current PCI tree looks like this

Code: Select all

# lspci -t -v -n
-[0000:00]-+-00.0  8086:2590
           +-01.0-[01]----00.0  1002:5460
           +-1c.0-[02]----00.0  14e4:167d
           +-1c.2-[03-0a]--
           +-1d.0  8086:2658
           +-1d.1  8086:2659
           +-1d.2  8086:265a
           +-1d.3  8086:265b
           +-1d.7  8086:265c
           +-1e.0-[0b-0e]--+-00.0  1180:0476
           |               \-02.0  8086:4224
           +-1e.2  8086:266e
           +-1f.0  8086:2641
           +-1f.2  8086:2653
           \-1f.3  8086:266a
# lspci -n | egrep '1c.0|1c.2|1e.0'
00:1c.0 0604: 8086:2660 (rev 03)
00:1c.2 0604: 8086:2664 (rev 03)
00:1e.0 0604: 8086:2448 (rev d3)
#

Or with names

Code: Select all

# lspci -t -v
-[0000:00]-+-00.0  Intel Corporation Mobile 915GM/PM/GMS/910GML Express Processor to DRAM Controller
           +-01.0-[01]----00.0  ATI Technologies Inc M22 [Mobility Radeon X300]
           +-1c.0-[02]----00.0  Broadcom Corporation NetXtreme BCM5751M Gigabit Ethernet PCI Express
           +-1c.2-[03-0a]--
           +-1d.0  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #1
           +-1d.1  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #2
           +-1d.2  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #3
           +-1d.3  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB UHCI #4
           +-1d.7  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) USB2 EHCI Controller
           +-1e.0-[0b-0e]--+-00.0  Ricoh Co Ltd RL5c476 II
           |               \-02.0  Intel Corporation PRO/Wireless 2915ABG [Calexico2] Network Connection
           +-1e.2  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) AC'97 Audio Controller
           +-1f.0  Intel Corporation 82801FBM (ICH6M) LPC Interface Bridge
           +-1f.2  Intel Corporation 82801FBM (ICH6M) SATA Controller
           \-1f.3  Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) SMBus Controller

# lspci | egrep '1c.0|1c.2|1e.0'
00:1c.0 PCI bridge: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) PCI Express Port 1 (rev 03)
00:1c.2 PCI bridge: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) PCI Express Port 3 (rev 03)
00:1e.0 PCI bridge: Intel Corporation 82801 Mobile PCI Bridge (rev d3)
#

Ricoh is CardBus controller (for PCMCIA, PCCard; as Express Card slot independent from it and connects to the 1c.2 (PCI Express Port 3)). So I guess, ThinkPad Dock II will plug in just below it Mobile PCI Bridge (in 1e.0), as next PCI bridge, and there will be CMD 648 IDE controller, Ricoh PCMCIA Controler (my guess) and one (or more) more things depending what I plug into empty PCI slot there. It will essentially be sharing bandwidth and interrupts with WIFI in notebook. Right? Can anybody with Dock II, and PCI card show us some informations (like me above).

Thrid.

Will T43p (14.1 and 15" version) with 6 or 9 cell batter will work with this Dock?

I cannot find any manual for this Dock on Lenovo age, and IBM pages for it are dead.
I was searching for any discussion about Dock II, but only found page on thinkwiki.org

Thanks in advance.

[RealBlackStuff]

You might be interested to convert your T43/T43p/R52 hard disk to SATA:
http://forum.thinkpads.com/viewtopic.php?f=2&t=94308
http://mikejmoffitt.com/sata/

[baryluk]

You might be interested to convert your T43/T43p/R52 hard disk to SATA:
http://forum.thinkpads.com/viewtopic.php?f=2&t=94308
http://mikejmoffitt.com/sata/

Thanks!

I know that in principle it was possible, but would need some small reverse engineering and soldering. Nice to see that somebody actually done this!

However, single disk, or even two disks (second in Ultrabay), may not be enough for me. SSD would probably be nice option, but it is still low capacity and very expensive. I know that there is also few other options like using dual eSATA controler in the Express Card. It will be much faster, however will not be so nice as Dock solution (cumberstom cables, and SATA multipliers).

[RealBlackStuff]

If you need more HD capacity, forget about SSD for another 3-5 years.
If you convert to SATA (I have done it also, 1/2 hour to remove chip, 1 hour to remove IDE and put in SATA connector), you could have a 1TB drive in the main bay, and another 1TB in a SATA adapter in your Ultrabay.
If 2TB is not enough, you need to seriously consider other options.

[baryluk]

If you need more HD capacity, forget about SSD for another 3-5 years.
If you convert to SATA (I have done it also, 1/2 hour to remove chip, 1 hour to remove IDE and put in SATA connector), you could have a 1TB drive in the main bay, and another 1TB in a SATA adapter in your Ultrabay.
If 2TB is not enough, you need to seriously consider other options.

I'm not considering SSD. It is low capacity, expensive, and will not show full power in T43. I also do not trust them too much

It is not about HD capacity, rather performance. All I am asking is about PCI speed in the Dock II. Nothing more.

I'm already have few 2TB SATA drives in 3.5" form factor, and was thinking about using them. I want to use them as storing some big files, backups and to accelerate development. I have many applications which are I/O intensive, and number of IO per second is quite important. So I will rather be more happy with 4x320 GB than 2x1TB actually. I have ExpressCard SATA controler, but I do not like cables on the side of the laptop. Dock would be easier to use.

It is however not discussion what is better. I'm just asking for answers for the question in the initial post.

Thanks.

[ajkula66]

How about installing a UltraBay 2000 SATA HDD adapter in the Ultrabay slot on the dock?

That should get you where you need to be without major harassment/investment.

No clue about transfer speeds on the dock, though.

Good luck.

[RealBlackStuff]

You are supposing WAY too much of such a laptop.
It's 6 year old technology and rather underpowered for what you seem to want.
I do not have any figures for throughput etc., these are IBM-internal and were never published.
IMHO you should look for a workstation-type laptop or rather, a desktop.

[baryluk]

Thanks,

it is not any secret - it can be simply measured by anyone having IDE or SATA controler in Dock II, and I know there is many such T4x users.

As, of upgrading system, it will not really help me, as newer hard drives doesn't have some much better IO per second. And additionally I have SATA controler, cables, few big hard drives, and Dock II is cheap (I can find one for 25$).

Anyway, I like pushing hardware to their limits, and above

[automobus]

I have a Dock I and a X22. You might not be interested because my hardware so different that yours. Do you want me to run some tests?

[baryluk]

I have a Dock I and a X22. You might not be interested because my hardware so different that yours. Do you want me to run some tests?

Well, not much. This is very different Dock (you wrote Dock I, which actually doesn't work with X22). It doesn't have PCI slot, only ISA, and you will not really runny any useful test in it (until you have some SCSI controller in this ISA slot and some disks connected to it, which I doubt). Maybe you are talking about Dock II, or Mini-Dock?

If you have some time and Linux then you can do (as root):

lspci

;

lspci -n

;

lspci -t -v

;

lspci -v

; will be useful anyway (you can skip last one, it is pretty big). With dock connected. I can analyses it and put some notes on thinkwiki.org. If you have Dock II and some storage (ide, sata, usb) connected to it, then

hdparm -tT /dev/name-of-disk-in-dock-like-sdb

; will measure quickly bandwidth of the PCI bus (between buffer of the controller on the PCI and main memory and CPU) and the hard disk.

But if this is too hard, skip it, I'm essentially interested in Dock II, and T4x (faster CPU, faster memory, and PCI Express main bus, instead of just PCI), as measurements will be much different. If you want, you can post, it will not hurt, and maybe will help somebody else (and google).

[automobus]

No, not the really old Dock I. I have the "2000" Dock I. It is nearly the same thing as "2000" Dock II, which you have (Dock II is differently sized to fit T40 generation). If you have a look at Docking_Solutions, you will see there was a "1990" Dock I and II; SelectaDock I and II and III; then "2000" Dock I and II.

[baryluk]

No, not the really old Dock I. I have the "2000" Dock I. It is nearly the same thing as "2000" Dock II, which you have (Dock II is differently sized to fit T40 generation). If you have a look at Docking_Solutions, you will see there was a "1990" Dock I and II; SelectaDock I and II and III; then "2000" Dock I and II.

Oh. Never seen this. Thanks for information. It will not work with T4x probably, however It will be interesting to see some diagnostics. It should be essentially the same.

I even suppose with some tricks Dock I could work with T43, by analysing multiple docs I guess that Dock I and Dock II, have same connector between dock and the laptop (for example T30 or X24 models works with both). Probably only part of functionality will be not supported, and maybe laptop will not fit properly (for example only 14.1" models will work but not 15"). Just a theory.

[automobus]

Will T43p (14.1 and 15" version) with 6 or 9 cell batter will work with this Dock?

I forgot to answer you sooner. Yes, there is a flap shaped like the big battery, that folds away. The battery fits, no problem.

Code: Select all

in UltraBase Ultrabay
average read 53.2 (min 29.8 max 72.9) "MB/s"
average access 17.0 ms
hdparm -t -T /dev/sdb:
 Timing cached reads:   428 MB in  2.00 seconds = 214.01 MB/sec
 Timing buffered disk reads: 180 MB in  3.02 seconds =  59.56 MB/sec

in Dock Ultrabay
average read 50.4 (min 30.7 max 57.1) "MB/s"
average access 16.9 ms
hdparm -t -T /dev/sdb:
 Timing cached reads:   438 MB in  2.00 seconds = 218.67 MB/sec
 Timing buffered disk reads: 140 MB in  3.02 seconds =  46.41 MB/sec

in local CardBus
average read 53.4 (min 29.5 max 72.9) "MB/s"
average access 17.1 ms
hdparm -t -T /dev/sdb:
 Timing cached reads:   434 MB in  2.00 seconds = 216.98 MB/sec
 Timing buffered disk reads: 180 MB in  3.02 seconds =  59.56 MB/sec

in Dock CardBus
exactly the same
as local CardBus
hdparm -t -T /dev/sdb:
 Timing cached reads:   442 MB in  2.00 seconds = 220.98 MB/sec
 Timing buffered disk reads: 178 MB in  3.01 seconds =  59.18 MB/sec

Of note, Dock I is from year 2000. The ATA controller in Dock I is capable of Ultra DMA 4, Ultra ATA/66, ATA-5.

Those numbers are with the fastest drive I have at hand. I used SATA CardBus card and Ultrabay 2000 HDD adapter. I can tell you from experience that the PCI slot is full-featured, capable of add-in BIOS and bootable. I have used a RAID card and video card before. Right now the PCI slot is empty. Since the limit in my equipment is the hard drive itself, I did not go through the effort to test PCI.

Code: Select all

lspci -v -nn -s 09:
09:01.0 IDE interface [0101]: Silicon Image, Inc. PCI0648 [1095:0648] (rev 01) (prog-if 8f [Master SecP SecO PriP PriO])
	Subsystem: Silicon Image, Inc. PCI0648 [1095:0648]
	Flags: bus master, medium devsel, latency 64, IRQ 11
	I/O ports at 8020 [size=8]
	I/O ports at 8014 [size=4]
	I/O ports at 8018 [size=8]
	I/O ports at 8010 [size=4]
	I/O ports at 8000 [size=16]
	Capabilities: <access denied>
	Kernel driver in use: pata_cmd64x

09:02.0 CardBus bridge [0607]: Texas Instruments PCI1420 PC card Cardbus Controller [104c:ac51]
	Subsystem: IBM ThinkPad A20m [1014:0148]
	Flags: bus master, medium devsel, latency 168, IRQ 11
	Memory at 52000000 (32-bit, non-prefetchable) [size=4K]
	Bus: primary=09, secondary=0a, subordinate=0d, sec-latency=176
	Memory window 0: 34000000-37fff000 (prefetchable)
	Memory window 1: 2c000000-2ffff000
	I/O window 0: 00003000-000030ff
	I/O window 1: 00003400-000034ff
	16-bit legacy interface ports at 0001
	Kernel driver in use: yenta_cardbus

09:02.1 CardBus bridge [0607]: Texas Instruments PCI1420 PC card Cardbus Controller [104c:ac51]
	Subsystem: IBM ThinkPad A20m [1014:0148]
	Flags: bus master, medium devsel, latency 168, IRQ 11
	Memory at 53000000 (32-bit, non-prefetchable) [size=4K]
	Bus: primary=09, secondary=0e, subordinate=0e, sec-latency=176
	Memory window 0: 38000000-3bfff000 (prefetchable)
	Memory window 1: 30000000-33fff000
	I/O window 0: 00003800-000038ff
	I/O window 1: 00003c00-00003cff
	16-bit legacy interface ports at 0001
	Kernel driver in use: yenta_cardbus

Code: Select all

lspci -t -v
-[0000:00]-+-00.0  Intel Corporation 82830 830 Chipset Host Bridge
           +-01.0-[01]----00.0  ATI Technologies Inc Radeon Mobility M6 LY
           +-1d.0  Intel Corporation 82801CA/CAM USB Controller #1
           +-1d.1  Intel Corporation 82801CA/CAM USB Controller #2
           +-1d.2  Intel Corporation 82801CA/CAM USB Controller #3
           +-1e.0-[02-17]--+-03.0  Ricoh Co Ltd RL5c476 II
           |               +-03.1  Ricoh Co Ltd RL5c476 II
           |               +-05.0  Agere Systems L56xM+S [Mars-2] WinModem 56k
           |               +-06.0-[09-0f]--+-01.0  Silicon Image, Inc. PCI0648
           |               |               +-02.0  Texas Instruments PCI1420 PC card Cardbus Controller
           |               |               \-02.1  Texas Instruments PCI1420 PC card Cardbus Controller
           |               \-08.0  Intel Corporation 82801CAM (ICH3) PRO/100 VE (LOM) Ethernet Controller
           +-1f.0  Intel Corporation 82801CAM ISA Bridge (LPC)
           +-1f.1  Intel Corporation 82801CAM IDE U100 Controller
           +-1f.3  Intel Corporation 82801CA/CAM SMBus Controller
           \-1f.5  Intel Corporation 82801CA/CAM AC'97 Audio Controller

[baryluk]

Wow. This is extremely useful information! Thanks you.

It looks that PCI slot and connection between dock and laptop is actually PCI 66MHz (PCI 33MHz, have maximum transfer128MB/s)! Thank you also for latencies (average access time), this is even more important to me, but just assumed, testing this will be to hard for you It looks that bridge chip does not introduce any noticeable latency.

Also PCI tree structure confirms my theory. If you could post also just "lspci", "lspci -n", it would be everything. (lspci -t unfortunately doesn't show bridges in PCI bus, like this marked as 06.0 - I even reported this issues to pciutils maintainers http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=640239 ).

As of booting I was also speculating in my head it is possible, but do not really need it. Thanks for confirmation.

So Dock II could be even better, or at least equal, it is probably just compatibility with other laptop models. And maybe slightly different chips (we can compare this with "lspci", "lspci -n" then).
Will get Dock II in few days, and will also post my results.

I will also probably use SiL SATA chip, mostly because I have used them in many machines, and know that most issues there was some time (especially in non-Linux OSes), are mostly solved. They are also VERY cheap, and supports FIR based SATA multiplexing (in case of using SATA multipliers - which I really do not need yet, as SiL 3124 have 4 SATA ports - enough for me ).

[automobus]

No, not 66 MHz. I think the cached reads mislead you. The cached numbers give you an idea of my computer's memory transfer speed, not the potential of PCI. Here are bridge details.

Code: Select all

lspci -nn | grep '\[06'
00:00.0 Host bridge [0600]: Intel Corporation 82830 830 Chipset Host Bridge [8086:3575] (rev 02)
00:01.0 PCI bridge [0604]: Intel Corporation 82830 830 Chipset AGP Bridge [8086:3576] (rev 02)
00:1e.0 PCI bridge [0604]: Intel Corporation 82801 Mobile PCI Bridge [8086:2448] (rev 41)
00:1f.0 ISA bridge [0601]: Intel Corporation 82801CAM ISA Bridge (LPC) [8086:248c] (rev 01)
02:03.0 CardBus bridge [0607]: Ricoh Co Ltd RL5c476 II [1180:0476] (rev 80)
02:03.1 CardBus bridge [0607]: Ricoh Co Ltd RL5c476 II [1180:0476] (rev 80)
02:06.0 PCI bridge [0604]: Texas Instruments PCI2032 PCI Docking Bridge [104c:ac22]
09:02.0 CardBus bridge [0607]: Texas Instruments PCI1420 PC card Cardbus Controller [104c:ac51]
09:02.1 CardBus bridge [0607]: Texas Instruments PCI1420 PC card Cardbus Controller [104c:ac51]



The host bridge is not 66MHz.
00:00.0 Host bridge [0600]: Intel Corporation 82830 830 Chipset Host Bridge [8086:3575] (rev 02)
	Subsystem: IBM ThinkPad A/T/X Series [1014:021d]
	Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort+ >SERR- <PERR- INTx-
	Latency: 0
	Region 0: Memory at d0000000 (32-bit, prefetchable) [size=256M]
	Capabilities: <access denied>
	Kernel driver in use: agpgart-intel



The main PCI bridge is not 66MHz.
00:1e.0 PCI bridge [0604]: Intel Corporation 82801 Mobile PCI Bridge [8086:2448] (rev 41) (prog-if 00 [Normal decode])
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
	Status: Cap- 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
	Latency: 0
	Bus: primary=00, secondary=02, subordinate=17, sec-latency=96
	I/O behind bridge: 00003000-00008fff
	Memory behind bridge: b0200000-c08fffff
	Prefetchable memory behind bridge: e8000000-efffffff
	Secondary status: 66MHz- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- <SERR- <PERR-
	BridgeCtl: Parity- SERR- NoISA+ VGA- MAbort- >Reset- FastB2B-
		PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-



One Ricoh (ThinkPad onboard) CardBus has busses 10-13 and the other has 14-17.
The Dock has busses 09 through 0f.
One Texas Instruments (inside Dock) CardBus has busses 0a-0d and the other has just 0e.



02:06.0 PCI bridge [0604]: Texas Instruments PCI2032 PCI Docking Bridge [104c:ac22] (prog-if 01 [Subtractive decode])
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
	Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
	Latency: 64, Cache Line Size: 32 bytes
	Bus: primary=02, secondary=09, subordinate=0f, sec-latency=68
	I/O behind bridge: 00003000-00003fff
	Memory behind bridge: 2c000000-33ffffff
	Prefetchable memory behind bridge: 34000000-3dffffff
	Secondary status: 66MHz- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort+ <SERR- <PERR-
	BridgeCtl: Parity- SERR- NoISA+ VGA- MAbort- >Reset- FastB2B-
		PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
	Capabilities: <access denied>

[automobus]

I got my hands on a faster disk: WD10EARX. In all of the following configurations, I measured sustained sequential transfers in the first 20 Go (lowest LBAs, fastest portion of disk). First, I tested it on a host with native (1500 Mbit) SATA (Intel ICH6). The sustained transfer rate was at least 120 Mo/s. (It might have been higher, but I did not write down the result and I am going from memory. All other numbers were recorded and are accurate.)

Code: Select all

Intel ICH6 native SATA : 120 Mo/s

X22 Ricoh + SiI3112 : 94 Mo/s
X22 Ricoh + INIC1620 : 98 Mo/s

X22 + Dock TI PCI1420 + SiI3112 : 89 Mo/s
X22 + Dock TI PCI1420 + INIC1620 : 89 Mo/s

T43 TI PCI1510 + SiI3112 : 110 Mo/s

T23 TI PCI1420 + SiI3112 : 110 Mo/s

My card with Initio chip might be defective. It operates properly in ThinkPads X22 X30 X41t (all Ricoh chips), and Dock (TI PCI1420). It does not operate in T23 (TI PCI1420) and T43 (TI PCI1510): the system freezes while card is inserted, and resumes when card is ejected. dmesg reports only one second between insertion and removal, no matter how much real time spent frozen. Windows behaves the same way: operating system seems stable after multiple insertion and removals. T60 with Windows behaves differently, the operating system does not recover gracefully. Inserting the card is not safe: freeze while inserted, and BSOD upon removal.

It is strange that the Dock works with the card: it contains TI chipset similar to T23. If my card is not defective, then perhaps this model in general is abnormal. I very much want to know what the model is. My only guess is Sunix CBSA20I, but the board within has no hints. StarTech's part number is CBESATA2.

Who would like to own this card? Someone with a X60, X61? I want to get rid of it.

[automobus]

I got my hands on a faster disk: MMCRE28GTMXP. According to ltwbook, Lenovo ThinkPad R60e has TI PCI1512 CardBus bridge (PCI ID 104C:AC56 subsystem 17aa:2012). Host PCs are ThinkPad T20 R60e and T61p, and Dell Latitude E4200.

Code: Select all

T20 440BX + PIIX4E + PCI1450 + SiI3112 : 107 Mo/s
T20 440BX + PIIX4E + PCI1450 + SiI3124-1 : 103 Mo/s

R60e 945GM + ICH7M native SATA : 135 Mo/s
R60e 945GM + ICH7M + PCI1512 + SiI3512 : 113 Mo/s
R60e 945GM + ICH7M + PCI1512 + SiI3124-1 : 108 Mo/s

T61p PM965 + ICH8EM + SiI3132 : 140 Mo/s
T61p PM965 + ICH8EM + R5C847 + SiI3112 : 94 Mo/s
T61p PM965 + ICH8EM + R5C847 + SiI3512 : 94 Mo/s
T61p PM965 + ICH8EM + R5C847 + SiI3124-1 : 101 Mo/s

E4200 GS45 + ICH9M-SFF native SATA : 220 Mo/s
E4200 GS45 + ICH9M-SFF + SiI3132 : 140 Mo/s

edit:correct typing mistake

[automobus]

Again, I tested read performance from a single disk, Samsung MMCRE28GTMXP. This time, I have an actual PCI card and I borrowed a desktop: Dell OptiPlex GX260.

Code: Select all

sustained read speed
GX260 ICH4 + SiI3124-2 : 126 Mo/s
Dock PCI2032 + SiI3124-2 : 101 Mo/s

random seek 500000 average
GX260 ICH4 + SiI3124-2 : 0,45 ms
Dock PCI2032 + SiI3124-2 : 0,53 ms

If the speed of a typical SATA PCI card (32@33) is 126 Mo/s, the speed of a SATA card in ThinkPad Dock is about 101 Mo/s. The PCI slot in ThinkPad Dock affords better performance, than CardBus as implemented in many laptops. Even better than some desktop PCs, with 25 MHz PCI bus speed.

edit to remove misleading false information

[FrankL]

Again, I tested read performance from a single disk, Samsung MMCRE28GTMXP. This time, I have an actual PCI card and I borrowed a desktop: Dell OptiPlex GX260.

Code: Select all

sustained read speed
GX260 ICH4 + SiI3124-2 : 126 Mo/s
Dock PCI2032 + SiI3124-2 : 101 Mo/s

random seek 500000 average
GX260 ICH4 + SiI3124-2 : 0,45 ms
Dock PCI2032 + SiI3124-2 : 0,53 ms

If the speed of a typical SATA PCI card (32@33) is 126 Mo/s, the speed of a SATA card in ThinkPad Dock is about 101 Mo/s. The PCI slot in ThinkPad Dock affords better performance, than CardBus as implemented in many laptops. Even better than some desktop PCs, with 25 MHz PCI bus speed.

Do not be mislead by my test results: Ricoh CardBus bridge chips are not very good. Write speed is not consistent. In my X22, average write speed does not ever exceed 18 Mo/s. Typical to one disk is 14 Mo/s. Writing to two disks simultaneously, 5 Mo/s each. Read from one while write to another: 45 and 8 Mo/s. As a mere computer user, with no insight as a driver writer or engineer, I do not understand Ricoh chip behaviour. Why does it slowly receive commands from host, but it is capable of transferring to the host at high speed? This is a "show stopper" for creative use. I agree with all the audio and video professionals: Ricoh's CardBus and IEEE 1394 controller chips suck.

are you comparing to a Pentium 100 MHz system? I don't know of any machines newer than that, which run on 25 MHz PCI.

If so, I doubt you are going to find many people still running a computer like that, let alone in a situation where they care about their IO throughput.

[automobus]

I did put my foot in my mouth. I now know there is a big difference in performance between the first PCI systems and a typical year 2000 system. I am wrong to conclude "something sucks" having observed only few devices. I did begin to measure write speeds instead of only read. Every laptop CardBus which I tried (bridges include TI or Ricoh or O2), seems to allow disk-write at a rate nowhere near disk-read rate. One Ricoh which I tried allowed 91.1 read and 66.9 write: that is rather good, I think.