|
Guest
|
 |
« on: November 28, 2002, 05:57:51 PM » |
|
Holy! I have GA-7VAXP. I believe they have enabled PCI parking with the latest F9 bios. Because i just tried the HD test and my burst speed is now 88mb/s compared to 66mb/s i use to get.
|
|
|
|
|
Logged
|
|
|
|
|
Guest
|
|
« Reply #1 on: November 28, 2002, 05:57:51 PM » |
|
PCI PARKING
|
|
|
|
|
Logged
|
|
|
|
|
nelamvr6
Regular
Posts: 31
Join Date: Sep, 2002
|
|
« Reply #2 on: November 28, 2002, 06:27:29 PM » |
|
What exactly is PCI parking?
Can it be enabled outside of the BIOS?
|
|
|
|
|
Logged
|
|
|
|
|
GeorgeBreese
Ace
Posts: 2,088
Join Date: Dec, 2001
|
|
« Reply #3 on: November 29, 2002, 11:38:37 AM » |
|
<< What exactly is PCI parking? >>
PCI Bus Parking is a cool concept. I just wrote a long description that may end up in my FAQ, and sent it to a PCI guy for proofreading. In the meantime, here's the short version of the story.
Normally, VIA's PCI controller will grant each PCI device a few microseconds in which to use the bus. VIA's PCI controller keeps choosing different devices to receive the grant, and never chooses the same device twice. This works best when there are a lot of PCI devices and each PCI device only needs a little time on the bus to transfer its data. And, by the way, VIA isn't the only chip maker to do it this way.
The normal VIA PCI behavior doesn't work well when a PCI device needs a lot of time on the bus. ATA133 IDE controllers from HighPoint and Promise made this obvious. An ATA133 drive can deliver data at 133 MB/second, which happens to be the same speed as a the PCI bus. So, VIA needed a way to allow a single PCI device to use the bus as long as it wants, or at least until some other PCI device really needs it.
Intel had run into this problem a few years ago. They solved it by modifying their PCI controller chips. Remember, I said that VIA's PCI controller would never choose the same device twice? Intel's new idea was to allow the PCI controller to choose the same device again and again, until another PCI device really needs to use the bus.
<< Can it be enabled outside of the BIOS? >>
My early testing seems to say: YES! I'm adding it to my PCI Latency Patch. But, the hard part is, how will my patch detect whether PCI Bus Parking is supported in your chipset?
|
|
|
|
|
Logged
|
|
|
|
|
nelamvr6
Regular
Posts: 31
Join Date: Sep, 2002
|
|
« Reply #4 on: November 29, 2002, 01:03:51 PM » |
|
Thanks for the reply George. I'll keep looking for updates of your patch, if it can be done I'm sure you're the one who will figure out how!  |
|
|
|
|
Logged
|
|
|
|
|
Guest
|
|
« Reply #5 on: November 29, 2002, 02:17:38 PM » |
|
Could you be kind enough to tell us the relevant bits to adjust in wcpredit?
It would be very greatly appriciated here for sure, I looked at the early kt400 pcr files certainly very usefull but they dont included these settings unless im mistaken.
I may not be correct but it should be any board with the VT8235 that support parking but you probably know this anyway, is it possible to get this info from the board?
Thank you for youre time.
|
|
|
|
|
Logged
|
|
|
|
|
GeorgeBreese
Ace
Posts: 2,088
Join Date: Dec, 2001
|
|
« Reply #6 on: November 29, 2002, 11:13:01 PM » |
|
My PCI Bus Parking tweak might not be complete yet. Here's what I have so far. FYI, all of this was done without access to VIA datasheets.
1) In WPCREDIT, set 75=08 and 76=50 -- IN THAT ORDER. The system will freeze if bit 6 of register 76 is set while 75 is not.
2) Yes, all indications are that all VT8235 chips will support PCI Bus Parking.
3) I wrote VIA to ask how to detect support for PCI Bus Parking in present and future chipsets. I received a short answer, "Rx76 bit 6". I suspect that my question was lost during a transation to Chinese and back. (I have the same problem with Japanese translation in my current contract job.)
4) At this time, I believe that I can detect the presence of a VT8235 without any special tricks. Scan the PCI bus, and look for the ISA bridge. Its ID will be 1106:3177 for a VT8235, or 1106:3147 for a VT8233A or below. (In WPCREDIT, press Ctrl+D to scan the PCI bus.)
5) Since 50% of my paperwork with VIA is done, I can receive datasheets from them but I can't say what is in those datasheets.
|
|
|
|
|
Logged
|
|
|
|
|
Guest
|
|
« Reply #7 on: November 30, 2002, 04:06:05 PM » |
|
Thank you very much for taking the time to reply.
It seems that Abit`s latest bios for the KD7`s also enables PCI parking, they have added a bios setting "Enhance PCI performance" wich changes the values 75=01 and 76=10 to 75=09 and 76=50
Im not sure if the diffrence between 75=08 and 09 is sgnificant, but i do seem to be getting a fair improvement in disk performance particullarly compared to older bios`s.
|
|
|
|
|
Logged
|
|
|
|
|
DawnWolf
Green
Posts: 16
Join Date: Feb, 2002
|
|
« Reply #8 on: November 30, 2002, 09:24:07 PM » |
|
Excellent. Thanks, George! I do hope most manufacturers will make this a BIOS option, but it is very reassuring to see that it can be done through WPCREDIT and that you'll have your patch enable it, too, if a VT8235 is detected. I feel much more secure now in considering a board with a VIA southbridge for an upgrade. Shawn |
|
|
|
|
Logged
|
|
|
|
|
