• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

PCI Latency Timer

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Foxie3a

Normal Member
Joined
Sep 7, 2003
I saw the option in my BIOS and decided to read about it. I think I learned something out of it, but I have a few more questions.

The timer is measured in clocks, right? Lets say it's at 32 clocks.

The timer is used to make sure each device gets a fair share of the PCI bus, right? So for every 32 clocks, it switches to another PCI device unless if there is nothing waiting in line then it can take another share.

So if I'm at 3ghz and it switches every 32 clocks, there will be 93 shares in that second.

Is it trying to get a share so that it can go to the NB(to use cpu ram or whatever)?

I guess it doesn't seem like a lot considering how much resources stuff uses, but there isn't a limit on how much it can use in those 32 clocks, and there are buffers.

What am I missing here? I don't fully understand this.
 
Yes, the PCI Latency Timer measures how long a device can hold te bus and the time is measured in clock cycles, but the PCI bus runs at 33MHz, not the speed of your cpu. I also think your math is off 3GHz/32=93750000, not 93.
33MHz/32=1031250. Remember Giga means 10^9

The PCI Bus is shared, so only one device on it can transmit data at once. The PCI Latency TImer controlls how long each device has control over the bus before another can be switched to. Remember, sound cards, IDE controllers, NIC and other devices are all on the PCI bus, some need to send data to the CPU, some directly to ram. When they have control over the PCI bus, their data goes to the PCI controller on the SB, then through some sort of link to the NB (V-link on Via chipsets, HyperTransport on Nvidia Chipsets) where it goes into memory (if it's a dma transfer), or directly on to the CPU. The latency timer keeps one device from tying up all of the PCI bus, when there are other devices that need to use it also.

I guess I'm not really sure what your question was, I hope I answered it.
In case I didn't, some links:
http://www.reric.net/linux/pci_latency.html
http://www.rojakpot.com/default.aspx?location=8&var1=0&var2=138
http://computer.howstuffworks.com/pci.htm

PCI specs:
http://www.pcisig.com/specifications/conventional/
 
i have found lower latency makes a system feel more snappy...
however things like HDD's can and will suffer if the value is set to low
 
Back