To help keep things moving at a quick
pace, the SIS 655FX places six request/response channels (three for
each channel) between itself and the SIS 964 Southbridge. That is a total
of 1.2GB/s of bandwidth between the two core logic chipsets. With traditional
chipsets, there are just two channels (one upstream and one downstream),
so all data, no matter how trivial, is treated equally and passed
through with the same priority.
With HyperStreaming, not only is the
data being computed in parallel, certain data streams receive higher
priority, and are thus worked on first and have more channels between the
bridges. For instance, HDD performance of the SiS964 is among the highest
PCstats has ever seen thanks to the fact that hard drive streams
have their own independent channels to the Northbridge.
As with life, not all
computer transactions are created equal. SiS is a bit vague
describing Specific Stream with Prioritized Channel, but it
seems with the HyperStreaming, the data that is most critical to the system will
be computed first while less essential data will have to wait. For instance SiS
claims that HyperStreaming will give access higher access to high bandwidth, low
latency programs such as video/audio (which can suffer sync problems) and
internet accesses. All priority transactions are tagged
with an isochronous bit and then get to pass through the isochronous channel
while the rest of the non essential transactions like keyboard strokes,
printer commands and mouse movements run through the non
isochronous channel.
Think of it as an express highway with
its own exit points. The express data doesn't run any faster than the collectors
(non iso), but as it's less congested, the pieces of data end up getting to
their final destination faster. When HyperStreaming
begins to work with another architecture based on different protocols,
efficiency and transition between the two becomes very important in order to
avoid redundancy. The first part of Smart Stream is Intelligent
Arbitration. This feature prioritizes
streams according to their data types and readies the data to be sent.
Non-post and response request are higher
up on the food chain than Post requests because non-post and response requests
use more aggressive insurances and directly reduce the wait state of the
processor and other devices. Smart Stream allows for
a higher cache hit rate and if that fails system memory hit rate is also much
higher. HyperStreaming allows for two levels of intelligent arbitration to
schedule commands and data, and improves bus utilization and
efficiency.
Smart Flow Control is
used to improve transfer rates between the processor and chipsets/devices. With
HyperStreaming, no matter how complex the commands are they only use one cache
line. If you'd like to
learn a bit more on SiS's HyperStreaming please check out SiS' informative little flash
movie. Now let's take a look at how the SiS 655FX
chipset performs against the competition.
RSS Newsletter
