I agree, AMD is waking up. The phenom II looks impressive.
However...
Intel's i7 is a better quality cpu to build around for most every application.
It takes 2 4870's to perform near the same level as a single gtx280
The 2 dinosaur companies came up w/ 6Gb sata first, but everyone else will be right behind.
You shouldn't blame WD for having the fastest and most reliable HD's. I thank them.
The SATA/6G standard is needed to eliminate the
channel bottleneck when solid-state devices are
connected to SATA controllers.
I include both flash memory and SDRAM in the term "solid-state devices".
Assuming a worst-case controller overhead of 20%
of rated bandwidth, 4 x SATA/6G channels in RAID 0
are theoretically capable of a combined bandwidth
of 4 x 600 MB/second = 2,400 x 0.80 = 1,920 MB/second.
Call it 2 GB/sec.
Now we're talking greatly improved file system performance,
via standard SATA ports on standard motherboards.
Moreover, compare the above with the theoretical bandwidth
possible with something like OCZ's new "Z-drive", which
uses an x8 lane PCI-Express slot:
8 lanes @ 250MB/sec x 0.8 = 1.6GB/second (in one direction, e.g. READs)
Or, if the Z-drive's controller is measurably faster, use 0.9 controller efficiency:
8 lanes @ 250MB/sec x 0.9 = 1.8GB/second (in one direction, e.g. READs)
However, as seen from the review above, the Z-drive is doing READs
presently at only 600 MB/second. This may see FAST, however it is
not very close to real channel capacities.
So, the decision to go with a device like the Z-drive,
as opposed to 4 x solid-state devices in RAID 0,
will likely turn on the bottom-line cost to achieve
comparable performance.
Where do you think the bottleneck is in this new OCZ Z-drive?
My guess is the flash memory chips, not in the controller
-- a Highpoint RocketRAID 3520:
In terms of price-performance, a P45 system with 16GB of RAM
or Core i7 system with 24GB of RAM, can be enhanced with a
single copy of RamDisk Plus at $50 - $100, resulting in file system
performance approaching 25,000 MB/second (i.e. Core i7 at stock settings).
Given the rock bottom prices of DDR2 presently,
a 16GB P45 system sounds like an excellent way
to deliver outstanding file system performance
when most recently used files are "cached"
in a ramdisk.
We have already measured 2.7GB/second READs
using a Corsair DDR2-800 that is now 3+ years old
to host a ramdisk using an earlier version of
RamDisk Plus!
For the cost premium of a Core i7 system,
a huge leap in file system performance is
obtained by dedicating an upper subset of
12-24GB of RAM to a ramdisk.
Your thoughts are most appreciated here.
MRFS
Message edited by MRFS on 03-10-2009 at 12:50:55 AM
@MRFS....I've actually used Superspeed Supercache and Ramdisk+ in my machine. My typical layout for using the 16GB I have is 4GB Supercache on the OS Drive, then I create 8GB partitions on my "Program" array and I'm able to at will clone any of them to RAM on the fly. That leaves me with only 4GB left for the system. It's just ridiculously fast. In IE, Office, opening PDF's and general use there is no perceptible delay between "double click" and fully opened window. Unfortunately, it is an addiction and I'm jonesing for more RAM
Awesome video link BTW...
------------------------------I always have to learn the hard way, it's just not enough fun doing it the right way.
Reply to ShadowFlash
You are about to answer a thread that has been inactive for more than 6 months. If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.