 |
 |
Is 3.0Ghz the top of the hill at the moment?
|
 |
|
 |
|
Professional Poster
Join Date: Jan 2001
Location: Salt Lake City, UT USA
Status:
Offline
|
|
I poked around a bit, and I haven't found any chips with a higher clockspeed... So I'm kind of wondering if, we're finally back where we were in the glory days, where we had clock speeds to match those on the other side of the platform fence...
Are there PC makers sticking faster chips in their machines?
|
|
2008 iMac 3.06 Ghz, 2GB Memory, GeForce 8800, 500GB HD, SuperDrive
8gb iPhone on Tmobile
|
| |
|
|
|
 |
|
 |
|
Mac Elite
Join Date: Aug 2001
Location: Near Boulder, CO
Status:
Offline
|
|
intel is the king (arguably) of the PC market, the woodcrest is the top dog at the moment... the mac now has the top of the line as far as processors go.... apple built a pro machine, they would naturally put in the best of the best.
Zach
|
|
|
| |
|
|
|
 |
|
 |
|
Posting Junkie
Join Date: Oct 2005
Location: Houston, TX
Status:
Offline
|
|
3Ghz is the fastest Core microarchitecture (Xeon 51xx sequence or Core 2 Duo) chip available.
A high-end Opteron may win a few benchmarks here and there, but the 3Ghz Xeon wins the majority.
|
|
|
| |
|
|
|
 |
|
 |
|
Moderator 
Join Date: Sep 2001
Location: Arizona
Status:
Offline
|
|
Apple/Intel are likely going to remain at the top of the heap into next year, until AMD's K8L debuts. K8L, along with Intel's Clovertown, will be 4-core CPUs. But AMD's core implementation will be a bit better and their HyperTransport bus is more suited to supplying sufficient bandwidth to those four cores than Intel's NetBurst bus is.
|
|
I like chicken
I like liver
Meow Mix, Meow Mix
Please de-liv-er
|
| |
|
|
|
 |
|
 |
|
Mac Elite
Join Date: Nov 2005
Location: Seattle, WA, USA
Status:
Offline
|
|
There are P4 processors clocked up to 3.4 GHz, but they pale in comparison to even the 2.66 GHz Core 2 Duo processor. The 3.0 Core chip is definitely the fastest intel processor available, and will probably win out against AMD's best more often than not.
|
Any ramblings are entirely my own, and do not represent those of my employers, coworkers, friends, or species
|
| |
|
|
|
 |
|
 |
|
Posting Junkie
Join Date: Oct 2005
Location: Houston, TX
Status:
Offline
|
|
Originally Posted by Lateralus
Apple/Intel are likely going to remain at the top of the heap into next year, until AMD's K8L debuts. K8L, along with Intel's Clovertown, will be 4-core CPUs. But AMD's core implementation will be a bit better and their HyperTransport bus is more suited to supplying sufficient bandwidth to those four cores than Intel's NetBurst bus is.
They (AMD's 4 core and Clovertown) each have their ups and downs. Clovertown has the FSB potential bottleneck (see below) and is really just two dual core chips on a single package.
HyperTransport is nice, but the NUMA architecture also has it's downsides; if one core needs to access memory that is local to a core on a different chip, it has to traverse several HT links to get there. Also the L2 caches are still per-core, not even shared among two cores.
Anandtech compared performance at 1066Mhz and 1333Mhz FSB in about a dozen real-world apps to see what impact FSB had. The average gain was 2.4% by increasing the FSB 25%, with the best gain being 7.5%.
I expect the difference to be a bit bigger with 4 cores per socket, but the current architecture doesn't look to be bandwidth starved.
|
|
|
| |
|
|
|
 |
|
 |
|
Moderator 
Join Date: Sep 2001
Location: Arizona
Status:
Offline
|
|
Originally Posted by mduell
They (AMD's 4 core and Clovertown) each have their ups and downs. Clovertown has the FSB potential bottleneck (see below) and is really just two dual core chips on a single package.
HyperTransport is nice, but the NUMA architecture also has it's downsides; if one core needs to access memory that is local to a core on a different chip, it has to traverse several HT links to get there. Also the L2 caches are still per-core, not even shared among two cores.
Anandtech compared performance at 1066Mhz and 1333Mhz FSB in about a dozen real-world apps to see what impact FSB had. The average gain was 2.4% by increasing the FSB 25%, with the best gain being 7.5%.
I expect the difference to be a bit bigger with 4 cores per socket, but the current architecture doesn't look to be bandwidth starved.
Indeed. As I stated in the other thread;
Originally Posted by Lateralus
Yes, Clovertown is supposed to be a drop-in replacement for Woodcrest.
I'm not particularly enthused about it though. For starters, it isn't a true 4-core chip. It's essentially two dual-core dies rigged onto the same processor, each having their own caches rather than sharing one. Basically, it's a repeat of Intel's initial move to dual-core with Pentium D 8xx line. Two separate cores with independent caches occupying the same chip. It's a nice quick and dirty solution, but as far as performance efficiency it's far from the perfect solution.
Beyond that though, Intel's current bus architecture for the Core is derived from the NetBurst/Pentium 4 family, and it is sincerely lacking in a number of areas. It is adequate for the moment but it doesn't provide enough throughput to support four cores effectively. So you'll see performance per core drop with the debut of Clovertown.
Intel's next generation bus, dubbed CSI, looks very promising but unfortunately it's still a year or so off. But in my mind, it's the next big thing to look forward to on the Intel front.
|
|
I like chicken
I like liver
Meow Mix, Meow Mix
Please de-liv-er
|
| |
|
|
|
 |
 |
|
 |
|
|
|
|
|

|
|
 |
Forum Rules
|
 |
 |
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
|
HTML code is Off
|
|
|
|
|
|
 |
 |
 |
 |
|
 |
|