Parallel Processing, Part 1: CPU Cores

Clock speed is only one part of the story. More and more PC components are laid out with multiple units to increase performance via parallelism. In part 1 of this article series, we compare single, dual and quad cores CPUs using identical architectures and clock speeds.

http://www.tomshardware.com/2007/1 [...] index.html

While i agree that it would have been interesting to see the performance of AMDs offerings in the comparisons too, it would have shifted the focus of the article.
In addition it does not make sense to compare intel and amd clock for clock. Their products are offered at different prices and even their platform prices are different. The components would need to be comparable too, i.e. Memory, chipset etc. which is simply not possible. It would be like comparing vehicles by horsepower.

yeah - not many people have a clean install ... that's for sure.

Interesting article but no real surprises I suppose. If anything, it just highlights the need for more apps to be multi-threaded. I believe that the conclusion of, ": if you can get a quad core of similar clock speed for only a little premium over the dual core, go for it; if not, stick to a reasonably priced dual core." to be spot on. My opinion though, I still say quad core simply for the sake of having a quad core (ahem...braggin rights) is a waste of money, no matter how "cheaply" you can buy one.; moving into '08 & '09 may be another story. However, I think that is largely up to Developers.

I agree that it would make no sense to have AMD procs in the mix. It's still hurry and up wait for Phenom and until it's up to speed and readily available, using comparable dual/quad cores from Intel are the best baseline for an article like this.

As side note and as an opporunity to Vista bash, I would like to have seen the testbed run on Vista 32 and/or Vista 64 rather than WinXp SP2. One of the much hyped "advantages" of Vista over XP SP2 was better multi-threading support and the supposed ability for Vista to load balance across multiple cores. Oh well, maybe THG will introduce Vista as the test OS in the subsequent parts of this article.

I wish they had benched a game that takes advantage of multi threaded cpu's to see how much of an increase you get over a single cpu.

I run a dual core and when I am gaming I turn off any extra applications so they won't interfere with the game. Those people who play games while running a virus scan, down load music, burn a cd, etc. etc.... need to be checked by a psychiatrist for ADD. But I do agree they could have at least opened up a few windows and apps and then tested them to show the impact. Some how these articles always seem to always come up short in one department or another.

Where are the AMD's?

True and we didn't see any test on differences in programs exiting back to windows or program startup. How much faster is program startup and shutdown on quads?

Why do I feel that... Toms has became a lot more.... n00b proof?

I'll say this much about startup speed and shutdown since I have a Q6600. I am running Vista and after Post and the loading screen Vista is up pretty much in about 2-3 seconds with almost every progam up and running. Shutdown is roughly 5-7 Seconds.

Total startup time without Post is about 15-25 seconds(that includes the Vista loading screen.

I'll compare this to my old system which was a single core P4 EE running XP. Startup after Post and XP loading screen was about 10 seconds shutdown was 15 seconds depending on what I had running.

Without Post and with the XP loading screen it was roughly 45-60 seconds startup.

I personally think that a Quad core is worth the investment right now since you can get a Core 2 Q6600 for $277.99 where as a Core 2 Duo E6600 is $229.99 making it a $48 dollar difference and a even higher clocked Core 2 Duo E6850 is $279.99 or $2 dollars more.

So basically the investment is worth it since it will allow you to keep the same PC for longer without having to spend more money on a new CPU and Mobo when games/applications start to be enhanced for multiple threads.

I got all my prices at www.newegg.com and are current as of 10/08/07 so some may be lower/higher than I quoted if found on another site.

My basic principle is when you design and build a new PC try to make one that will last for at least 2 years instead of having to build a new one in 6-12months.

I for one love my Q6600. I OC'ed it to a 300MHz FSB(1200MHz) so it is running at 1.8GHz when idle and 2.7GHz when under load and it only runs at about 1c hotter than at stock speeds which is 31-34c idle and 40-44c load(if even that). That is a G0 stepping though and I still have plenty of headroom to OC with my Zalman CPNS 9700 which I recommend to get maximum air cooling for OC'ers.

come on what kind of an article is that? show me some power user tests! like running a benchmark while ripping an mp3 and decompressing a rar file couple that with some internet download and some youtube buffering and have another pc copy a file from it while printing and installing some software in the background.. phew. give me the ultimate CORE STRESS TEST cause im already stressed

dam, i did not see this i hate double posts!

well lock my thread!

this article has all the same amd bias as most cpu articles.

This is the kind of article we need to help choose between CPUs now. Despite the fact that it was Intel-only it helps you see where quads do and do not provide significant performance gains. That's what I wanted to know.

One nit: please keep each CPU type at the same place on each graph! Rearranging each graph to have the highest performer on top each time is really confusing and just makes it harder to see how a processor performs overall.

Arbie

I don't buy the Vista loading on a quad in 2-3 seconds thing. XP is just as fast on a P4 with a fresh install. The longer you run any OS, the longer it takes to startup and shut down. Thus, it's really hard to compare startup and shut down times from one system to the next, let alone attribute the difference to hardware.

with all this virtualization trend, what I would like to see is a setup up with some, lets say 4, virtual machines, and benchmarks running inside them.

Maybe I missed it, but I'd be curious to see how a configuration of 2 dual core proecessors (2x2) would perform along side a single quad core processor (1x4). Anyone know the answer or have any insight to this?

There is no Core 2 Solo processor, only Core Solo, which is based on Pentium M.

Technically speaking, you use "Core 2" to represent all Core uarch processors, that include Core 2 Duo, Core 2 Quad, and Core 2 Extreme.

Ooops my bad. I stand corrected.

What test?