MORE...

• NEWS
• PODCASTS
• BLOGS
• SYSTEMS
• MULTIMEDIA
• MISCELLANY
• ARCHIVES
• ABOUT US

My post yesterday about the new Celerons' lackluster performance in some tests generated quite a bit of interest, so I'm pleased to be able to follow up on it. To recap, I had speculated (with prompting from an observant reader) that perhaps Intel's new Celeron processors based on the Coppermine core are somehow rigged to run slower than the Coppermine-based Pentium IIIs. We know the Celerons have half the L2 cache and a 66MHz stock bus speed, but those differences didn't seem to account for the performance differences reported at some web sites. Or did they? Mad discussion ensued.

Dave at ComputerNerd was kind enough to run some benchmarks to help shed some light on the situation. He grabbed a Coppermine Pentium III and a new Celeron off the shelf, slapped 'em each into an Epox KP6-BS board (one at a time, though it is dual-slot board). Then he set the bus speeds as follows:

	Celeron	Coppermine
Multiplier	8.5	5.5
Bus MHz	66.6	103
Nominal MHz	566.7	566.5
Reported MHz	568	567

So what we're getting here is a test of these two CPUs running at almost the exact same clock frequency—at very different bus speeds. I'm not sure these numbers will prove conclusively whether or not the Celeron II is crippled in some way beyond what me know about it (the smaller L2 cache size and lower default bus speed). Ideally, I'd like to see a Celeron 566 pitted against a Pentium III 850E, with the two CPUs running at the same bus speed, in order to determine whether the Celeron is somehow throttled. However, I do think these tests shed some light on the situation.

With that said, let's look at the benchmark results, which I threw into graphs for easy reading:

The scores are very close together here. The Celeron II certainly doesn't seem throttled in the least in these tests.

The cache and bus speed differences manifest themselves in Wintune's "Advanced" tests. The numbers stack up rather neatly, showing that both cache size and bus speed affects performance. At least, that's how I read it. Now let's look at the memory test...

Once again, the Celeron runs very close to the Coppermine both with and without the L2 caches enabled. Although the Celeron is slower thanks to its lower bus speed, it appears to benefit from use of its L2 cache just as the PIII does.

As I read these numbers, the Celeron II doesn't look like it's been crippled beyond its slower L2 cache and—especially—its crazy low bus speed. But I'm sure some of you may read the numbers differently. (Heck, Dave did.)

Thoughts?