News, info, and opinion by Mac users, for Mac users.

April 28, 2008

news

3GHz iMac’s CPU overclocked, not really next-gen

Posted Apr. 28, ’08, 4:00 PM PT by David Dahlquist
Category | Apple » News

overclocking.jpgFor those of you frothing at the mouth for the new 3.06GHz iMac, we have some news that you may find a bit disappointing. It turns out that the new iMacs are part of a special run of Intel’s existing technology rather than an early introduction of the anticipated Centrino 2 technology.

Essentially, the 3.06GHz processors are current generation chips modified to take in more power (55 watts instead of the typical 35-45 watts) and to support the faster system bus speeds (1,066MHz up from 800MHz) so that they can be revved up to the impressive 3.06GHz Apple is advertising.

This is a big difference from the upcoming Centrino 2 processors which are expected to consume 25W to 35W and to achieve the faster processor speeds through all natural, organic, free-range methods rather than through barbaric, unnatural overclocking. Either way, a 3.06GHz processor is still a 3.06GHz processor regardless of how the speed is achieved, and that’s fast, so we’re not complaining!

[Via Electronista]



8 Comments

Goobimama said:

I wonder how much noise this thing will make then. Suddenly it doesn't seem like an excellent deal (which it still is!).

April 28, 2008 5:06 PM
wesg Author Profile Page said:

...as long as the iMac doesn't melt through the desk.

April 28, 2008 6:53 PM
josh said:

I thought the "old" 2.8Ghz processors were actually overclocked 2.6Ghz processors. So wouldn't that mean these are new chips, just overclocked as well?

April 28, 2008 9:11 PM
Anonymous said:

Isn't this blog affiliated with Macworld.com? If so, you should ask Intel about the processors. I don't think these parts are overclocked, Electronista may be using the term in a loose way.

http://forums.macrumors.com/showpost.php?p=5373072&postcount=643
http://valid.x86-secret.com/show_oc.php?id=353588

If this poster at Macrumors really ran CPU-Z on the new iMac, and if the app can identify this particular CPU, then it's an Intel Core 2 Duo E8435, a Wolfdale part (E8000 series) intended for desktop computers. It can operate at 3+GHz and support a 1333MHz front side bus. But a desktop part should have a different socket, not Socket P, the mobile processor socket.

April 28, 2008 10:18 PM
Anonymous said:

I have been waiting for the upgrade and the new chip design that has been reported- the Montivine- for more than one year. The tech specs were that it had a front end bus of 1066 and speeds up to 3.0 GHz. When the Monday announcement came, I placed my order. Only in the evening did I see the story that the chip is not the next generation chip I had previously read about and expecting in June. I just thought Apple moved up the release of the next generation chip.

Is this an issue of truth in advertising? Acts of omission?

Other than pulling less power, and therefore running cooler, are there other differences in the chip design? Is the next generation chip really going to come in less than 2 months?

Is there any advantage to waiting?

April 28, 2008 10:56 PM
MacClassic said:

The new iMac chips are 45nm penryns so they are in fact 'next gen." The next generation chips won't deput until the Nehalem platfrom drops at the end of the year. The Montevina platform due in June wouldn't provide any additional performance benefits as this iMac revision includes all of the important features the forthcoming platform: 1066mhz bus, 800mhz RAM and 6MB L2 cache. Power consumption is not important in the iMac as it's not a laptop. Remember, Intel has started this tick-tock cadence meaning 1st we get a new platform and then new chips follow. EX: We got the Santa Rosa platform with Merom chips. Then, Penryns replaced Meroms on Santa Rosa (current MBs & MBPs) and soon Penryns will move to Montevina.

April 29, 2008 8:02 AM
molintorch said:

If this poster at Macrumors really ran CPU-Z on the new iMac, and if the app can identify this particular CPU, then it's an Intel Core 2 Duo E8435, a Wolfdale part (E8000 series) intended for desktop computers.


That was my post, and here are links to the other benchmarks I ran for people. As far as noise goes, I ran 3dmark three times, and never heard the fans kick on. I have not had time to install anything CPU/GPU intensive yet, but will tonight and then see how the sound is.

8248 3Dmarks with default bootcamp drivers
SM2.0 score: 3501
HDR/SM3.0: 3328
CPU Score: 2769
xBench: http://db.xbench.com/merge.xhtml?doc2=286743
GPU-Z: http://www.techpowerup.com/gpuz/43pxg/
CPU-Z: http://valid.x86-secret.com/show_oc.php?id=353588

April 29, 2008 9:43 AM
D'bug said:

I think it is irresponsible to use the term "overclocked" to describe what Apple is doing. Some might even call it slander.

Overclocked simply refers to operating beyond the clock rate a chip is tested/certified/sold to run at. Make no mistake about it, anything Apple is providing here is with Intels' blessing.
Apple and Intel have worked closely together and that can produce new variations of a CPU. For example watch Apples' Quicktime stream of this years' MacWorld keynote where the MacBook Air was introduced. They used an Intel CPU in a different package than previously offered. It is really so surprising that Apple could go beyond mobile-class performance in a compact desktop?

Whatever evolutionary stage the new CPUs in the iMacs are at, there is no evidence to indicate or suggest that they're something not fully supported by Intel. Customers are not getting out-of-spec components.

Note that the FSB speeds of earlier CPUs were not an inhernet limitation in the chip, but a design power/heat decision for chips used in mobile equipment. Essentially the same chips certainly could handle higher FSB and operating rates if the combination of thermal resistances from chip to package to heatsink to internal air to ambient are low enough to dissipate the power used at a given speed. Even then there is some flexibility since the C2D series CPUs support thermal throttling - automatic slowdown if they're getting to toasty. Power consumption scales up with speed for a given chip design. Apple certainly is known for attention to design detail, so it's a safe bet that all design tradeoffs between speed and power/heat underwent careful review by the engineers involved.

I've used an early C2D CPU in a homebuilt desktop and had no trouble overclocking by raising the FSB bandwidth from 1066 to about 1300 MHz.
The CPU actually could have been sped up by a much greater percentage, but I was limited by the memory controller chip on a very modestly priced motherboard.

Apple going beyond mobile chip performance in a desktop doesn't require anything unreasonable or unsupported. Of course if other cost effective chips offer similar performance at lower power levels we can expect Apple will again migrate to save energy, possibly enhance reliability (from lower operating temperatures) and maybe pick up a little more speed too.

The type of overclocking that did warrant some public outcry was that done by some G3/G4 CPU upgrade vendors for older Macs with replaceable CPUs or daughtercards. Some of those were using chips beyond the vendor-marked ratings. Note that most worked just fine, but users expecting to overclock what they bought generally found there wasn't more performance available to be had.

Instead of condemning Apple on unfounded speculation, we should be happy to see enhanced performance with no change in pricing. With the weaker dollar we're lucky the prices haven't gone up.

April 29, 2008 12:30 PM

Leave a comment

 



Visit other IDG sites: