Where's A_Pickle to defend Intel?

I have already sent an email to the author of the article (and to the Inquirer) with the following content:

You have completely missunderstood this issue.

Intel mentions in erratum AA30 that it is possible for BIOS to contain WORKAROUND. It is already included in ASUS latest BIOS-es. This workaround is actually a microcode update for the affected processor steppings. Workaround presumably disables EIST and C1E reporting via CPUID instruction so they don't get activated by the operating system.

This erratum cannot be fixed by the BIOS update, it will be fixed in new revision of B1 stepping cores which will be out in Q2/2006 with different S-Spec numbers, at least that is what Intel specification update document says.

As for how much this affects people:

Pentium 920 already runs at x14 multiplier and it cannot go any lower. Pentium 930 could go 200Mhz down (from x15 to x14) but that doesn't make any palpable difference. Only those chips with higher TDP could be affected by the lack of EIST.

As for C1E, it requires Windows XP SP2 installed to be used (newer CPU ACPI driver) and it affects only idle power consumption.

As for TDP, it says in Intel document that it is the _maximum_ number, not typical. It also says that individual chips may differ and have less than maximal TDP.

I am using Pentium 630 with EIST and C1E disabled even though they exist and work because I don't need them -- chip is cool anyway. New chips are even cooler than this one.

So to summarize, this may affect only those people who buy higher performance chips such as 940, 950, and 955. They should perhaps wait for the chips with new S-Spec numbers due to be released in Q2/2006. You can find those S-Specs on page 12 table 1 of the specification update document.

Regards,
Igor

P.S. Since most people who buy higher performance chips also overclock them, they obviously do not need those power saving features and may even consider them detrimental.

if you want to know how each defines TDP read here
http://www.silentpcreview.com/article169-page3.html

, one shop is saying that because EIST and other power-saving features are not fully functional with this initial lot of processors, the TDP (thermal design power) might go up as much as 10 - 15W.]

Pretty sure TDP is based on peak, not idle.

So this claim doesn't make alot of sense to me.

Intel's TDP is based on peak sustained power. And for Dothan it's a family TDP that's over-rated by about 25%.

Yee-haw, this story has been linked by The Inq...

http://www.theinquirer.net/?article=28841

:D

This stuff isn't related to the problems I experienced with the 955 XE getting too hot under load. The XE doesn't have EIST or the C1E halt state since Presler doesn't have the lower multipliers capable of bringing it down to more moderate speeds on a 1066MHz front-side bus.

I believe the problems that I ran into had to do with the thermal solution, the cooler retention mech, and possibly some flexing of the motherboard, so that the cooler didn't make solid, uniform contact with the CPU across the surface of its metal cap. It is still something of a mystery, even to Intel, who saw this problem on multiple review kits--not just mine--but not on all of them. I think perhaps those folks who used the TIM included with the thing and didn't switch around CPUs multiple times were OK. The TIM wasn't perfect for us--things still got very hot--but it did stave off (just barely) throttling under load in my thermal testing.

With the Zalman cooler and its very different retention mech, the 955 XE ran cool enough and consumed less power than the 840 XE under load.

Given that these errata affect processors with C1E halt and SpeedStep, the 955 XE should not be affected, FWIW.

Blaming the whole 900 series is abit harsh. Its true with the EIST support. But seriously..thermal issues...gimme a break.

Other hardware review sites are able to run even EE with stock HSF at 65degrees, 79 at 4.25Ghz...I´m sure you can do the same.

Sure its not the 40-50degrees like a 95W X2.

Any idiot would know the missing EIST features only affects idle operation.

It wouldn't be a netburst without some kind of thermal issues, right? It's kind of a tradition. Not one I'll miss, but a tradition all the same.

Edit: not to worry.

review on monday?