As a FYI, I usually include a Specs/Design page in my CPU upgrade reviews that lists the part number (speed rating) of the CPU chip and Cache chips on the upgrades I've reviewed. The only Sonnet CPU upgrade I have personally seen however is the Duet Dual G4/500 I installed in my Cube (which did have 500mhz rated G4s on it). I have not however seen other models of Sonnet upgrades to examine personally.

I have written to IBM's tech support for their comments on what their policy is on running high temperature CPUs at higher than the marked/rated speeds. (Many readers, and myself also have run CPUs at higher than rated speeds by choice, but that is not really the core issue here in my opinion.) BTW: One thing I noticed in all these photos is the black stripe under the CPU markings. I have not seen this personally in any IBM G3. The photos in the reviews here don't have any CPU with this sort of marking/stripe.

Update: Here's a photo from Sonnet of the remarked "333" Mhz CPU that shows a G3/450 marking also. (This photo is of a board/chip before final assembly, where the top strip over the contact area of the CPU would be removed. I have never seen a CPU chip with a strip like that over the CPU.) Looking closely at the black stripe area you can see the impressions of a "450" mhz rated marking under the black stripe which was remarked as a 333. Of course some may say this is a Photoshop enhancement (I do not see any similar marking in this OWC closeup photo of a board from stock). I wish I had an actual card to examine personally. I'd be tempted to try and remove the black stripe from the CPU chip personally. (Again I have never seen this sort of stripe/remarking on any CPU chip before.) Sonnet how has updated their FAQ to comment on the issue of remarked CPUs and selling G3/466 hi-temp chips as G3/500 upgrades.

(from OWC's email)

" **************************************************
Does a Sonnet G3/500 really have a G3/500?*
***************************************************

It was standing in our MacWorld booth that I learned something disturbing concerning Sonnet and processor speeds. Responding to an anonymous insider tip, we did a closer inspection on the Sonnet Upgrade models we had in stock and found that the Sonnet G3/500 PCI and G3/500 L2 Sonnet upgrades actually had 466MHz marked IBM G3 processors and even more of a stretch, the Sonnet G3/450MHZ PCI had a 333MHz marked IBM G3 processor.

I approached an available Sonnet manager on this issue and she did not seem to have any knowledge of this practice, but did say she would find out what was going on. I expected it to have been an error and was completely shocked when she did get back to me saying it was normal and that they had high-temp rated chips and that was why they could use them for a faster speed than their markings.

Most people would never know anything of this since the processor is hidden under a secured heat sink on the Sonnet models in question. While XLR8 and PowerLogix have made the processor easily accessible, Sonnet has advised removing their heatsink voids the warranty. Is there a consumer advantage to the parts Sonnet is using? I am not aware of any. Do I think many people are surprised by this? Possibly - I sure was! Was there some sort of advantage to Sonnet? There must be or they would have stuck with processors marked to the speed they are selling the upgrades running at. As a result of this, we have suspended the sales of the affected items. It just doesn't seem right to me that you get a true G3/500 processor on XLR8's 500MHz upgrade for $299, you get a true G3/500 processor on PowerLogix's 500MHz upgrade for $229, but you get a G3/466MHz on Sonnet's 500MHZ upgrade for $299. And don't even get me going on the 333MHz marked processor we found on their G3/450MHz upgrades.

OWC has sold Sonnet Product since about April of 1998. I think Sonnet has been a real asset to the Macintosh community and I do wish them continued success. This was information the consumer should now and Sonnet should have shared it, not hidden it. They barred us from their MacWorld event upon our publicizing the information, so clearly it would seem it was something they didn't want the public knowing and points with them we definitely did not earn. Sonnet makes some excellent upgrades and we will to sell Sonnet, but only those upgrades that do not have the lower speed marked processor in comparison to the speed of the upgrade itself.

See the products we won't sell and pictures of them and the processors here:
http://eshop.macsales.com/Search/Search.cfm?Column=Description&Criteria=ADVISORY

About 4 or 5 years ago, there was a case with XLR8 (prior to their current ownership) selling an upclocked upgrade. They got called on it, and have never since sold an upgrade that has a processor other than what the upgrade itself is to be running at. We have confirmed that all of our XLR8 and as well our PowerLogix upgrades have at least a processor of the speed of the upgrade speed being sold. I am not sure what Sonnet calls running a 466MHz G3 marked processor at 500MHZ or a 333MHz set to run 450MHz, but hopefully their official release will answer that. At the party we were barred from, they did say that they would have an official release - that was a week ago tonight and I eagerly await it. this, because they really is no need for something like this to arise.

If you have a Sonnet card with a slower marked processor and would like to report it, please e-mail simplyslower@macsales.com starting 10AM CST tomorrow. Picture files may be included.
[end of OWC's Comments] "

Sonnet's FAQ On Remarked CPUs and Running 466Mhz marked CPUs at 500MHz: Sonnet has updated their Support page, CPU Upgrades FAQ with their replies on the issues of remarked CPUs and high temperature 466mhz marked parts sold as 500Mhz upgrades.

IBM's Tech Support Comments: I received an email from IBM PowerPC tech support with comments on running hi-temperature rated G3/466 CPUs at 500MHz.

" Hello Mike :-)
Sonnet posted a reply to your concern on their FAQ. [See Sonnet's CPU Upgrades FAQ-Mike] The statements that they make about these IBM parts are correct.
IBM customers, reps, disti's, etc with PowerPC technical questions can visit our website, or send questions to ppcsupp@us.ibm.com.

The following is a reply to further questions I had which are in bold/black. Dale (IBM PowerPC tech support) replies are noted below the questions. -Mike

[Mike] Dale, I have looked at every IBM G3 spec - nowhere do I see that IBM recommends/suggests that a high-temp part can be overclocked if run at less than the max temp rating.

[Dale] That makes sense. This level of information about our parts is of no use to anyone except PowerPC design engineers, who know how to access the required information & processes.
Also, can we define overclocking as running the CPU faster than spec, so that the guardband is reduced? This makes more sense in a situation where one part number could have several max speeds depending on the application conditions. Under this definition, running this particular part number at 500/65C has the same guardband as running it at 466/105C, so it is not overclocking. In other words, there is as much margin in one application as the other.

[Mike] I do see notes that a CPU should be derated if run at higher temperatures (i.e. if a 366 65C part is run at 85C derate it to lower speeds) - but where is there an IBM spec that shows the reverse is true? [That IBM notes that is acceptable]

[Dale] First, only certain parts are specified in this way, and only in certain applications. Note that any derating information that you have seen is part number-specific. The rest of the process is IBM Confidential, so we cannot discuss it without a Non Disclosure Agreement.

[Mike] If true, isn't this an OK from IBM on overclocking G3s (running at higher than rated speeds) as long as the junction temperature is kept below the max chip rating?

[Dale] Not at all. IBM sells our parts as described in the Datasheet for the part. The PowerPC design community has access to additional information that allows them to work with IBM PowerPC Applications Engineering to qualify particular part numbers in uprated applications under certain conditions. But our offering to the general public is limited to Datasheet specification.
If you have additional questions, let me know.
Dale Elson,
Team Lead PowerPC 60x/7xx Applications Engineering "

Based on a recent reader email regarding this spring 2001 purchased Sonnet G3/500 CPU card, these remarked/high-temp 466s must be a change since then, since he sent an email with a link showing a G3/500 CPU on his Sonnet upgrade. (but note the black stripe again - I have never personally seen an IBM G3 with a black stripe under the CPU's marking/rating. Photos of the IBM G3's I've reviewed are in my CPU upgrade review's specs/design page.) His email from Jan 20th follows:

" Hello Mike,
I just read the news at your page. And remove the Heatsink of my Sonnet PCI G3 500 very carefully.

This is the same card thats in your data base dated 3/15/2001 that i install on my 8500.

Anyway feel free to copy the picture and info as you see fit.
see this page for his photo of his CPU chip
All the best
Ponciano
I asked when he bought the upgrade and he replied
Ordered it first week of January 2001.Arrived Mid February 2001. Ran it for a few weeks before submiting review on 3/15/2001 "

His photo of the card he bought last year shows a G3/500 marked CPU. (Unlike the photos linked above taken from Sonnet upgrades in OWC's stock this month.)

BTW - I received an email (under an admitted fake name via a hotmail account) from a claimed 'reseller/competitor' of OWC's calling Larry basically 'the devil' and saying he should have never went public with this, etc. Why, (for legal reasons and being honest to customers) would a dealer of these CPU upgrades keep quiet on this? (For other than the obvious profit motive - OWC sells many Sonnet upgrades and clearly they didn't profit from posting it.) They could have "kept quiet" on this issue and not burned any bridges. If profit was their motive, then why would they have made this issue public?
And to the "reseller" that sent the email under a (admittedly) fake name - at least I posted both sides of the story and *went the extra mile* to actually write IBM tech support for their comments on the issue for posting.
Here's a summary of things I have sent to the very few readers [two so far, and one was from a 'competitor'] that think OWC should have never publicly noted this issue and/or think I should have never mentioned it here.

1) OWC did discuss this with Sonnet, at the expo and reportedly via phone before going public. (Based on emails, phone messages, etc. from OWC) I also wrote Sonnet about this on the 10th, the day I received the info.

2) I'm not a lawyer, but the issue may also be a potential legal one. For a reseller/dealer to know about 466 marked parts on a 500MHz labeled upgrade and continue to sell them without a disclaimer (or explaination) may have legal implications. If all OWC cared about was profit, they could have kept quiet. I think it's going to have a negative effect (long term) on OWC's business, as it's almost certainly burned any bridges with Sonnet I suspect. (OWC sold Sonnet's PCI controller cards as well as their CPU upgrades.) And to the 'competitor' that wrote under a fake name, OWC does not have any equivalent Mercury line of cards to the models that were noted. There is no OWC L2 cache slot upgrade, nor any OWC Mercury branded G3/500 or /450 CPU card - only ZIFs and no Sonnet ZIFs were listed in the 'alert'. )

3) Like any dealer, OWC makes profit from selling upgrades - if their motives were profit only, they would have never went public on this, since they stopped any sales of the affected upgrades. They also incurred labor costs for opening inventory and removing heatsinks, taking photos, updating web pages, etc.

The real issue in my opinion is honesty (and perhaps legality).
Would I personally be concerned about running a high-temp G3/466 marked part at 500Mhz? No, not if it was reliable.
Would I be happy about not being told the CPU was a high-temperature 466 marked part? No.
Without this info being public would the typical end user be upset if they saw the 466 marked chip and didn't know about the details noted above? Probably.
If nothing else this has at least cleared the air on the issue, with comments here from IBM on these CPU chips. I posted both sides of the story here and spent time to personally check IBM G3 specs on the web, and also took time to personally contact IBM about this issue and post their comments (above). To not post it would be seen as trying to cover something up.

I don't like being in the middle of this, but I think if I were a dealer I would also have done the same thing -for legal and ethical reasons.-Mike