General Category > World News

New evidence of Chinese tampering with Supermicro hardware ‘found in US telecoms company’

<< < (2/2)

Smokin Joe:

--- Quote from: PeteS in CA on March 27, 2024, 08:09:29 pm ---1. This claim is patently absurd, for a very simple reason. Data signal lines in computer motherboards have been carefully impedance-matched since the days of the 80386 or 80486 microprocessors. They are transmission lines, not just conductors. At those clock frequencies a wire or PCB trace is no longer just a conductor. Because motherboards are finely tuned, matched to each component in each circuit, "planting malicious chips" into impedance-matched circuits would unbalance the signal lines and disrupt performance at best, and probably render the motherboard inoperable.

Further, modern server motherboards have control software that continually monitors power consumption. An extra "chip" that is not part of the motherboard's design would be detected quickly because of its added power consumption and the motherboard would be reported to the system or network as defective in some way, and the motherboard would probably be disabled.

I would not expect a "journalist" to understand this, but I would expect a real journalist to find some knowledgeable people to factcheck their expert's claims.

2. So, how well did Supermicro weather this storm? This quarterly revenue chart gives the broad picture, https://www.macrotrends.net/stocks/charts/SMCI/super-micro-computer/revenue . Supermicro is doing quite well.

In the years since this "planting malicious chips" claim Supermicro has built and brought online several new manufacturing buildings in San Jose on a site purchased in 2013, and last month purchased another large site in San Jose, https://therealdeal.com/sanfrancisco/2024/02/14/super-micro-computer-buys-former-frys-hq-in-san-jose/ . Whether they will use the existing building on that site or tear it down and build new (as they did with the 1960s vintage building in the 2013 purchase) has not been in the news (that I know of).

--- End quote ---
I claim no expertise in this field, but couldn't those circuits be rebalanced by other components (or even the same components) in the balancing side? You can set the baseline on a wheatstone bridge for example by simply adjusting the baseline to compensate for sensor drift or age, why not do the same with the circuits that you want to compromise?

Not arguing, just asking, thanks.

PeteS in CA:

--- Quote from: Smokin Joe on March 27, 2024, 10:13:06 pm ---I claim no expertise in this field, but couldn't those circuits be rebalanced by other components (or even the same components) in the balancing side? You can set the baseline on a wheatstone bridge for example by simply adjusting the baseline to compensate for sensor drift or age, why not do the same with the circuits that you want to compromise?

Not arguing, just asking, thanks.

--- End quote ---

Hypothetically, yes, for one or two easily accessible signal traces. Since we are talking of dozens or scores of traces in one or several inner layers of a multilayer printed circuit board, no. It's impractical to the point of impossibility. There is also the fact that the motherboard and system software would detect the increased load current from an added "chip". Whether the system disabled the MB or reported it as defective, this would be detected during system integration, which Supermicro does in San Jose, not China.

Bloomberg's story is false. Whoever sold it to them, for whatever reason, it is absurd.

Navigation

[0] Message Index

[*] Previous page