0.5 - 6% of Ryzen 5000 Processors Faulty

[humbug]

Never heard of PowerGPU until now.. so I guess their marketing worked My WD 10GB HDD was DOA, I returned it and got a working one, been fine since. Thats a failure rate of 50% based on the two samples I used. That's what a warranty is for.

I'm not sure tweeting out "Our PC's have a high failure rate" is good marketing.

 

[philo-sofa]

I think scale will have an impact on these rates.

PowerGPU are a system builder, in the short time they have been building Ryzen 5000 PC's how many have they sold? a few hundred? so they order several CPU batches of 50 units, the first two batches they may have had one dead CPU, so "0.5%" they order another two batches and there are 3 dead CPU's "3%" and another two batch has 4 dead in one 50X batch and two in the other "6%"...


The thing here is if you're only shifting a few 100 units you don't really have a sample size that's big enough for a conclusion, if you're shifting tens of thousands you probably do and they ain't necessarily going to agree with eachother.

We do actually have enough data to start doing this now. Am a statistician by trade so did a 99% confidence interval for the PowerGPU 5950X and 5900X on the back of a napkin and it gives a range 3.6% to 20.4% rejection rate for their CPUs. However combining PowerGPU with Mindfactory.de's dataset you get a much narrower failure rate: a 99% chance it's between 0.44% to 0.67%.

The trouble is we also need PC world's data, given they suggest a large manufacturer has a rejection rate of 2.9%.

Looking at it overall I'd suggest there's both low volumes for some outfits leading to unusual findings and also different failure rates for different batches (e.g. whomever PC world's large 2.9% source is, vs mindfactory). On top of that there are obviously differing definitions of failure.

My sense is that the true 'fail rate' is about of 0.44% to ~3.2% most likely, but I really need all the raw data.

 

[humbug]

We do actually have enough data to start doing this now. Am a statistician by trade so did a 99% confidence interval for the PowerGPU 5950X and 5900X on the back of a napkin and it gives a range 3.6% to 20.4% rejection rate for their CPUs. However combining PowerGPU with Mindfactory.de's dataset you get a much narrower failure rate: a 99% chance it's between 0.44% to 0.67%.

The trouble is we also need PC world's data, given they suggest a large manufacturer has a rejection rate of 2.9%.

Looking at it overall I'd suggest there's both low volumes for some outfits leading to unusual findings and also different failure rates for different batches (e.g. whomever PC world's large 2.9% source is, vs mindfactory). On top of that there are obviously differing definitions of failure.

My sense is that the true 'fail rate' is about of 0.44% to ~3.2% most likely, but I really need all the raw data.

What are you doing? no we don't.

The 2.9% is not a failure rate, its the rate the CPU's failed "to meet this particular system builders standards" a bit like 8Pack might bin CPU's for his systems, if they don't meet his standard they don't go in, that's what that is. PowerGPU said their CPU's were DOA.

Adding in the two system builders one of which isn't claiming to have 2.9% of "dead" CPU's is not going to make the blindest bit of difference to large mainstream retailers selling tens of thousands of CPU's each, adding PowerGPU to mindfactory isn't going to shift their 0.6% failure rate by so little as a fraction.

 

[G J]

This isnt really an issue for system builders as any bad chips would not get to customers if they test the systems before shipping them which I'm guessing near all do.

I'm sure AMD (or any other company) will replace the items and do their own internal investigation as maybe a few slipped by QC which they will resolve.

Unless something like this is a massive widespread issue and your rasing awareness if the company isnt listening then nothing really good comes from doing what PowerGPU have.

 

[Pleiades]

So more like 0.001% for intel.

Also far more issues posted on these forums about amd cpus than intel. So fanboy defence is gonna have a hard time going against the facts.

Made up stats alert.

The hypocrisy of saying 'fanboy defence'!

 

[philo-sofa]

What are you doing? no we don't.

I did a confidence interval for a binomial probability function, based on the Bernoulli distribution. Jargon aside it’s a dead simple, robust calculation.

Adding in the two system builders one of which isn't claiming to have 2.9% of "dead" CPU's is not going to make the blindest bit of difference to large mainstream retailers selling tens of thousands of CPU's each, adding PowerGPU to mindfactory isn't going to shift their 0.6% failure rate by so little as a fraction.

The confidence interval for the combined data set suggests that 99% of the time, the true failure rate of all Ryzens will be within 0.44% and 0.67%.

Most of the variability of the range of my estimate comes from the sample size. The point of adding PowerGPU to mindfactory data is just to use all available data.

There’s a separate estimate from a ‘major manufacturer’ of 2.9% (as per your link) that can’t be ruled out either yet if we trust that journalist.

 

[humbug]

I did a confidence interval for a binomial probability function, based on the Bernoulli distribution. Jargon aside it’s a dead simple, robust calculation.


The confidence interval for the combined data set suggests that 99% of the time, the true failure rate of all Ryzens will be within 0.44% and 0.67%. However there’s another estimate from another dataset from a ‘major manufacturer’ of 2.9% (as per your link) so we can’t rule that out either yet if we trust that journalist.

Most of the variability of the range comes from the sample size. The point of adding PowerGPU to mindfactory data is just to use all available data.

You keep including that 2.9% figure knowing this is not a failure rate.

 

[philo-sofa]

You keep including that 2.9% figure knowing this is not a failure rate.

I’ve already noted the specifics of this a few posts up man.

To reiterate there are differing definitions of failure here which complicates things. I’d add that when a CPU can’t run high frequency or low latency memory as PCWorld reported that’s not exactly a pass.