B3 with different RAM chips

[John Watlington]

On Apr 20, 2010, at 8:30 PM, Tiago Marques wrote:

> On Tue, Apr 20, 2010 at 8:54 PM, John Watlington <wad at laptop.org> wrote:
> 
> On Apr 19, 2010, at 8:51 PM, Tiago Marques wrote:
> 
> > Hi John,
> >
> > I'm trying out some new USB Flash drives from Sony that are quite fast(random write speed) compared to the SD cards shipping in the XO 1.5 and I'd just like to ask you something about the test script running on the Sandisk Extreme III:
> >
> > [edit] Sandisk Extreme III
> > There are four XOs at 1CC running the tests on a SanDisk Extreme III 4-GB SD card. Build 8.2-760 was freshly installed on the laptops.
> > The current test rates are roughly 3.8 sec/test step 4.1, and 4.1 sec/test step 4.2. This translates roughly into a 17 MByte/s read rate, and a 5.7 MByte/s write rate.
> > Laptop        Serial #        Test    Total Written
> > SAN1  SHF725004D1     Wear & Error    30,550
> > SAN2  SHF7250048F     Wear & Error    28,335
> > SAN3  SHF80600A54     Wear & Error    28,730
> > SAN4  CSN74902B22     Wear & Error    29,250
> >
> > Am I getting this right? They endured 30TiB of writes and still did ok?
> 
> No, that was 30K cycles (times 20MB/cycle) or just 600 GB.
> 
> That's too little :|
> On that page it says that "Total Written" represents GiB, not cycles. Is that incorrect?
> 
> 
> > Did they start marking bad blocks as unusable, thus shrinking the drive, or did they remained with the same 4GB without failing?
> > What are the results you obtained with the SD cards for XO 1.5 and which cards did you settled with?
> 
> An SD card sets aside between 5 and 10% of the raw storage for use in fixing bad blocks.
> Thus, you will see no change in the size of the card over time, until it runs out of spare blocks
> and fails completely.
> 
> 
> I'm asking this because I've ran into the shrinking behavior before, with other devices(http://www.siliconmadness.com/2008/03/asus-eee-pc-has-ssd-reliability-problem.html), which to me seems a better trade-off.
>  
> We didn't settle on "one card", and even if I tell you the manufacturer and size it won't help
> as most have various OEM models for a given size. 
> 
> I see. So you're saying that if you order the same card model and size, you will still get different FTL and/or NAND? 

No.   When OLPC specifies a card vendor and model, we require that the vendor not change either FTL or NAND
without changing the model number.    Quanta is large enough to get this assurance from vendors.

When you buy a card in the retail market, you have little or no assurance that the
card model number means anything.   See http://www.bunniestudios.com/blog/?p=918
for an example.

>  > I haven't checked the brand of the sample in my B2 but it's been running rock solid as far as I can tell, the only problem is it being so slow with random writes because otherwise the 1.5 is quite a capable machine.
> 
> Are you running into the SD card errors on an XO-1 ?
> 
> Just on cards I bought after market, either on the XO-1 or 1.5.
> Only one SD card from Sandisk has been working perfectly for me, in the XO-1. I'm looking at ways to test faster USB Flash drives, since the SD card on the XO 1.5 becomes very slow at times. Installing Linux on an external HDD is like breathing life into the device, I became very impressed with performance when I tried it.

There a number of known issues.   First, the performance of the external SD port on the XO-1
was belatedly discovered to have a high bit-error rate.   This depends on the particular laptop
and the SD card being tested.   The bit error rate is low enough that it didn't show up in Quanta's
QA testing, but it does appear in my long term testing.

On XO-1.5, we ran into similar issues.  I fixed it for most cards, but there are notable exceptions.
I was not able to eliminate the bit-error rate for any random card...

wad