Error error 92 (0x5c)

[anakayub]

Hi, I've been repeatedly getting this error on multiple runs:

Failed to read from 'fx_0_170' work file with error 92 (0x5c)

Failed to read from 'fx_0_205' work file with error 92 (0x5c)

Monterey 12.4, Intel 2019 16 inch MacBook Pro

Using a 4.48 TB synthesized Fusion Drive (mSATA Kingston UV500 480 GB (too small for Bladebit) + 4 TB WD Green (works separately, but I want a bit more speed obviously)) via USB

Seeing if I get problems on Madmax, as I suspect problems with my SSD (consistently failing at Phase 3 with fread() errors), or using APFS as the file system

Thanks

[anakayub]

I guess the real problem is a combination of using APFS with this particular SSD

I presume that the controller is unable to cope with the CoW/other fast i/o mechanisms with this file system, as even using cache, the total file sizes exceed the capacity

The plotting works when using HFS instead (I can observe the fact that writes > used space (with around 10 GB left prior to Phase 1 completion), while with APFS with the SSD alone it would always fail before table 7 of phase 1

bladebit -t 14 -f ... -c ... --cache 16G -t1 /Volumes/Kingston\ 480\ GB /Volumes/WD\ Green\ 4\ TB\ 3/Chia\ Plots on a 9980HK 2019 MacBook Pro, 32 GB RAM via USB 3.0, thermally modded to improve cooling

Cache size : 16.00 GiB - I noticed that BladeBit doesn't use all of the allocated RAM, at most around 14GB out of 20GB total allocation (could not get a graph to see real time peaks, just looking from time to time)
Finished Phase 1 in 5908.90 seconds ( 98.5 minutes )
Finished Phase 2 in 745.37 seconds ( 12.4 minutes )
Finished Phase 3 in 5209.80 seconds ( 86.8 minutes )
Finished plotting in 11864.83 seconds ( 197.7 minutes )

Cache size : 8.00 GiB
Finished Phase 1 in 6663.94 seconds ( 111.1 minutes )
Finished Phase 2 in 762.32 seconds ( 12.7 minutes )
Finished Phase 3 in 5965.41 seconds ( 99.4 minutes )
Finished plotting in 13391.69 seconds ( 223.2 minutes )

In either case, the I/O is not maxed out (based on the CPU swings), maybe due to the drive used.

I/O


CPU

Table 5, Phase 1

Not as fast as MadMax via USB 3 2.5 Samsung Evo 870 (130 min plot times when I best tuned things), but I'm just using what I have, and the Samsung is now repurposed as a laptop drive elsewhere (but still used for plotting from that machine, as the drive is so reliable). And I'm getting input not sorted errors on MadMax with the Kingston (maybe related to APFS use as well, I haven't tested it out with HFS), which didn't happen with the Samsung. One thing I liked about MadMax was that I could plot with 250 GB capacity still left in the SSD. But the real story was that this SSD sucks (initially bought it for a Drobo 5N2 cache).