amos2ace gives up at 2Gb?

[Mike DS]

Has anyone successfully converted an .afg assembly file (>= 3Gb) to .ace using amos2ace? I am unable to do this on a macbook pro (8 Gb RAM) as the output .ace file is truncated.

Details:
I have been given an .afg file assembly, and can convert it to .bnk (with hawkeye), but when I want to convert to .ace (using amos2ace) the resulting .ace file is truncated part way through the phd file descriptions. So when I try and import into Consed, it very rightly complains that there are very many missing .phd files.
The original .afg is 3 Gb, and amos2ace gives a (truncated) .ace file of 1.93 Gb. When I look inside it, it looks fine until it just stops (see below).
- Mike DS

-----------------------just the last bit below-----------------
QA 1 251 1 251
DS CHROMAT_FILE: 5387375 PHD_FILE: 5387375.phd.1 TIME: Thu Dec 12 00:14:39 2013
RD 5393237 228 0 0
CTGGCTGTCTATGCCTTCCCACATCGTTTCCCACTTAACCATGACTTTGG
GACCTTAGCTGGCGGTCTGGGTTGTTTCCCTCTTCACGACGGACGTTAGC
ACCCGCCGTGTGTCTCCCGTGATAACATTCTTCGGTATTCGGAGTTTGCA
TCGAGTTGGTAAGCCGGGATGGCCCCCTAGTCGAAACAGTGCTCTACCCC
CGAAGATGAATTCACGAGGCGCTACCTA

QA 1 228 1 228
DS CHROMAT_FILE: 5393237 PHD_FILE: 5393237.phd.1 TIME: Thu Dec 12 00:14:40 2013

Wamos2ace $Revision$
Run by root on Thu 12 Dec 2013 00:15:39 EST
}

[Mike DS]

follow up - converted to BAM

So, I gave up trying to get an .ace file. I converted the .afg to .bnk format, then .bnk to SAM, then SAM to BAM, and can now view the assembly very nicely in Tablet.
Now I can see the velvet assemblies, I notice many have crazy alignments for the first 250 nt; then an abrupt change to perfect alignment.
- Mike DS