After reading several of these I'm and pretty much convinced this is our problem. Subaru's tend to kill rod bearings, not main bearings. We have an odd setup that is ok whilst the engine runs at reasonable speeds but will starve the rod journals at high rpms.
Correct, the OEM oil system is designed for long engine life at moderate rpm, that is why the Subaru STi crank (2.0 L) has a different drilling pattern to it. Cross drilling was originally used in diesel engines as I understand it, and gives great oiling and very long bearing life at moderate engine rpms, but basically shuts off oil flow when engine rpm hits a critical rpm where centrifugal force in the crank main oil gallery equals engine supply oil pressure.
Changing the crank drilling pattern to one that is more friendly to high rpm is less complicated than a high tech oil pump system.
In my opinion the OEM oil clearances on the rods are a bit tight. The OEM clearence is .0008 - .0018 with .0020 being the max limit on a 1.8891 - 1.8898 crank pin.
A rule of thumb for connecting rod oil clearance for performance engines is .0012 clearance per inch of crank pin diameter. Using that rule on the max allowable crankpin size of 1.8898 we get a suggested clearance of .00267 clearance on the rod journals. If a person buys an OEM short block and it happens to be a tight one fitted at .0008 rod clearance, you are way tight on oil clearance and at high rpm as the connecting rod pinches in at the top of the stroke as it sees maximum loading you will squeeze out all your oil clearance.
If I were building an engine today I would fit the rods at OEM max oil clearance or even slightly larger if I was going to be running serious rpm.
You also need to measure your actual rod big end diameters and standardize them to a uniform standard. For example John Lingenfelter, has two standard rod sizes he used on the Chevy V-8 engines he built. The OEM size spec for the small journal rod (nominal 2.00 diameter) was 2.1247 - 2.1252, he spec'd all his rods for this journal size at exactly 2.1250. On the large end rods (nominal 2.100 diameter) the allowable range was 2.2247 - 2.2252 from the factory, he spec'd these rods at 2.2250.
Note he did not spec a range of rod sizes there was just one rod size -- it is this level of detail that made him one of the premier engine builders in the industry.
Now for comparison, the OEM oil clearance spec on the 2.100 rod journal size for the Chevy 400 was .0013- .0035. (this equals .000619 / inch of crank pin diameter to .001666/inch of crank pin diameter.)
If you scale those clearences down to a 1.8898 crank pin used in the Subaru, our clearence which would be equivalent to the OEM Chevy clearence would be -- .00169 -- .00315
compared to our stock clearances of .0008 - .0018 .
In the for what it is worth department, the big name engine builders for high power V-8 engines tend to use oil clearances of about .003 - .0035 on the rods on their high rpm race engines. Any tighter than that and they do not live. In short they run max allowable or very slightly larger oil clearances to allow for the stresses in a high rpm environment.
It seems to me on a street build for a high performance engine a rod clearence of .0020 would be sensible, and for a no holds barred 9000 rpm engine build I would probably open them up a tad more to .0025 - .0030 on the Subaru rod journal size.
Disclaimer -- this is just my educated guess based on my experience, not a gurantee that this is the best possible setup. Consider this only with careful dimensional control on rod big end size, and crank journal size and at your own risk.