> > Actually, it did compile when I removed -O2. I haven't tried it yet > > That's good news. For awhile, it wouldn't compile no matter what I > tried. I put the whole matter on the shelf back then because nobody was > complaining about it anyway :-)
Well, I am =) > > What kind of performance hit will happen with -O2 removed? > > Shouldn't be much. Not sure what kind of hit a kernel-space driver would > take if it's not -O2'ed. If you run any tests, please report the results > as I suspect that I'm not the only curious party. I rebuilt my alpha, I think I'll put the old back on. I saved an exact image of the disk (2gb isn't that hard to store =) I see no advantage of the way it's configure now to the way it was. > > On my alpha, I have firmare 2.70 (I modded the module the other guy compiled > > for me so it would work. thankfully it was just a string I changed from > > 2.73 to 2.00) and it works w/o any problems. 9.89mb/sec read from a raid0 > > of 3 2gb disks. > > > > A friend of mine on a SMP pc is running the same card (nec branded firmware) > > with firmware 2.39 and is getting speeds about 2mb slower than me but he's > > using a raid 5 on 4 4gb disks > > Not bad...not phenominal, but not bad at all (I'm getting really used to > fibre-channel stuff lately, so ignore me :-P). You have to think too, this is an old fast wide scsi controller, not an ultra wide, u2w, not u160/u320... But hey, for free, do you complain? =) I gave 100$ for this alpha. has a 266mhz cpu, 160mb ram, 4x dec (toshiba) cdrom (replaced with a piece of junk 40x toshiba), a 4gb dec (archive python I think) tape drive (replaced with an older 2gb drive as I used the 4gb elsewhere), 2 1gb dec wide disks (quantum) which have both died and replaced with a wde 2gb, 3 2gb dec wide disks (seagate), a 2.88mb floppy, 2 dac960pd (1 channel) raid cards, dec tulip 10/100 nic, built in 7 bay storage works cabnet, external 7 bay storage cabnet (redundant ps capable, but it broke on me). it's an as1000a 4/266. hey, you can't beat the price!!! > > what about gcc 3.0? It doesn't work well on my alpha, but is it not > > recommended? > > I haven't had many problems with 3.0 yet (I test it fairly frequently), > but it does have problems that have since been fixed upstream for the > future 3.1. I've built kernels with it and never had problems with those, > but given the amount of flux that is still going on from a C++ standpoint, > I haven't been comfortable enough to say "let's go with it starting today > by default". Basically, it's fun to play with and compare to 2.95.4, but > I'm not ready to rely on it yet. I see. with 3.0.3, I saw tons of ECC errors. They aren't in the log so I can't paste them for you. (Now if someone says I have a hardware problem, I'll shoot you! >=) -- Lab tests show that use of micro$oft causes cancer in lab animals
